Anticipated acquisition by

Microsoft of Activision
Blizzard, Inc.
Final report

26 April 2023
© Crown copyright 2023

You may reuse this information (not including logos) free of charge in any format or
medium, under the terms of the Open Government Licence.

To view this licence, visit www.nationalarchives.gov.uk/doc/open-government-

licence/ or write to the Information Policy Team, The National Archives, Kew, London
TW9 4DU, or email: publicsectorinformation@nationalarchives.gov.uk.

Website: www.gov.uk/cma
 Members of the Competition and Markets Authority
who conducted this inquiry

Martin Coleman (Chair of the Group)

John Thanassoulis

Humphrey Battcock

Ashleye Gunn

Chief Executive of the Competition and Markets Authority

Sarah Cardell

The Competition and Markets Authority has excluded from this published version
of the report information which the Inquiry Group considers should be excluded
having regard to the three considerations set out in section 244 of the Enterprise
Act 2002 (specified information: considerations relevant to disclosure). The
omissions are indicated by []. Some numbers have been replaced by a range.
These are shown in square brackets. Non-sensitive wording is also indicated in
square brackets.
Contents
Page

Anticipated acquisition by Microsoft of Activision Blizzard, Inc. ................................. 1

Summary .................................................................................................................... 4
Findings.................................................................................................................... 24
1. The reference ..................................................................................................... 24
2. The Parties, the Merger, and the rationale.......................................................... 25
The Parties ......................................................................................................... 25
Microsoft ........................................................................................................ 25
Activision ....................................................................................................... 28
The Merger ......................................................................................................... 31
Valuation ....................................................................................................... 32
The rationale ....................................................................................................... 33
Parties’ submissions ...................................................................................... 33
3. Relevant merger situation ................................................................................... 35
Enterprises ceasing to be distinct ....................................................................... 35
Jurisdiction test ................................................................................................... 36
Conclusion on the relevant merger situation ....................................................... 36
4. Industry background ........................................................................................... 37
Introduction ......................................................................................................... 37
Overview of the industry ..................................................................................... 37
Game development, publishing, distribution and marketing................................ 38
Game Development ....................................................................................... 39
Game Publishing ........................................................................................... 39
Game Distribution .......................................................................................... 40
Game Marketing ............................................................................................ 40
Gaming hardware ............................................................................................... 41
Gaming devices ............................................................................................. 41
Supporting infrastructure ............................................................................... 42
Game purchasing................................................................................................ 44
Cloud gaming ...................................................................................................... 45
Overview of cloud gaming ............................................................................. 45
Cloud gaming monetisation models............................................................... 45
Cloud gaming infrastructure and ecosystem ................................................. 46
5. Introduction to competitive assessment .............................................................. 48
Treatment of evidence ........................................................................................ 48
Market definition.................................................................................................. 53
Framework..................................................................................................... 53
The supply of gaming hardware (and associated gaming distribution) .......... 55
The supply of cloud gaming services............................................................. 66
The supply of game publishing services ........................................................ 74
6. Counterfactual..................................................................................................... 84
Framework for assessment of the counterfactual ............................................... 84
The Parties’ submissions on the relevant counterfactual .................................... 85
Our assessment .................................................................................................. 86
7. Theory of harm 1: Vertical effects in console gaming ......................................... 87
Introduction ......................................................................................................... 87
Competitive landscape in console gaming services ............................................ 87
Introduction .................................................................................................... 87

1
 Features of the console gaming market ........................................................ 88
Shares of supply in the downstream market .................................................. 97
Competition among gaming consoles.......................................................... 100
Availability of Activision’s content on MGS services ......................................... 108
Parties’ submissions .................................................................................... 108
Our assessment .......................................................................................... 109
Framework for assessment ............................................................................... 117
Ability to foreclose ............................................................................................. 120
Importance of content .................................................................................. 121
Importance of CoD to SIE ............................................................................ 124
Alternatives to CoD...................................................................................... 146
Contractual arrangements and negotiations ................................................ 157
Partial foreclosure........................................................................................ 160
Conclusion on the Merged Entity’s ability to foreclose................................. 163
Incentive to foreclose ........................................................................................ 165
Introduction .................................................................................................. 165
Parties’ views............................................................................................... 166
Third parties’ views ...................................................................................... 167
Our assessment .......................................................................................... 168
Conclusion ................................................................................................... 190
Effect of foreclosure on competition .................................................................. 191
Conclusion on TOH1......................................................................................... 191
8. Theory of harm 2: Vertical effects in cloud gaming services ............................. 192
Framework for assessment ............................................................................... 192
Future development of cloud gaming ................................................................ 194
Parties’ views............................................................................................... 195
Our assessment .......................................................................................... 196
Competitive Landscape in cloud gaming services ............................................ 206
Features of the cloud gaming market .......................................................... 206
Overview of competitors .............................................................................. 211
Microsoft’s strengths in cloud gaming ............................................................... 218
Operating systems....................................................................................... 219
Cloud infrastructure ..................................................................................... 228
First and third party content ......................................................................... 237
Microsoft’s overall strengths in cloud gaming services relative to rivals ...... 240
Availability of Activision’s content on cloud gaming services ............................ 247
Parties’ views............................................................................................... 247
Our assessment .......................................................................................... 249
Ability to foreclose ............................................................................................. 260
Importance of Activision content .................................................................. 261
Microsoft’s agreements with NVIDIA, Boosteroid and Ubitus ...................... 281
Conclusion on Ability ................................................................................... 283
Incentive to foreclose ........................................................................................ 285
Parties’ views............................................................................................... 286
Our assessment .......................................................................................... 287
Effect of foreclosure .......................................................................................... 295
Introduction .................................................................................................. 295
Strength of different cloud gaming services ................................................. 295
Barriers to entry and expansion ................................................................... 300
Conclusion on effect .................................................................................... 303
Conclusion on TOH2......................................................................................... 304

2
9. Countervailing factors ....................................................................................... 305
Entry and expansion in game publishing .......................................................... 305
Introduction .................................................................................................. 306
Barriers to entry and expansion ................................................................... 307
Recent entry or expansion and prospects of further entry and expansion ... 318
Conclusion on entry and expansion in game publishing .............................. 321
Efficiencies ........................................................................................................ 323
Parties’ views............................................................................................... 324
Our assessment .......................................................................................... 325
Conclusion on efficiencies ........................................................................... 327
10. Conclusion on SLC ........................................................................................... 328
11. Remedies .......................................................................................................... 329
Introduction ....................................................................................................... 329
Our remedy consideration process ................................................................... 329
CMA remedies assessment framework ............................................................ 331
Overview of remedy options ............................................................................. 333
The Parties’ responses to the Remedies Notice and Remedies Working Paper
................................................................................................................... 335
Effectiveness of prohibition ............................................................................... 336
Description of remedy .................................................................................. 336
Views of Parties and third parties ................................................................ 336
Effectiveness of prohibition .......................................................................... 338
Effectiveness of partial divestiture..................................................................... 339
Effectiveness of partial divestiture ............................................................... 339
Effectiveness of behavioural remedies ............................................................. 340
Description of remedy .................................................................................. 340
Views of the Parties and third parties .......................................................... 345
Assessment of the effectiveness of a behavioural remedy .......................... 357
Conclusions on the effectiveness of the Microsoft Cloud Remedy .............. 371
Conclusions on remedy effectiveness............................................................... 372
Proportionality ................................................................................................... 372
Relevant customer benefits ......................................................................... 373
Proportionality assessment ......................................................................... 404
Remedy Implementation ................................................................................... 414
Final decision on remedies ............................................................................... 415

Appendices

A: Terms of reference
B: Conduct of the inquiry
C: Shares of supply
D: Our assessment of the evidential value of the CMA Survey
E: Our assessment of the Parties’ incentives to foreclose
F: Foreclosure of rival PC operating systems
G: CMA Incentives model
H: Summary of responses from members of the public

Glossary

3
 Summary

What we have found

1. The Competition and Markets Authority (CMA) has found that the anticipated
acquisition (the Merger) of Activision Blizzard, Inc. (Activision) by Microsoft
Corporation (Microsoft) (together, the Parties) may not be expected to result
in a substantial lessening of competition (SLC) in console gaming services in
the UK. However, the CMA has found that the Merger may be expected to
result in an SLC in cloud gaming services in the UK.

2. In relation to console gaming services, we found that Xbox (Microsoft) and

PlayStation (Sony) compete closely with each other, and that Activision’s Call
of Duty (CoD) is important to the competitive offering of each. The evidence
suggests, however, that Microsoft would not find it financially beneficial to
make CoD exclusive to Xbox after the Merger. We also found that making
CoD available on Xbox on better terms than on PlayStation would not
materially harm PlayStation’s ability to compete. On this basis, we found that
the Merger would not substantially reduce competition in console gaming
services in the UK.

3. In relation to cloud gaming services, we found that Microsoft already has a

strong position. It owns a popular gaming platform (Xbox and a large portfolio
of games), the leading PC operating system (Windows), and a global cloud
computing infrastructure (Azure and Xbox Cloud Gaming), giving it important
advantages in running a cloud gaming service. With an estimated 60-70%
market share in global cloud gaming services, it is already much stronger than
its rivals.

4. We found that the Merger would make Microsoft even stronger and
substantially reduce competition in this market. We found that Activision’s
titles—including CoD, World of Warcraft, and Overwatch—will be important for
the competitive offering of cloud gaming services as the market continues to
grow and develop. We found that, after the Merger, Microsoft would find it
commercially beneficial to make Activision’s titles exclusive to its own cloud
gaming service. Given its already strong position, even a moderate increment
to Microsoft’s strength may be expected to substantially reduce competition in
this developing market, to the detriment of current and future cloud gaming
users.

5. Microsoft offered a behavioural remedy to address our concerns in cloud

gaming services (the Microsoft Cloud Remedy). The Microsoft Cloud
Remedy did not aim to restore competition to the level that would have

4
 prevailed absent the Merger, but rather to impose obligations on Microsoft to
support cloud gaming service providers using certain business models (such
as by allowing them to stream certain games purchased by users through
certain storefronts) for a ten-year period.

6. We found that the Microsoft Cloud Remedy had several shortcomings

connected with the growing and fast-moving nature of cloud gaming services.
In particular, the scope of the remedy was limited to cloud gaming providers
with specific business models. As a result, Microsoft would not have to supply
Activision’s full range of games to providers that, absent the Merger, may have
entered into a different type of commercial relationship with Activision (eg,
through exclusive content, joint marketing arrangements, or a multi-game
subscription service like Game Pass) or to cloud gaming providers that may
decide to operate using a non-Windows PC operating system (eg Linux). It
also did not provide for competition through differentiation in content. The
complexity of the remedy, in the context of a dynamic market that is evolving,
also meant that it had a high risk of circumvention, and that it would have been
difficult to monitor effectively. In light of these shortcomings, we could not be
sufficiently confident that the Microsoft Cloud Remedy would have addressed
our concerns, and we found that the only effective remedy to the SLC is to
prohibit the Merger.

7. At a late stage in our process—over a year after the Merger was announced
and after competition authorities in the EU, the UK, and the US expressed
concerns about the Merger—Microsoft told us that it had entered into
agreements with Nintendo and three cloud gaming service providers to allow
certain Activision content to be made available on their platforms after the
Merger. Microsoft told us that these agreements, along with Microsoft's plan to
enter the mobile gaming market and its intention to place Activision's content
on Game Pass (Microsoft's multi-game subscription service), were relevant
customer benefits (RCBs) that would make prohibition disproportionate.

8. We found that most of these did not qualify as RCBs. We found that the
Merger itself would not increase—rather, it would decrease—the incentive that
Activision would have had absent the merger to enter into these agreements.
We also found that the impact of these agreements was highly uncertain, and
we could not be confident that they would lead to material benefits for
customers. As for Microsoft’s plans to enter the mobile gaming market, we
found that these plans were far from certain, especially in current
circumstances where the largest mobile OS—Google’s Android and Apple’s
iOS—either currently prohibit rival mobile gaming app stores or impose strict
limits on their ability to monetise content.

5
9. We found that Microsoft’s intention to place Activision’s games on Game Pass
on the date of their release amounts to an RCB. However, we found that this
benefit would likely be limited. Having Activision’s content on Game Pass
would represent a new option to pay for content that is already available on a
buy-to-play basis on Xbox, and it would only represent better value than the
status quo for some consumers (which, in any event, would only start to
accrue some time after the Merger completes). Moreover, we expect Microsoft
to have the incentive to increase the price of Game Pass commensurate with
the value enhancement of adding Activision’s valuable content to it, and we
found that even a modest price increase would significantly reduce or
eliminate any potential RCB.

10. We compared the RCBs that would be foregone from prohibition to the harm
that would arise from the SLC. We believe the likely future growth, competitive
dynamism and innovation in the cloud gaming market that would be
substantially reduced as a result of the Merger would lead to a significantly
greater level of harm to UK consumers than any RCBs foregone. We also
considered other factors such as the broader international context and extra-
territorial impact of prohibition, but we found no effective remedy that would
address the SLC in the UK without having an impact outside of the UK.
Furthermore, given our SLC finding and the absence of an alternative effective
remedy, we consider that prohibition is not disproportionate in order to protect
competition and consumers in the UK.

11. On this basis, the CMA has decided to prohibit the Merger.

About the gaming industry

The same three companies have been the only major suppliers in the
console gaming market for the past 20 years

12. The gaming industry is the UK’s largest revenue-generating form of

entertainment. It is bigger than pay TV, home video (including streaming),
cinema, music, or books. In 2022, it generated around £5 billion in revenue in
the UK.

13. For the past twenty years, the same three companies have been the only
significant suppliers of console gaming – Microsoft (Xbox), Sony (PlayStation)
and Nintendo (Switch being the current generation console), with little or no
entry from new rivals.

14. Part of the difficulty in entering and expanding in the console gaming market is
the existence of strong network effects. Console providers such as Microsoft
compete to attract users who want to play high-quality games, as well as high-

6
 quality content from game developers. Consoles with a large user base attract
more users, especially those who want to play multi-player games with their
friends and other users (ie, direct network effects). Consoles with a lot of users
attract better content, which in turn attracts more gamers to that console,
which in turn attracts better content, and so on (ie, indirect network effects).
This self-reinforcing mechanism makes it more difficult for new entrants
without a large user base or good pre-existing gaming content to enter and
grow in the market.

15. Gaming consoles compete against each other across a wide range of
parameters, including price, quality, and game range. Price is determined by
the console manufacturer, both for the console itself and for the console
provider’s own games on its console. Quality reflects mainly a console’s
technical specifications (eg, CPU, GPU, RAM, storage, video output, audio
output, connectivity, networking features, etc). These can affect the range of
games that can run on the console and the quality of gameplay. Game range
is determined by the titles available from the console manufacturer (first-party
titles), together with the titles from other publishers (third-party titles) available
on that console. In general, the console provider’s first-party titles are less
likely to be available on other consoles, whilst third-party titles are more likely
to be available across different consoles.

16. The most important games for a console are typically referred to as ‘AAA’,
which is a loosely defined term to denote the most popular, costly and/or
graphically intensive games in the industry. Although there are thousands of
games available on console and PC, only a handful of AAA games, including
CoD, account for the majority of gametime and revenues on Xbox and
PlayStation.

17. In recent years, gaming consoles have also started to compete on the basis of
their multi-game subscription offerings. Unlike the traditional buy-to-play
model, where users pay an up-front fee for lifetime access to a game, these
services allow gamers to access a catalogue of games for a fixed, often
monthly, fee. Although some multi-game subscription services have extensive
gaming catalogues, several AAA games (such as CoD) are either not currently
available on these services or only available in older versions. While most of
the revenue in the industry continues to be generated from the purchase of
individual games and in-game purchases, multi-game subscription services
are expected to grow over the next few years.

18. In addition to consoles, people play games on PCs and mobile devices.
Consoles and PCs designed for gaming can usually process more complex
and technically demanding games (such as CoD). Mobile devices currently
lack the technical capabilities to run most console games locally, and people

7
 can use them to play more casual games specifically designed for mobile
devices (such as Candy Crush or Call of Duty Mobile).

Cloud gaming services are growing as a potential alternative to consoles

19. In recent years, cloud gaming has started to emerge as an alternative to

gaming consoles and PCs. Unlike consoles and PCs, where gamers typically
download and run games locally on their device, cloud gaming services allow
complex games to be accessed on remote servers and streamed directly to a
range of devices.

20. The evidence we have seen suggests that cloud gaming may be an important
disruptive force in the gaming industry. Since games are executed remotely,
gamers can play using devices that can be less powerful, and are often
cheaper, than consoles or gaming PCs (such as mobile phones, smart TVs,
less powerful PCs, or tablets). This widens the pool of potential customers—
including to those not willing or able to buy a gaming console or PC—and
introduces new ways to compete that could facilitate new entry. Besides
Microsoft, recent new entrants into cloud gaming include Amazon Luna,
NVIDIA GeForce Now, Boosteroid, Shadow, and Google Stadia (now shut
down).

21. Several industry experts predict that cloud gaming will continue to grow
significantly in the coming years. While estimates vary, market reports forecast
that global cloud gaming revenue will increase to $6.1–11.4 billion by 2025,
and $11.9–13.5 billion by 2026. This suggests that the UK market will be worth
$0.6–1.1 billion by 2025, and $1.2–1.3 billion by 2026. UK cloud gaming
monthly active users more than tripled from the start of 2021 to the end of
2022.

22. The evidence we have seen suggests that, to succeed, cloud gaming
providers will need to offer a strong gaming catalogue. For new entrants
without an existing gaming console (including its games and operating
system), we have found that this catalogue is most likely to come from games
that are currently available on PC OS, as these can be streamed from any
cloud gaming service that runs that OS (provided that adequate licensing
arrangements are in place). As such, these cloud gaming service providers
will either need a license for a proprietary PC OS—such as for Windows, the
OS for which most PC games are designed—or they will need to operate their
service using an open-source PC OS such as Linux. They will also need
access to cloud infrastructure.

8
Who are the businesses and what services do they provide?

Microsoft has a strong gaming ecosystem

23. Microsoft is a global technology company offering a wide range of products

and services, with a global turnover of around £150 billion in FY2022. Since
2001, it has released several generations of Xbox gaming consoles. Gamers
typically download digital copies of the games they want to play on Xbox from
Microsoft’s Xbox Store. They can also pay a monthly fee to gain access to a
library of downloadable and cloud-based content via Xbox Game Pass,
Microsoft’s multi-game subscription service.

24. Microsoft is also a game publisher and currently owns 24 game development
studios, several of which it acquired in recent years. These studios make
games such as Minecraft, Forza, Elder Scrolls, and Halo for Xbox and other
consoles, PC, and mobile devices. Many of Microsoft’s first-party titles are
available exclusively on Xbox and PC, and some are licensed to rival console
providers.

25. Microsoft has other business areas that are important to gaming. One is Xbox
Cloud Gaming, Microsoft’s current cloud gaming service, which is powered by
custom Xbox Series X hardware. Another is Azure, a leading cloud platform
(ie a network of data centres and cloud computing infrastructure) that offers a
wide range of services across several industries, including gaming. Another is
Windows, the leading PC OS. Many people play games on a PC rather than a
console, and most of them use Windows OS. Because of its popularity, game
developers generally make PC games that are designed and optimised for
Windows OS.

Activision creates some of the most popular gaming content

26. Activision is a game developer and publisher with global turnover of

£6.1 billion in FY2022. It develops gaming content for consoles, PC, and
mobile. Activision’s three most popular franchises—CoD, World of Warcraft
and Candy Crush—account for most of its revenue. It publishes these games
through three separate business divisions, ie, Activision, Blizzard, and King,
respectively.

27. CoD, in particular, is widely regarded as one of the most successful gaming
franchises of all time. For more than a decade, its releases have ranked in the
top games available on console and PC. The latest game in the franchise,
CoD Modern Warfare II, was released in November 2022 to what Activision
described as the #1 top-selling opening weekend ever in the franchise.

9
Our Assessment

Why did we review this merger?

28. The CMA’s primary duty is to seek to promote competition for the benefit of
consumers. It has a duty to investigate mergers that could raise competition
concerns in the UK, provided it has jurisdiction to do so.

29. Microsoft announced in January 2022 that it had agreed to acquire Activision
for a purchase price of USD 68.7 billion. The Merger was conditional on
receiving merger control clearance from several global competition agencies,
including the CMA.

30. While both Microsoft and Activision are US-based entities, the question for the
CMA is whether the Merger may have an impact on competition in the UK.
This link to the UK can be established based on the turnover of the business
being acquired in the UK (ie whether the UK turnover of that business is more
than £70 million). In this case, we concluded that the CMA had jurisdiction to
review this Merger because Activision met that threshold in FY2021.

How did we examine this merger?

31. In deciding whether a merger may be expected to result in an SLC, the

question we are required to answer is whether there is an expectation—a
more than 50% chance—that the merger will result in an SLC within any
market or markets in the UK.

32. To determine whether this is the case, we have gathered information from a
wide variety of sources, using our statutory powers to ensure that we have as
complete a picture as possible under the constraints of the statutory timetable
to understand the implications of this Merger on competition.

33. We have focused on two ways, or ‘theories of harm’, in which the Merger
could give rise to an SLC:

(a) The first considers whether Microsoft would be able to harm gaming
console rivals now or in future, to the detriment of consumers, by making
CoD exclusive to Xbox (or by only making it available to rivals on worse
terms), whether it would be commercially beneficial to do so, and what
the impact would be on competition in the market for console gaming
services in the UK.

(b) The second considers whether Microsoft would be able to harm cloud
gaming rivals now or in future, to the detriment of consumers, by making

10
 CoD and other Activision games, such as World of Warcraft and
Overwatch exclusive to its cloud gaming offering, whether it would be
commercially beneficial to do so, and what the impact would be on
competition in the market for cloud gaming services in the UK.

34. We concluded that the Merger is not likely to give rise to an SLC in console
gaming services in the UK, but it is likely to give rise to an SLC in cloud
gaming services in the UK. This is discussed in further detail below.

What evidence have we looked at?

35. In assessing this Merger, we looked at a wide range of evidence that we

considered in the round to reach our decision. The evidence we have
gathered has been tested rigorously, and the context in which the evidence
was produced has been considered when deciding how much weight to give it.

36. We received a significant volume of evidence from the Parties. In response to

targeted information requests, we received over 3 million internal business
documents from Microsoft and Activision, including key strategy documents
and email communications among senior staff. These documents which, for
the most part, were created in the ordinary course of business, set out the
Parties’ views of the console and cloud gaming markets, as well as their future
commercial strategy.

37. The Parties also had several opportunities to make submissions and comment
on our emerging thinking throughout the investigation. In October 2021, the
Parties submitted a response to our phase 1 decision. They subsequently
submitted a response to our Issues Statement, where we set out the theories
of harm on which we planned to focus our phase 2 investigation. We held a
site visit with each of the Parties, where their senior business staff gave us
several presentations on the nature of their businesses, the rationale for the
Merger, and answered our questions relating to our investigation. We then
produced working papers and an annotated Issues Statement with our
emerging thinking, and the Parties submitted their views on that material. We
held formal hearings with each of the Parties, in which we spoke to the
Parties’ senior management about topics that we were exploring in our
investigation. In February 2023, we published our Provisional Findings and
Notice of possible remedies. We held formal hearings with each of the Parties
to discuss our Provisional Findings and possible remedies and held a hearing
with Sony to discuss possible remedies. In March 2023, in light of the
additional and updated evidence that we received from the Parties, we
published an Addendum to the Provisional Findings in which we provisionally
found that the Merger may not be expected to give rise to an SLC in console
gaming in the UK. We continued to discuss possible remedies with the Parties

11
 and third parties, focusing on cloud gaming. We also shared a supplementary
evidence paper with the Parties covering further evidence gathered and
analysed in the period following the Provisional Findings. In addition, we had
several calls and considered a number of other submissions setting out the
Parties’ views on our theories of harm and possible remedies at different
points in our investigation.

38. We gathered evidence from other gaming console providers, game publishers,
and cloud gaming service providers. We sent out over 90 requests for
information, held several calls and meetings, and gathered hundreds of
internal documents from these third parties. In our calls, we spoke to senior
staff and business experts across the industry to have a better understanding
of the competitive landscape and likely future developments in these markets.

39. We sought views from the public. In response to our Issues Statement of 14
October 2022, we received and reviewed over 2,100 emails containing views
on the transaction. We considered those views and published a summary of
these responses on 21 December 2022. In response to our Provisional
Findings, our Addendum to the Provisional Findings, and Notice of possible
remedies, we received and reviewed around 160 emails and submissions from
the public. We considered those views and have included a summary of these
responses as an appendix to our final report.

40. We engaged an independent market research company to conduct an online

survey. The survey polled a random sample of PlayStation CoD gamers—
defined for the purposes of the survey as those who played at least 10 hours
or spent at least $100 on the game between July 2021 and June 2022—to get
a sense of how important this game franchise is to them, and what they might
do if it became partially or totally exclusive to Xbox after the Merger.

41. While there are no pre-defined measures for assessing whether a merger may
be expected to result in an SLC, market shares are a commonly used
measure in merger control cases. There is a high degree of product
differentiation in some of the markets in which Microsoft and Activision
operate, which means that in this case market shares may not be the best
indicator of how closely businesses compete with each other. As such, when
assessing the impact of the Merger on competition, we have considered the
evidence on market shares alongside other evidence on how closely the
Parties compete with rivals (either currently or in the future). As well as the
Parties’ market shares, our assessment has taken account of the type of
games that Activision offers, of the technical specifications of different
consoles (and the types of games that users play on them), and of Microsoft’s
potential strengths in cloud gaming arising from its broader multi-product
ecosystem. We have also taken account of the strength of competitive

12
 constraints on the Parties, and the extent of past entry and exit from the
relevant markets.

42. Finally, as well as looking at how competition works currently (and the Parties’
current market positions), we recognise that markets, and in particular markets
for digital products and services such as those offered by the Parties, change
over time. Our assessment is therefore forward-looking and considers how
markets are evolving and the Parties’ plans for their businesses in future.

What would have happened absent the merger?

43. To determine the impact that the Merger may have on competition, we have
considered what is likely to happen absent the Merger. This is known as the
counterfactual.

44. For an anticipated merger such as this, the counterfactual may consist of the
prevailing conditions of competition or conditions involving stronger or weaker
competition than under the prevailing conditions. In this case, based on the
evidence we gathered, our conclusion is that the counterfactual is the
prevailing conditions of competition.

45. We recognise that, as part of the prevailing conditions of competition, markets

may continue to evolve and develop. In this case, our view is that the market
is likely to develop in important ways absent the Merger in the near future (ie,
within the next five years): in relation to gaming consoles, we consider the
evidence shows that multi-game subscription services would continue to grow
but would be unlikely to offer Activision’s most valuable games on the date of
their release (we note that several AAA games currently make most of their
sales in the first 12 months after release). In relation to cloud gaming services,
we consider that the evidence shows that the market would continue to grow,
but we believe that at least some of these cloud gaming providers—especially
those with a buy-to-play or bring-your-own-game offering—would have
Activision’s most valuable games available on their platforms on the date of
their release in the foreseeable future. We explain the relevance of these
findings in our assessment below.

What did the evidence tell us?

…about the importance of Activision’s gaming catalogue

46. We have gathered substantial evidence from Microsoft, Activision, and third
parties to assess the significance of Activision’s gaming portfolio. This
evidence consistently points towards Activision’s content, especially CoD, as
being important and capable of making a material difference to the

13
 competitiveness of rivals’ gaming platforms. Activision invests significant time
and capital in creating regular CoD releases, which consistently rank as some
of the most popular games. These titles require thousands of game
developers and several years to complete, and there are very few other
games of similar popularity. Moreover, CoD’s popularity has been consistent
over time and is continuing. For example, Activision reported that the release
of CoD Modern Warfare II on 28 October 2022 was the franchise’s best-ever
opening weekend, delivering more than $800 million worldwide in the first
three days from its release.

47. Activision also offers PC games and mobile games. Through its Blizzard
division, its most popular release is World of Warcraft, a massively multiplayer
online role-playing PC game. Through its King division, it offers Candy Crush,
a free-to-play casual game available on mobile and PC. Although Activision’s
mobile games are not relevant to our SLC assessment on console or cloud
gaming services, we found that some of Activision’s broader catalogue of PC
and console games, such as World of Warcraft and Overwatch, are popular
games that may be important for cloud gaming services, thereby adding to
Activision’s already strong catalogue in this market.

…about the impact of the Merger on gaming consoles

48. Our assessment under this theory of harm has focussed on whether Microsoft
would have the ability and incentive to limit access to CoD, and whether this
‘foreclosure’ would impact rivals’ ability to compete with Microsoft in gaming
consoles. In terms of ‘ability’, we considered whether limiting access to CoD
would harm the competitiveness of Xbox’s rivals. In terms of ‘incentive’, we
considered whether Microsoft stands to gain from this strategy. And in terms
of ‘effect’, we considered how this would impact overall competition in the
market for gaming consoles.

49. CoD is currently available on two gaming consoles – Xbox and PlayStation.
We found that these consoles compete closely with each other in terms of
content, target audience, and console technology. We found that Nintendo’s
consoles compete less closely with either of Xbox or PlayStation, generally
offering consoles with different technical specifications, and with its most
popular titles tending to be more family- and child-friendly. Nintendo does not
currently offer CoD, and we have seen no evidence to suggest that its
consoles would be technically capable of running a version of CoD that is
similar to those in Xbox and PlayStation in terms of quality of gameplay and
content.

50. The evidence we gathered shows that the CoD franchise is important to
PlayStation.

14
51. First, the evidence shows that CoD accounts for a significant proportion of
PlayStation’s overall gametime, implying that making it exclusive to Xbox
would represent an important reduction in range of games offered on
PlayStation.

52. Second, the large majority of our survey respondents (ie, CoD gamers as
described above) indicated that the content available on a console is important
to their choice of console, and around 24% of them said they would divert
away from PlayStation if CoD were no longer available on that platform. The
level of switching in this analysis, which indicates that a significant proportion
of all PlayStation gamers would switch away from the platform, suggests that
PlayStation gamers would be affected by not having access to CoD,
notwithstanding the availability of other games on PlayStation.

53. Third, even CoD gamers who would remain on PlayStation could be harmed by
the reduction in choice in that console. They would also likely spend less time
and money on PlayStation than they did before, which the evidence suggests
would have a material impact on PlayStation’s revenue and ability to compete.

54. We also found that Microsoft would not have the ability to foreclose PlayStation
solely through partial foreclosure strategies. This is because PlayStation would
not lose the full extent of the range that CoD represents; rather, that part of its
range would suffer a quality deterioration (or price increase) that would likely
amount to only a small fraction of the value that gamers derive from CoD. As
such, and considering PlayStation’s broader gaming catalogue, a partial
foreclosure strategy would amount to a deterioration in a small fraction of
PlayStation’s overall range.

55. As to what Microsoft would do with CoD, we have found that it would not have
an incentive to make it exclusive to Xbox.

56. First, we found that making CoD exclusive to Xbox would result in significant
financial losses for Microsoft over a five-year time period. Given the magnitude
of those losses, we placed considerable weight on this quantitative evidence.

57. Second, we found that Microsoft’s past behaviour in relation to acquisitions of

other gaming studios was inconclusive. We found that most of these
acquisitions were, in effect, acquisitions of talent. The majority of studios that
Microsoft has acquired (with some notable exceptions such as Bethesda) did
not have regular releases of popular gaming franchises available on different
platforms. As such, Microsoft did not have to decide whether to make
multiplatform games with a large customer base exclusive to Xbox following

15
 these acquisitions; it acquired those studios with the specific purpose of
making exclusive games for its platform. Although the evidence shows that
console providers, including Microsoft, place significant value in having
exclusive content to differentiate their platform and attract more users, there
seem to be exceptions to this rule. For example, when Microsoft acquired
Minecraft (a multi-player franchise that was available on different platforms at
the time of acquisition, although significantly different from CoD in many
respects, such as its pricing model), it kept it on PlayStation and Nintendo.

58. Third, we found that there is a range of other potential gains and losses from a
foreclosure strategy that are more difficult to quantify on a comparable basis.
They include (i) furthering Microsoft’s strategy of expanding Game Pass,
(ii) any reputational impacts (good or bad), (iii) the strength of the Xbox brand
and user loyalty, (iv) the impact of network effects (including for games that
allow cross-play), and (v) the potential for entry, expansion, or repositioning by
rivals to disincentivise foreclosure. We found that, on balance, these factors
tend to contribute to Microsoft’s incentive to engage in a foreclosure strategy.
In light of the magnitude of the potential losses that Microsoft would incur from
a total foreclosure strategy, however, we found that these longer-term
strategic incentives, together with all other available evidence, are not
sufficient to show that Microsoft would have an incentive to make CoD
exclusive to Xbox post-Merger.

59. On this basis, we found that Microsoft would not have an incentive to make
CoD exclusive to Xbox post-Merger. As such, we believe the Merger may not
be expected to result in an SLC in console gaming services in the UK.

…about the future of cloud gaming

60. The evidence we found suggests that cloud gaming could be transformative
for the gaming industry in the next few years, helping to reach new customers
and improve choice for existing customers (potentially replacing expensive
consoles and gaming PCs altogether for some of them).

61. Cloud gaming has historically faced some unique challenges relative to
consoles. It requires users to have a fast and stable internet connection
capable of streaming graphically complex games. It must overcome latency
(ie, the time it takes for data to travel from a gaming device to a cloud server
and back), which can introduce delays and affect gameplay. The computing,
bandwidth, warehousing, and utilities costs associated with cloud gaming are
high, and this has led some in the industry to question whether it can ever be
profitable.

16
62. The evidence we have gathered indicates that cloud gaming service providers
already have, or soon will, overcome these challenges. In terms of demand,
as set out above the market is already sizeable, and the evidence indicates it
is poised to continue growing in the next few years. In terms of latency, some
providers noted that they have already successfully streamed graphically
complex games, such as CoD, with good results in terms of gameplay. As for
profitability, although providers have had mixed results and continue to
explore different avenues to monetise their service and gaming content, the
evidence suggests that costs will continue to fall as demand grows and
providers are able to scale their offering. These expectations are backed up by
considerable amounts of investment into this market by a range of market
participants.

…about Microsoft’s position in cloud gaming

63. Microsoft already holds a strong position in the gaming industry through its
established Xbox console, which has a large user base, and a strong
catalogue of gaming content. It has been steadily strengthening its gaming
ecosystem in line with the evolution of the gaming industry, including by
acquiring independent gaming studios (such as Bethesda in 2021), expanding
Game Pass, and developing its cloud infrastructure to better support its
gaming activities.

64. In relation to cloud gaming services, Microsoft has a combination of assets

that we consider is difficult for other cloud gaming service providers to match.
By owning Windows, the OS for which the vast majority of PC games are
designed, Microsoft could stream games from Windows servers without
having to pay a Windows licensing fee or adapt games designed for Windows
to an alternative OS. By having Xbox Cloud Gaming and Azure, Microsoft has
both a short-term and a longer-term solution to host cloud gaming, leveraging
its large and well distributed global cloud infrastructure to stream its games
without having to pay a fee to third-party cloud platforms. And by having an
existing console ecosystem, Microsoft has a range of popular games that it
can offer. As such, we consider that Microsoft has a strong position in cloud
gaming services and will remain an important competitor as the market
expands and evolves.

…about the impact of the Merger on cloud gaming

65. Our assessment under this theory of harm focused on whether Microsoft
would have the ability and incentive to limit access to Activision’s titles, and
whether this ‘foreclosure’ would impact rivals’ ability to compete with Microsoft
in cloud gaming services. In terms of ‘ability’, we considered whether limiting

17
 access to Activision’s games would harm the competitiveness of Microsoft’s
cloud gaming rivals. In terms of ‘incentive’, we considered whether Microsoft
stands to gain from this strategy. And in terms of ‘effect’, we considered how
this would impact overall competition in the market for cloud gaming services.

66. We have found that Activision’s games are likely to be important for the
growing market for cloud gaming services. Given that cloud gaming services
aim to achieve a similar quality of gameplay as consoles and gaming PCs, we
would expect customers’ preferences in cloud gaming to be similar to their
preferences in consoles and gaming PCs. As explained above, CoD is already
one of the most important games for consoles. We have also seen evidence
that CoD is a popular game in the PC market, and that it is consistently one of
the most requested titles by current cloud gaming users. A range of the
evidence that we gathered, including from multiple third parties, suggests that
CoD could make a material difference to the success of a cloud gaming
provider. And we found evidence that Activision has other games, such as
World of Warcraft (a PC-only game) and Overwatch that are currently popular
in consoles and/or gaming PCs and, as such, could also be important to cloud
gaming. Overall, therefore, we found that Activision’s titles are likely to be an
important input for the success of cloud gaming services, as they are today for
consoles and gaming PCs.

67. As to what Microsoft would do with Activision’s titles, we found that it would
have an incentive to make them exclusive to its cloud gaming service.

68. Cloud gaming is a relatively new market characterised by some elements of

direct and/or indirect network effects. In this type of market, success is highly
uncertain, and there’s an opportunity (and strong incentive) for incumbents to
develop a unique offering in a bid to gain market power (which can itself be
reinforcing as a result of network effects and scale advantages). One way for
Microsoft to achieve this would be to offer exclusive games on its cloud
gaming service. The evidence suggests that this is already part of Microsoft’s
cloud gaming strategy – except for a few Bethesda titles, most of which were
old or already available on rival cloud gaming services before Microsoft
acquired Zenimax, Microsoft has not made its games available on rival cloud
gaming platforms. We do not consider that the agreements Microsoft has
entered into with NVIDIA, Boosteroid and Ubitus change this, as we consider
there is material uncertainty around the scope, terms, and enforceability of
these agreements. They also apply only to a few existing cloud gaming
service providers, rather than to the full spectrum of actual and potential rivals.

69. We are concerned that making Activision’s titles exclusive to Microsoft’s cloud
gaming service would harm competition, particularly since our view is that
Microsoft already holds a strong position in this market by virtue of its

18
 unparalleled advantages through its ownership of Windows, its cloud
infrastructure, and its existing catalogue of first party titles. There are a few
emerging rivals with their own respective strengths, such as Amazon, Sony,
and NVIDIA, but none seem to be as well positioned as Microsoft in this
market. We consider that Google’s recent decision to shut down its own cloud
gaming service, Stadia, shows that merely having some strengths relevant to
cloud gaming is not enough to guarantee a platform’s success. The evidence
also indicates that there are significant barriers to entry and expansion,
including the cost of cloud infrastructure, the cost of acquiring content, and the
need for economies of scale in order to drive down costs. Since Microsoft
already appears to face limited competitive constraints from current and
potential rivals, we are concerned that withholding Activision’s content from
rival cloud gaming platforms is particularly likely to harm competition now and
in the foreseeable future.

70. On this basis, we found that the Merger may be expected to result in an SLC
in cloud gaming services in the UK, as a result of vertical effects in the form of
input foreclosure.

…about the overall impact of the Merger on consumers

71. Our statutory duty is to assess whether the Merger may be expected to result
in an SLC within any market or markets in the UK for goods or services. Any
such reduction in competition can have a potential impact on consumers.

72. In this case, we are concerned that the Merger will ultimately harm current and
future gamers. By stifling competition in the growing and dynamic market for
cloud gaming services, the Merger could alter the future of gaming. The
market for cloud gaming seems poised to grow and become an important
conduit for playing games, both for new users who are unable or unwilling to
buy an expensive console or gaming PC, and for existing gamers looking for
an alternative to these devices. Absent the Merger, strong competition in this
market could make cloud gaming better and more affordable for consumers.
By contrast, we found that the Merger would make an already strong
incumbent in this market even stronger, which could result in Microsoft
retaining a big share of the market and facing limited competition from current
and potential rivals. This reduction in competition could harm consumers, such
as by increasing prices and reducing quality, innovation, and choice over time.

What remedy did Microsoft offer?

73. To address our concerns, Microsoft offered the Microsoft Cloud Remedy.
Under this remedy, Microsoft committed to license Activision games, including
CoD and World of Warcraft, royalty-free to certain cloud gaming providers for

19
 a period of 10 years. Microsoft proposed to update the consumer licenses on
its website, giving the right to any consumer who acquired an Activision game
in one of the online stores designated by Microsoft to stream that game in the
cloud gaming services that were covered by the remedy. Microsoft offered to
appoint a monitoring trustee to monitor and seek to ensure Microsoft’s
compliance with the remedy, and a fast-track dispute resolution mechanism
carried out under arbitration.

74. The CMA’s guidance sets out the established position that behavioural
remedies are, due to their overall risk profile, unlikely to deal with an SLC and
its adverse effects as comprehensively as structural remedies. Behavioural
remedies can operate satisfactorily in limited circumstances, such as where
the company operates in a regulated environment, where there are expert
monitors, or where the SLC is expected to have a short duration. In this case,
the market for cloud gaming is a new and unregulated sector. We have
nevertheless engaged in a detailed assessment of the proposed Microsoft
Cloud Remedy, including through multiple discussions with the Parties and
third parties to establish whether this could constitute an effective remedy in
the specific circumstances of this case.

75. We found two significant limitations in scope for the Microsoft Cloud Remedy.

76. First, it was limited to a model whereby gamers had to first acquire the right to
play certain games (eg, by purchasing them on certain stores or subscribing to
them on certain services) in order to stream those games on certain cloud
gaming services. It did not make any provision for a different type of
commercial relationship between cloud gaming service providers and the
game publisher (ie, Activision). As such, it restricts the ability of cloud gaming
service providers to access Activision’s games through other strategies and
business models (some of which we already see in the cloud gaming market),
such as joint marketing arrangements, exclusive or early access to content, or
multi-game subscription services. In our view, and consistent with our
competitive assessment, this is a dynamic market in which there is a
reasonable chance that different providers will compete using a range of
different business models, and that these providers would have had access to
Activision’s content absent the Merger.

77. Second, the Microsoft Cloud Remedy applies to current and future PC and
console versions of Activision games. The PC versions are those that are
developed to run on a Windows OS, as well as other PC OS versions as may
be released by Microsoft during the term of the remedy. We found that, absent
the Merger, Activision would seek to maximise the value that it can derive from
these games, which would have involved considering making non-Windows
PC versions of its games (as it has already done in some cases). However,

20
 after the Merger, Microsoft’s incentives to make these games compatible with
rival OS would be significantly lower, as this would both increase the
attractiveness of rival cloud gaming services and divert demand away from
Windows OS. This means that, in effect, cloud gaming services wishing to
stream these games would have to use, or be compatible with, the Windows
OS version of those games. This could exclude or restrict providers that may
wish to provide cloud gaming services using other operating systems (such as
Linux), either now or in the future. The Microsoft Cloud Remedy would
therefore put non-Windows based cloud gaming services at a disadvantage,
and potentially distort the choice of OSs for new entrants.

78. We also found limitations in terms of the duration, monitoring, and

enforcement of the proposed remedy. The fact that the remedy is only for 10
years represents a clear weakness in terms of its effectiveness as a
comprehensive solution to the SLC, which is not itself time limited. Since the
remedy applies only to a defined set of Activision games, which can be
streamed only in a defined set of cloud gaming services, provided they are
purchased in a defined set of online stores, there are significant risks of
disagreement and conflict between Microsoft and cloud gaming service
providers. Given the information asymmetry between Microsoft and any
monitoring trustee or the CMA, it would be difficult to monitor and enforce this
remedy, even with significant information gathering. We found several
additional concerns with the practicalities of implementing the remedy, which
are detailed in the Final Report.

79. Based on this evidence, we found that the only effective remedy to the SLC
and its adverse effects was to prohibit the Merger.

Is it proportionate to prohibit this Merger?

80. The CMA seeks to ensure that no remedy is disproportionate in relation to the
SLC and its adverse effects. An effective remedy to an SLC, such as in this
case prohibition, could be considered disproportionate if it prevents customers
from securing benefits resulting from the Merger where this is disproportionate
to the scale of the SLC and its adverse effects. Insofar as these benefits
constitute RCBs, we take them into account when we assess whether a
remedy is proportionate.

81. Microsoft submitted that the Merger would give rise to a number of RCBs.
During our merger investigation, Microsoft entered into agreements with
different console and cloud gaming service providers to place its content
and/or Activision’s content on their respective platforms. These agreements
included a 10-year agreement with Nintendo to develop and publish future
native console versions of the CoD titles for Nintendo platforms post-Merger,

21
 as well as 10-year agreements with NVIDIA, Boosteroid, and Ubitus to make
Activision’s content available on their service. In addition, Microsoft submitted
that the Merger would give rise to RCBs as a result of (i) Microsoft placing
Activision content on Game Pass (Xbox and PC) on the date of release, and
(ii) Microsoft expanding into mobile gaming.

82. We found that most of these did not amount to RCBs under the Enterprise Act
2002 (the Act). In relation to the agreements with Nintendo and cloud gaming
services providers, we found that nothing about the Merger—such as any
potential changes in the market structure or commercial incentives that arise
from Microsoft and Activision ceasing to be distinct—would increase
Activision’s incentive to enter into these agreements relative to the situation
pre-Merger. To the contrary, being part of a corporate group that owns a
competing console (Xbox) and cloud gaming service (Xbox Cloud Gaming)
would suggest that Activision’s incentive to enter into these agreements would
be significantly reduced post-Merger. There is also considerable uncertainty in
the scope, enforceability, and potential benefits brought about by these
agreements. Microsoft itself acknowledged in the context of its agreements
with cloud gaming platforms that such a rapidly evolving market could give rise
to unanticipated and unforeseeable future events over a ten-year period
beyond its control.

83. In relation to Microsoft expanding into mobile gaming, the chances of

Microsoft succeeding seemed low in circumstances where the two largest
mobile OS—Google’s Android and Apple’s iOS—either currently prohibit rival
mobile gaming app stores or impose strict limits on their ability to monetise
content. In any event, there seemed to be other, less anti-competitive ways,
through which Microsoft could reasonably attempt to enter this market, such
as by licensing mobile gaming content from publishers.

84. We consider that bringing Activision’s content to Game Pass would amount to
an RCB under the Act. The Merger would bring Game Pass and Activision’s
content under common ownership, creating an opportunity to exploit synergies
and eliminate double marginalisation. And we believe it’s unlikely that
Activision would have made its most valuable content available on Game Pass
on the date of release absent the Merger.

85. We found that the scale of this benefit, however, would be limited. Having
Activision’s content on Game Pass would represent a different way to pay for
the same content, which would not necessarily be cheaper for all consumers.
We would also expect Microsoft to have the incentive to increase the price of
Game Pass commensurate with the value enhancement of adding Activision’s
valuable content to it, and we found that even a modest price increase would
significantly reduce or eliminate any potential RCB.

22
86. We recognise that having Activision’s content available on Game Pass is an
attractive prospect to some customers and something that, based on the
comments that we received from the public during this investigation, seems to
explain much of the support for this Merger by those in favour of it. But, on
balance, we found that having this new option to pay for content that is already
available on a buy-to-play basis on Xbox would not outweigh the overall harm
to competition (and, ultimately, consumers) arising from this Merger in the
sizeable and rapidly expanding market for cloud gaming services.

Conclusions

87. As a result of our investigation and our assessment, we concluded that the
anticipated acquisition by Microsoft of Activision would result in the creation of
a relevant merger situation.

88. We have also concluded that the Merger may be expected to result in an SLC
in the supply of cloud gaming services in the UK due to vertical effects
resulting from input foreclosure.

89. We have found that the Microsoft Cloud Remedy proposal would not be
effective in addressing the cloud gaming services SLC that we found.

90. We found that the only effective remedy to this SLC and its adverse
consequences is to prohibit the Merger. We also found that this remedy is
proportionate in relation to the SLC and its adverse effects, including taking
into account any RCBs.

23
 Findings

1. The reference
1.1 On 15 September 2022, the Competition and Markets Authority (CMA) in
exercise of its duty under section 33(1) of the Enterprise Act 2002 (the Act),
referred the anticipated acquisition by Microsoft Corporation (Microsoft) of
Activision Blizzard, Inc. (Activision) (the Merger) for further investigation and
report by a group of CMA panel members (the Inquiry Group). Microsoft and
Activision are referred to collectively as the Parties or, for statements
referring to the future, the Merged Entity.

1.2 In exercise of its duty under section 36(1) of the Act, the CMA must decide:

(a) whether arrangements are in process or contemplation which, if carried

into effect, will result in the creation of a relevant merger situation (RMS);
and

(b) if so, whether the creation of that RMS may be expected to result in a
substantial lessening of competition (SLC) within any market or markets
in the United Kingdom (UK) for goods or services.

1.3 In assessing the competitive effects of the Merger, we must decide whether
there is an expectation (ie a more than 50% chance) that the Merger will
result in an SLC.

1.4 We are required to prepare and publish our final report by 26 April 2023.

1.5 Our terms of reference, along with information on the conduct of the inquiry,
are set out in Appendix A and Appendix B respectively.

1.6 This document, together with its appendices, constitutes the CMA’s Final
Report published and notified to the Parties in line with the CMA’s rules of
procedure. 1 Further information relevant to this inquiry can be found on the
CMA case page. 2

1 CMA rules of procedure for merger, market and special reference groups (CMA 17), Rule 13.
2 Microsoft/Activision Blizzard Case Page.

24
2. The Parties, the Merger, and the rationale
2.1 This chapter sets out:

(a) an overview of the Parties and their financial information;

(b) the background to the Transaction, including the valuation; and

(c) the Parties’ stated rationale for the Merger.

The Parties

Microsoft

Principal activities

2.2 Microsoft is a global technology company founded in 1975 and headquartered

in Redmond, Washington, US. 3 Microsoft is publicly listed on Nasdaq.
Microsoft’s global turnover in the financial year 2021 was close to £125 billion,
of which [] was generated in the UK. 4

2.3 Microsoft is organised into three operating segments: (i) Productivity and
Business; (ii) Intelligent Cloud; and (iii) More Personalised Computing. 5
Microsoft offers a wide range of products and services including:

(a) Windows OS. Microsoft Windows is a computer OS that can be installed

on a personal computer (PC) or server to provide a graphics-based
interface between the user and the computer’s hardware and software.
Over the years, Microsoft has released various versions of Windows
aimed at improving on features like speed and user interface. 6 Microsoft
offers two types of licences for Windows – Windows for desktop PCs
(Windows Client) and Windows for servers (Windows Server) []. 7

(b) Azure. Azure is Microsoft’s public cloud platform and associated services.
Azure offers over 200 Infrastructure-as-a-Service (IaaS) and Platform-as-
a-Service (PaaS) solutions including computing, storage, networking,
databases, operating systems, developer tools, and runtimes, to help
enterprises build and run their systems, analytics, and applications in the

3 Parties Final Merger Notice (FMN).

4 Parties FMN.
5 Parties FMN.
6 ‘From Windows 1 to Windows 10: 29 years of Windows evolution’, dated 2 October 2014, accessed on 4

August 2022.
7 Parties response to the CMA’s request for information (RFI).

25
 cloud. Customers pay consumption-based fees for the services they use. 8
Within Azure, Microsoft offers Azure PlayFab, a back-end platform for live
games, providing managed game services, real-time analytics, and live
operations, which enables game developers to build and operate games,
analyse gaming data, and improve overall gaming experiences. 9
Examples of games that run on Azure PlayFab include first-party games
such as Minecraft, Forza Horizon 4, Doom Eternal, and Microsoft Flight
Simulator, as well as some third-party games including Roblox, Astroneer
and Wasteland 3. 10

(c) Xbox Cloud Gaming. Microsoft currently offers cloud-based game

streaming through Xbox Cloud Gaming, which is composed of dedicated
Xbox consoles located in Microsoft data centres. 11 This is distinct from
Azure. Microsoft has deployed around [] Xbox console motherboards
worldwide across its data centres to provide Xbox Cloud Gaming.
Microsoft initially submitted []; 12 however, Microsoft has since noted
that it []. 13

(d) Xbox. Xbox is Microsoft’s gaming console. It connects to a television or

other display and allows users to play games specifically developed for
Xbox. Xbox first launched in 2001 and has since remained one of the
three main gaming consoles in the market (along with Sony Interactive
Entertainment’s (SIE) and Nintendo’s consoles). 14

(e) Xbox Game Studios. Microsoft is active as a developer, publisher, and

distributor of games. Microsoft publishes games for PCs, consoles and
mobile devices developed by Xbox Game Studios, a collection of 24 first-
party development studios, including the recently acquired ZeniMax
studios. Examples include games in the Minecraft, Forza, Elder Scrolls
and Halo game titles. 15

(f) Digital distribution. Microsoft distributes games in digital form. 16 Microsoft

operates the Microsoft Store on Windows (the Microsoft Store), an app
store on Windows PCs, through which it distributes its own first-party
games and third-party games for PC, as well as an Xbox-branded

8 Parties FMN.
9 Parties FMN.
10 Parties FMN.
11 Parties FMN.
12 Parties FMN.
13 Microsoft site visit.
14 Parties FMN.
15 Parties FMN.
16 Microsoft also distributes games in physical form through third parties, but does not have ‘bricks-and-mortar’

retail outlets in the UK.

26
 storefront (the Xbox Store), which can be accessed via an Xbox console,
web-browser, or the Xbox App for Windows. 17

(g) Gaming Subscription Services. Microsoft offers multi-game subscription

(MGS) services that include access to first- and third-party games (eg
Xbox Live Gold and Xbox Game Pass), online multiplayer capabilities (eg
Xbox Live and Xbox Live Gold) and cloud gaming functionality (Xbox
Cloud Gaming, which is available as part of the Xbox Game Pass top-tier
subscription and on a free-to-play (F2P) basis with Fortnite). 18

2.4 Microsoft publishes PC and console games developed by its 24 first-party

game development studios, as well as by second- and third-party
developers. 19

Key financials

2.5 Microsoft generated $198.1 billion USD in revenue in the 2022 financial year,
an 18% increase on the $168.1 billion USD earned in 2021. 20

2.6 As noted above, Microsoft is split into three business segments. Table 2.1
below sets out the revenue and operating income earned by Microsoft by
business segment in the years to June 2019, 2020, 2021 and 2022.

Table 2.1: Microsoft’s revenue and operating income split by business segment from 2019 to
2022

$m

2019 2020 2021 2022

Revenue
Productivity and Business
Processes 41,160 46,398 53,915 63,364
Intelligent Cloud 38,985 48,366 60,080 75,251
More Personal Computing 45,698 48,251 54,093 59,655
Total 125,843 143,015 168,088 198,270
Operating income
Productivity and Business
Processes 16,219 18,724 24,351 29,687
Intelligent Cloud 13,920 18,324 26,126 32,721
More Personal Computing 12,820 15,911 19,439 20,975
Total 42,959 52,959 69,916 83,383

Source: Microsoft, Annual Report 2021 ‘Summary Results of Operations’, accessed 3 November 2022 and Microsoft, Annual
Report 2022 ‘Summary of Results of Operations’, accessed 18 January 2023.
Note: Revenue represents the total revenue earned by each of Microsoft’s segments, while operating income represents the
profits after operational costs.

17 Parties FMN.
18 Parties FMN.
19 Parties FMN.
20 Microsoft, Annual Report 2022 ‘Summary of Results of Operations’, accessed by the CMA on 18 January

2023.

27
2.7 Microsoft’s gaming activities sit within the More Personal Computing business
segment. Microsoft noted that the increase in revenues earned by the More
Personal Computing segment was driven by gaming in the year to 30 June
2021. 21

2.8 In its 2021 Annual Report, Microsoft noted the growth in gaming revenue,
explaining that: (i) gaming revenue increased $3.8 billion or 33% in FY21
driven by growth in Xbox content and services and Xbox hardware; (ii) Xbox
content and services revenue increased $2.3 billion or 23% driven by growth
in third-party titles, Xbox Game Pass subscriptions, and first-party titles; and
(iii) Xbox hardware revenue increased 92% driven by higher price of consoles
sold due to the Xbox Series X|S launches. 22

2.9 Growth in gaming continued in the year to 30 June 2022. Microsoft noted that
Gaming revenue increased $860 million, or 6%, on a prior year comparable.
Microsoft explained that Xbox hardware revenue increased 16% due to
continued demand for Xbox Series X|S. Xbox content and services revenue
increased 3% driven by growth in Xbox Game Pass subscriptions and first-
party content, offset in part by a decline in third-party content. 23

2.10 Table 2.2 below further breaks down the revenues earned by Microsoft’s
gaming business unit in the years to June 2019, 2020 and 2021.

Table 2.2: Microsoft Gaming revenues from 2019 to 2021 (year to June)

[]
Source: CMA analysis of: []
*[]

Activision

Principal activities

2.11 Activision is a game developer and publisher founded in 2008 and

headquartered in Santa Monica, California, US. 24 Activision is publicly listed
on Nasdaq. Activision’s global turnover in the financial year 2021 was over £6
billion, of which approximately £716 million was generated in the UK. 25

2.12 Activision is active in the following areas:

21 Microsoft, Annual Report 2021 ‘Summary Results of Operations’, accessed by the CMA on 3 November 2022.
22 Microsoft, Annual Report 2021 ‘Summary Results of Operations’, accessed by the CMA 3 November 2022.
23 Microsoft, Annual Report 2022 ‘Summary Results of Operations’, accessed by the CMA 18 January 2023.
24 Parties FMN.
25 Parties FMN.

28
 (a) Game development and publishing. Activision develops games for PCs,
consoles, and mobile devices, and publishes them in most countries
around the world through three business units: (i) Activision Publishing,
Inc (the Activision segment); (ii) Blizzard Entertainment, Inc (the
Blizzard segment); and King Digital Entertainment (the King
segment). 26

(b) Digital distribution. In Europe, Activision provides warehousing, logistics,

and sales distribution services to third-party publishers of interactive
entertainment software and interactive entertainment hardware (as well as
its own publishing operations). Activision also has an online gaming digital
storefront for PC games, Battle.net, which facilitates digital distribution of
Blizzard and select Activision content. 27

(c) Display advertising. Activision operates digital display advertising within

some of its game content, particularly within mobile games offered by
King. 28

Key financials

2.13 Activision earned total revenues of $8.8 billion USD in the financial year
ended December 2021, a slight increase on the $8.1 billion USD in the year to
2020. 29

26 Parties FMN.
27 Parties FMN.
28 Parties FMN.
29 Activision Blizzard, 2021 Annual Report ‘Activision Blizzard, Inc. and Subsidiaries Consolidated Statement of

Operations’, page F-4 (78 of document).

29
Table 2.3: Activision net revenues and operating income by segment in the years ended 31
December 2019, 2020, and 2021 30

$m

2019 2020 2021

Revenue
Activision segment 2,219 3,942 3,478
Blizzard segment 1,719 1,905 1,827
King segment 2,031 2,164 2,580
Total 5,969 8,011 7,885
Operating income
Activision segment 850 1,868 1,667
Blizzard segment 464 693 698
King segment 740 857 1,140
Total 2,054 3,418 3,505

Source: Activision, Annual Report and Accounts for 2019, 2020 and 2021.
Note: Revenue represents the total revenue earned by each of Activision’s segments, while operating income represents the
profits after operational costs.

2.14 Table 2.3 shows Activision’s net revenues and operating income by operating
segment in the three years ended 31 December 2019, 2020, 2021. The
Activision segment (which produces Call of Duty (CoD) consistently
generated the most revenue and operating income across the three-year
period, followed by King (which produces Candy Crush). Revenue peaked in
the year to 31 December 2020 – despite a slight decline in revenue in 2021,
the results remained significantly higher than in 2019.

Table 2.4: Activision consolidated net revenues by platform in the years ended 31 December
2019, 2020, and 2021

$m

2019 2020 2021

Console 1,920 2,748 2,637

PC 1,718 2,056 2,323
Mobile and ancillary 31 2,203 2,559 3,182
Other 32 648 687 661
Total 6,489 8,086 8,803

Source: Activision, Annual Report and Accounts for 2019, 2020 and 2021.

2.15 Table 2.4 demonstrates that console gaming was the largest revenue platform
for Activision in 2019 and 2020, but this was overtaken by mobile gaming in

30 Note that the segment net revenues set out in Table 2.3 differ from the consolidated net revenues in Tables 2.4
and 2.5. Segment revenues set out in Table 2.3 do not include revenues from non-reportable segments (other
income and expenses outside of reportable segments, including Activision’ Distribution business and unallocated
corporate income and expenses), share based compensation expenses, restructuring costs, and other
accounting items (eg amortisation of intangible assets and recognition (or deferral) of deferred revenue).
Reconciliations between segment net revenues and consolidated net revenues for 2021 and 2020 can be found
on page 47 of the 2021 Annual Report and Accounts, and on page 41 of the 2019 Annual Report and Accounts
for 2019.
31 Net revenues from ‘mobile and ancillary’ include revenues from mobile devices as well as non-platform specific

game related revenues, such as standalone sales of toys and accessories.

32 Net revenues from ‘other’ primarily includes revenues from the Distribution business, the Overwatch League,

and the Call of Duty League.

30
 2021. Console gaming saw a slight decline in 2021, but the results do not yet
point to any continued change in revenue source.

Table 2.5: Activision consolidated statement of operations for the years ended 31 December
2019, 2020, and 2021

$m

2019 2020 2021

Net revenues 6,489 8,086 8,803

Costs and expenses 4,882 5,352 5,544
Operating income 1,607 2,734 3,259
Net income 1,503 2,197 2,699

Source: Activision, Annual Report and Accounts for 2019, 2020 and 2021.

2.16 In line with Table 2.3 and Table 2.4, Table 2.5 shows an overall increase in
Activision’s revenues from 2019 to 2021, which flows through to an overall
increase in both operating and net income.

The Merger

2.17 On 18 January 2022, Microsoft entered into an agreement with Activision, via
its direct wholly owned subsidiary Anchorage Merger Sub Inc., to acquire sole
control of Activision (as defined above, the Merger). 33 Under the terms of this
agreement, Microsoft agreed to pay USD 95 per share, representing a
purchase price of approximately USD 68.7 billion. 34

2.18 The Parties informed the CMA that the Merger is also the subject of review by
competition authorities in a number of other jurisdictions, including Australia,
Brazil, [], [], the EU, Japan, [], South Korea, and the US. 35 In August
2022, the Saudi Arabian General Authority for Competition cleared the
merger, 36 followed by the Brazilian Administrative Council for Economic
Defense (CADE) on 5 October and Chile's Fiscalía Nacional Económica
(FNE) on 29 December. 37 In March 2023, the Japan Fair Trade Commission
(JFTC) cleared the Merger. 38

33 Parties FMN. See here: press release issued by Microsoft.

34 Parties FMN.
35 Parties FMN.
36 Saudi Arabia General Authority for Competition on Twitter, Certificate of No Objection, 22 August 2022,

accessed by the CMA on 7 November 2022.

37 CADE, ‘CADE clears Microsoft's acquisition of Activision Blizzard’, 7 October 2022, accessed by the CMA on 7

November 2022 and FNE, ‘FNE aprueba la adquisición de Activision Blizzard por parte de Microsoft Corporation,’
29 December 2022, accessed by the CMA on 2 February 2023.
38 JFTC, press release, accessed by the CMA on 28 March 2023.

31
Valuation

2.19 Microsoft’s valuation model of January 2022 estimated Activision’s value to

Microsoft to be $[]. This []. 39

Standalone valuation (ie Activision’s current business)

2.20 In determining the ‘standalone’ value of Activision, Microsoft used []. 40

2.21 In determining the relevant cashflows for its valuation, Microsoft estimated
[], as set out in Table 2.6 below.

Table 2.6: []

[]
Source: Microsoft, Annex to the FMN.

2.22 Microsoft estimates that [] Activision operating segment revenue

[] 41[] 42.

2.23 In 2021, []% of the Blizzard operating segment’s revenues were generated
by [] content with [] distributed across its other franchises (eg [], []
and []). The proportion of Blizzard's revenue earned by these games is
expected to change by [], with [] dropping to approximately []% and
most of the remaining revenue being generated by []. 43

2.24 The King operating segment revenues are []. Microsoft expects []. 44

Synergies arising from the Merger

2.25 Microsoft identified the following [] synergies as being key value drivers for
the Merger:

(a) ‘[]. 45

(b) []:

(i) [];

39 Microsoft, Annex to the FMN.

40 Microsoft, Annex to the FMN.
41 Note that Table 2.3 sets out []. Source: Microsoft, Annex to the FMN and Microsoft, email to the CMA.
42 Microsoft, Annex to the FMN.
43 Microsoft, Annex to the FMN.
44 Microsoft, Annex to the FMN.
45 Microsoft, Annex to the FMN.

32
 (ii) []; and

(iii) []. 46

(c) []. 47

The rationale

Parties’ submissions

2.26 The Parties told the CMA that Microsoft’s rationale for the Merger is to:

(a) provide Microsoft with gaming content (including popular Activision

franchises like CoD, World of Warcraft (WoW), and Candy Crush Saga)
which will help Microsoft to execute a cross-platform strategy (allowing
gamers to play games on multiple devices); 48

(b) improve Microsoft’s presence in the mobile segment, where Activision

holds an established position (particularly through King); 49

(c) support Microsoft’s investments in its multi-game subscription service,

Xbox Game Pass 50 (XGP), and improve user engagement/adoption
amongst Xbox and PC users; 51

(d) improve Microsoft’s ability to create a ‘Universal Store’ (extending the

Xbox digital storefront across non-Xbox platforms and devices); 52 and

(e) increase the attractiveness of MSAN. 53

2.27 Microsoft described the acquisition as supporting the opportunity to ‘pivot

away from the device centric strategy of the past’ and ‘focus instead on a
consumer-centric, cross-platform approach that will allow gamers to play
wherever and on whatever device they wish.’ 54 Microsoft’s internal documents
from before and at the time of the Merger broadly support the rationale stated

46 Microsoft, Annex to the FMN.

47 Microsoft, Annex to the FMN.
48 Parties FMN.
49 Parties FMN.
50 For completeness, the CMA notes that Microsoft offers a number of pricing options for its subscription services,

for users on both PC and consoles, as well as a bundled subscription and cloud gaming offering (Xbox Game
Pass Ultimate). Throughout this report, references to Xbox Game Pass refer to the various offerings as a
collective, unless otherwise specified.
51 Parties FMN.
52 Parties FMN.
53 Parties FMN.
54 Parties FMN.

33
 above, with a particular focus on acquiring a broad range of differentiated
gaming content to help scale XGP. 55

2.28 Microsoft placed significant emphasis on the importance of mobile gaming to

its ongoing growth and presence in the gaming market in its submissions,
noting that [], and has identified this as a key driver of the transaction. 56
Microsoft’s focus on expanding its presence in mobile is reflected in internal
documents. 57

55 For example, Microsoft Internal Document; and Microsoft Internal Document.

56 Microsoft site visit.
57 For example, see Microsoft Internal Document.

34
3. Relevant merger situation
3.1 In accordance with section 36 of the Act and pursuant to our terms of
reference we are required to investigate and report on two statutory
questions:

(a) whether arrangements are in progress or in contemplation which, if

carried into effect, will result in the creation of an RMS; and

(b) if so, whether the creation of that situation may be expected to result in a
SLC within any market or markets in the UK for goods or services.

3.2 We address the first of the statutory questions in this section.

Enterprises ceasing to be distinct

3.3 The first element of the RMS test is whether the arrangements in progress or
contemplation will, if carried into effect, lead to enterprises ceasing to be
distinct. 58

3.4 The Act defines an ‘enterprise’ as ‘the activities or part of the activities of a
business’. A ‘business’ is defined as including ‘a professional practice and
includes any other undertaking which is carried on for gain or reward or which
is an undertaking in the course of which goods or services are supplied
otherwise than free of charge.’ 59

3.5 We have described the activities of Microsoft and Activision in Chapter 2

above. In light of this, we are satisfied that Microsoft and Activision are
‘businesses’ and that they each satisfy the definition of ‘enterprise’ in
accordance with the Act.

3.6 Section 26 of the Act provides that any two enterprises cease to be distinct if
they are brought under common ownership or common control. The
background to the Merger is described in Chapter 2 above. On completion of
the Merger, Activision will be under the common ownership and control of
Microsoft.

3.7 We are therefore satisfied that the Merger will result in Microsoft and
Activision ceasing to be distinct enterprises for the purposes of the Act.

58 Section 23 and section 24 of the Act.

59 Section 129(1) and (3) of the Act.

35
Jurisdiction test

3.8 The second element of the RMS establishes whether the Merger has
sufficient connection with the UK on a turnover or share of supply basis to
give the CMA jurisdiction to investigate.

3.9 The turnover test is satisfied where the value of the turnover in the UK of the
enterprise being taken over exceeds £70 million. As noted in Chapter 2,
Activision’s turnover in the UK exceeded £70 million in FY2021. We are
therefore satisfied that the turnover test in section 23(1)(b) of the Act is met.
As we have concluded that the turnover test is met, there is no need to
consider the share of supply test.

Conclusion on the relevant merger situation

3.10 In the light of the above, we have concluded that the Merger constitutes
arrangements in progress or in contemplation which, if carried into effect, will
result in the creation of a relevant merger situation.

36
4. Industry background

Introduction

4.1 This chapter sets out background information on the gaming industry relevant
to our assessment of how Microsoft and Activision operate in the industry.
The chapter is split into the following subsections:

(a) Overview of the industry;

(b) Game development, publishing, distribution and marketing;

(c) Gaming hardware;

(d) Game purchasing; and

(e) Cloud gaming.

Overview of the industry

4.2 The gaming industry is the UK’s largest revenue-generating form of

entertainment. It is bigger than pay TV, home video (including streaming),
cinema, music, or books. In 2022, it generated approximately £5 billion in
revenue in the UK. 60

4.3 Gamers can play games on PCs, gaming consoles or mobile devices. 61
Games played on mobile devices tend to be less demanding games in terms
of processing power.

4.4 Games are developed (ie conceived, designed, programmed and tested) by
game studios. While game studios and publishers may be different entities,
there can be, and often is, common ownership across the two. Once
developed, games are licensed, marketed, and released in different regions
by publishers. While game studios and publishers may be different entities,
there can be, and often is, common ownership across the two. Games are
distributed to gamers across various channels and platforms. 62

4.5 Games can be monetised in different ways. Over the past several years,
gamers have typically accessed games by paying an up-front fee and
downloading the relevant games from a digital storefront (such as the Xbox

60 BBC, Almost 90% of games sold in UK in 2022 were digital - ERA, 10 January 2023, accessed by the CMA on
30 January 2023.
61 Parties FMN.
62 Parties FMN.

37
 Store) to their console or device (such as a PC or mobile). This ‘buy-to-play’
(B2P) model remains the primary mode of delivering games on console.
Some console providers also monetise content by offering multi-game
subscription services. Unlike the traditional B2P model, these services allow
gamers to access a catalogue of games for a fixed, often monthly, fee. In both
B2P and multi-game subscription services, there are some games that allow
users to make in-game purchases, meaning they can purchase access to
additional features in the game. The ‘in-game’ purchasing model is a key
feature of free-to-play games which are free to access but generate revenue
from in-game purchases.

4.6 The gaming industry is characterised by strong network effects. Console

providers such as Microsoft compete to attract users who want to play high-
quality games, often with friends, as well as high-quality content from game
developers, who want to make games for consoles with a large user base.
Console platforms with a lot of gamers attract better content, which in turn
attracts more gamers to that console platform, which in turn attract better
content, and so on.

4.7 In recent years, the industry has seen the development of cloud gaming
services, a technology that allows complex games to be accessed on remote
servers and streamed directly to a device. Since games are executed
remotely, gamers can play using a range of devices that can be less powerful,
and are often cheaper, than consoles (such as mobile phones or tablets).
There have been several recent entrants into the gaming industry using this
disruptive technology, including Amazon Luna, Google Stadia, 63 Blacknut,
Boosteroid and NVIDIA GeForce Now. These entrants are exploring different
ways to monetise their offering, including through ‘B2P, ‘bring-your-own-
game’ (BYOG) (whereby customers can play games purchased on third-party
storefronts on the relevant cloud gaming service), and multi-game
subscription models.

Game development, publishing, distribution and marketing

4.8 This section sets out a summary of the lifecycle of a game from development
stage through to game publishing, distribution and marketing.

63Note that Google Stadia was shut down on 18 January 2023. Source: Google, ‘Thank you for playing with us’,
accessed by the CMA on 31 January 2023.

38
Game Development

4.9 The first stage in the game lifecycle is game development, which relates to
the development and production of the game. This includes the design, art,
programming and testing of a game, and is typically done by one or more
development studios.

4.10 Game development involves the use of software development tools (including
‘game engines’ and other tools such as audio and video middleware) which
may be created by a studio in-house or licensed from third parties. The core
functionality typically provided by a game engine includes a rendering engine
for 2D or 3D graphics, a physics engine (including collision detection and
collision response), sound, scripting, animation, networking, streaming,
memory management, threading, localisation support, scene graph, and
sometimes video support for cinematics. These different tools result in
different gaming experiences, in particular between PC/console games as
compared to native mobile games. Typically, PC and console games offer a
more advanced gaming experience and require more advanced game
development tools. 64

4.11 Developers may be first-, second-, or third-party:

(a) First-party developer: a first-party developer refers to development

completed internally by the game publisher, a subsidiary, or a fully
integrated studio, and then made available to the public by that developer;

(b) Second-party developer: a second-party developer is ordinarily a game

studio that enters into development contracts with a publisher, such as
SIE, Nintendo or Microsoft, and develops games exclusively for that
publisher;

(c) Third-party developer: a third-party developer is an independent

developer that develops and owns games that are then published by a
game publisher. 65

Game Publishing

4.12 Following game development, game publishers make video games available
to the public for sale or for free. They are responsible for licensing the rights in
relation to the game as well as for handling the advertising, marketing and
distribution of the game.

64 Parties FMN.
65 Parties FMN.

39
Game Distribution

4.13 Game distribution has traditionally occurred through gamers purchasing a

hard copy of their games on physical media from a retail store. There has
been a significant shift in recent years to games being purchased digitally. In
both instances, software is executed locally on the player’s gaming device.

4.14 Gamers may also access games via a streaming service. In this case, the
gaming software is executed remotely on cloud infrastructure and streamed
over the internet to the player’s end device. The end device can, in principle,
be anything from a mobile phone to a smart TV.

4.15 Some providers offer multi-game subscription services that allow gamers to
purchase access to a number of games for a fixed monthly fee (multi-game
subscriptions). Depending on the subscription, gamers may download or
stream these games.

Game Marketing

4.16 Games are often marketed by reference to their type or their genre:

(a) Game type: Definitions of game types are imprecise, but typical
categorisations used in the industry are: ‘AAA’, ‘AA’, and Indie.

(i) AAA games tend to be considered to be those which are typically

developed by large development studios and require significant
budget and time (eg a number of years). AAA status is not a well-
defined term, but in industry discussions is correlated with its budget,
complexity, popularity, or some combination of the three.

(ii) AA games tend to have significantly smaller budgets. While they are
still developed by large groups of developer staff that are spread
across multiple offices, they tend not to have the scale and reach of
AAA games.

(iii) Indie games are those developed by companies that are typically
small, self-funded and ‘independent’ from broader parent companies
which dictate the direction of the game developers. Indie studios are
often made up of smaller teams and can often depend on
crowdfunding and donations to fund their games.

(b) Genre: Games may also be classified by reference to their genre,

definitions of which can be broad. Some common genres of games
include: action; adventure; role-playing games (RPGs), shooter, sport,
strategy, open world, and resource management. Games are sometimes

40
 recognised as belonging to a genre, but several games have elements of
multiple genres, which can make game classification to some extent a
subjective exercise. Genres can be broken down into sub-genres; for
example, the shooter genre includes: first-person shooters, third-person
shooters or battle royale games.

Gaming hardware

Gaming devices

4.17 Gamers have historically used the hardware within one (or more) of the
following sub-categories in order to play games:

(a) PCs (eg desktops, laptops and gaming-specific PCs);

(b) Dedicated gaming consoles (eg SIE PlayStation, Nintendo Switch, and
Microsoft Xbox); and

(c) Mobile devices.

PCs

4.18 Gaming PCs are personal computers designed for playing video games and
are manufactured in such a way as to improve the gaming experience as
compared to gaming on a mainstream PC. The specifications and
components of a gaming PC make it more suited to delivering a high-quality
gaming experience, as compared to a mainstream PC.

Gaming consoles

4.19 The Parties estimate that worldwide revenues from console hardware sales
were [] in 2021, with [] generated in the UK. 66

4.20 Consoles were initially released by Atari 67 and Nintendo in the 1970s and
1980s, followed by SIE in 1995. Microsoft followed in 2001. Consoles are
currently in what is known as the ‘ninth generation’, 68 with SIE, Nintendo and
Microsoft being the leading suppliers. 69

66 Parties FMN.
67 Atari exited the console gaming market in the early 1980s.
68 Parties FMN.
69 Parties FMN.

41
Figure 4.1 Console generations from 1972 – 2022 by developer

Source: Parties FMN.

4.21 Recent generations of gaming consoles function as independent media hubs

with the ability to download and stream content, such as Netflix and Amazon
Prime Video. Generally, there is a period of around five to seven years
between consoles. []. 70 Older consoles tend to be supported for a period of
time following the release of a new generation console. 71 Consoles and PCs
can usually process larger and more complex games (such as Call of Duty).

Mobile devices

4.22 In addition to consoles and PCs, people play games on mobile devices.
Mobile devices currently lack the technical capabilities to run most console
games, and most people use them to play more casual games (such as
Candy Crush). We note that some console / PC games have developed
mobile versions, which are simplified versions of the game with different game
modes and functionality to their console/PC counterparts.

Supporting infrastructure

4.23 This section sets out the key supporting infrastructure in the development of
video games, including: (i) operating systems; (ii) Graphics Processing Units;
and (iii) network infrastructure for cloud gaming.

70 Microsoft response to the CMA’s RFI.

71 Parties FMN.

42
Operating systems

4.24 Gaming hardware requires an Operating System (OS). An OS provides a

graphics-based interface between the user and the device’s hardware and
software. 72

4.25 Microsoft owns Windows OS, which is the market leader in the supply of PC
OSs. Microsoft’s rivals, which are significantly smaller, include MacOS,
ChromeOS, and Linux-based OSs. 73 Because of the popularity of Windows
OS, game developers most commonly make games that are designed and
optimised for Windows OS. 74 Gaming consoles operate with exclusive and
proprietary console operating systems.

Graphics Processing Units

4.26 Graphics Processing Units (GPUs) are specialised processors designed to

accelerate graphics rendering, thereby improving the quality of gaming
visuals. In recent years, video games have become more computationally
intensive, with some featuring hyper-realistic graphics or vast and
sophisticated in-game worlds. With advanced display technologies (such as
4k screens) and the rise of virtual reality gaming, the demands on graphics
processing have grown significantly and continue to do so. With better
graphics performance, games can be set to a higher resolution, at faster
frame rates, or both. The leading suppliers of GPUs are AMD and NVIDIA.

Network infrastructure for cloud gaming

4.27 Cloud infrastructure to deliver cloud-based game streaming enables gamers

to access games across a range of different endpoints or devices. It is
therefore a critical asset in the provision of cloud gaming. A provider may
build their own cloud streaming infrastructure, partner with a cloud computing
provider to do so, or use a white-label service provided by a third-party
provider. Providers such as Microsoft, Amazon, Google, SIE, NVIDIA and
others have built their own infrastructure. 75 We set out further detail on the

72 Parties response to the CMA’s RFI.

73 Linux is an operating system engine which is free to use and modify resulting in multiple Linux based operating
systems. Windows’ share of user personal computer OSs is [70-80%] worldwide and [60-70%] in the UK (Source:
Parties response to the CMA’s RFI). The hypothetical market share of OS software for PCs used for gaming is
even higher than this, upwards of 95% (‘Steam Hardware & Software Survey: July 2022’, July 2022, accessed by
the CMA on 15 August 2022).
74 Note that Valve Software has released ‘Proton’, a compatibility layer which allows games developed on the

Microsoft OS to be run on other Linux-based operating systems.

75 Parties’ response to RFI.

43
 ecosystem of cloud gaming infrastructure within the ‘cloud gaming’ section
below.

Game purchasing

4.28 Gamers have traditionally purchased games under a B2P model in which they
pay a one-time upfront fee for each individual game. Increasingly, games
which are purchased upfront also offer additional purchasable content (eg
expansion packs or in-game purchases). For example, gamers will pay an
upfront amount for the core Call of Duty game but may make additional
payments as they play the game, eg to purchase in-game currency which they
can exchange for additional content, upgraded ammunition etc. 76 In 2021,
Activision earned 26% of revenues from product sales and 60% of revenues
from in-game purchases (both B2P and F2P games). The remaining 14% of
revenues were categorised as relating to subscriptions and other sources. 77

4.29 Some providers are now offering subscription services, which require ongoing
payments to play games. One popular subscription option is multi-game
subscription services. These enable gamers to: (i) access a catalogue of
games, which gamers can download or stream, effectively renting access to
games; and/or (ii) play with each other online in the form of multi-player
experiences across multiple games (although multi-player is not unique to
multi-game subscriptions).

4.30 Subscription services currently include: Microsoft Xbox (Xbox Game Pass and
Xbox Live Gold), Sony PlayStation Plus, Nintendo Switch Online, EA Play,
Apple Arcade, Google (Play Pass and Google Stadia 78), Amazon (Luna 79 and
Prime Gaming), Ubisoft Plus, Netflix Gaming, Utomik, and Blacknut.

4.31 However, for the purpose of our assessment we focus on multi-game

subscription services that offer a relatively wide range of games (eg Game
Pass, PlayStation Plus (PS+) Extra and PS+ Premium), rather than services
which mainly offer access to online multiplayer and/or only a few monthly
games (eg Xbox Live Gold and PS+ Essential).

76 Parties, FMN.
77 CMA analysis of: Activision 2021 Annual Report, accessed by the CMA on 18 January 2023. Product sales of
$2,311 million revenue (page F-4); in-game net revenue of $5,266m (page 45); subscription and other revenue of
$1,226 million (calculated as the difference between product sales and in-game net revenue and total revenue, of
$8,803 million). Note that this relates to revenues across the entirety of Activision content, including mobile as
well as console, meaning that in-game purchase percentages are likely to be higher than in the case of console
gaming only. This is due to the majority of revenue on mobile games being generated from in-game purchases
rather than upfront purchase costs.
78 As noted above, on 18 January 2023 Google closed down its Stadia service.
79 Amazon Luna is not currently available in the UK.

44
Cloud gaming

Overview of cloud gaming

4.32 Cloud-based game streaming services (cloud gaming services) are

consumer-facing services which allow games to be streamed over the internet
from gaming hardware in a data centre to a gamer’s choice of supported
device. This differs from traditional gaming on PCs, mobile, and console,
where customers are limited to playing games that have been downloaded to
that specific piece of hardware. Cloud gaming provides gamers the
opportunity to play technologically complex games on less powerful devices—
such as mobile devices—that may otherwise lack the computing power or
storage to support them. 80 There are several existing cloud gaming services
available to the public today. It is, however, a market that has emerged only in
recent years, and industry participants expect it to grow and develop
considerably. In that sense, cloud gaming is still a nascent market. 81

4.33 Microsoft offers cloud gaming services through Xbox Cloud Gaming (xCloud).
This is available to gamers as a bundled offering with its Xbox Game Pass
Ultimate (XGPU) multi-game subscription service, as well as separately on a
free trial basis for one game, Fortnite. 82 Activision does not currently allow its
games to be streamed via any cloud gaming service. 83

4.34 The key benefit to gamers of cloud gaming is the ability to stream games
without the need to purchase console (or other) hardware, which results in a
cost saving. Cloud gaming also allows a gamer to start gameplay without
waiting for a game to download locally to a device.

Cloud gaming monetisation models

4.35 Cloud gaming service providers are testing a range of different business
models to monetise their services. They all currently offer a subscription-
based model for access to their servers (and some offer a free tier with
advertising for this purpose). Different services monetise their gaming content
in different ways, including (i) the traditional B2P model, whereby users pay a
one-time fee to purchase a game and can only play it on that platform, (ii) the
BYOG model, whereby users pay a regular subscription fee for access to
cloud gaming servers and can play games bought in third-party storefronts,

80 Parties, FMN.
81 Further evidence on Microsoft and third-party submissions on the future of cloud gaming is set out in more
detail in Chapter 8.
82 Parties response to the CMA’s RFI.
83 Parties response to the CMA’s RFI.

45
 such as Steam and Epic Games Store (eg, NVIDIA GeForceNow), (iii) free-to-
play offerings monetised through advertising revenue and in-game purchases,
and (iv) multi-game subscription services, whereby users pay a subscription
fee for access to gaming servers and a catalogue of games (eg, Amazon
Luna and Xbox Game Pass Ultimate). These business models are not fixed,
and cloud gaming services can, and often do, explore different ways of
monetising their services. For example, Amazon Luna offers both an MGS
service and a BYOG feature for Ubisoft games. 84

Cloud gaming infrastructure and ecosystem

4.36 The ‘cloud’ refers to a network of servers which are accessed over the
internet. Cloud gaming utilises cloud servers, as noted above, to allow gamers
to play games without the need for gaming hardware (eg an Xbox or
PlayStation).

4.37 In addition to the supporting infrastructure required for game development

more broadly (as noted above), the cloud-gaming ecosystem is reliant on:

(a) Data centres: data centres are physical buildings located around the world
where servers are placed and connected to the internet.

(b) Servers: servers combine computing components to produce a machine

with significant processing power which run games in a central location,
which gamers connect to via the internet. Server components include:

(i) Central Processing Units (CPUs): The CPU is a processing device

which carries out instructions, allowing the computer to perform its
tasks. 85

(ii) GPUs: as set out above, the GPU allows for the processing of
graphics to improve the visual gaming experience.

(iii) Storage drives: The storage drive of the server is the core data
storage device, holding the OS, applications, data files etc. This can
be stored on solid-state drives (SSDs) or hard disk drives (HDDs). A
cloud server can connect to multiple drives hard drives as needed to

84 Amazon Luna – Cloud Gaming Service, accessed by the CMA on 23 December 2022.
85 The two main manufacturers of CPUs are Intel and AMD. In December 2020, Bloomberg reported that
Microsoft is in the process of designing its own chips and other articles have suggested that Microsoft may be
hiring staff to do so.

46
 allow you access to more games than you could store on a personal
device. 86

(iv) OSs: The OS running on the data server requires compatibility with
the games being played by end-users. As noted above, the majority
of PC games are developed using Microsoft’s Windows OS, while
console games are developed for proprietary console OS.

4.38 Accessibility to each of these core components is essential for the provision of
cloud gaming. The reliance of cloud gaming providers on third parties in order
to access each of these core components varies. For example, across its
broader offering Netflix uses Amazon’s AWS servers to host its content on the
cloud. Conversely, Microsoft hosts its Xbox Cloud Gaming internally, on
dedicated console hardware residing in Microsoft’s own data centres.

Some of the key manufacturers of hard drives include Hewlett-Packard (HP), Kingston Technology, Seagate,
86

Western-Digital and SanDisk.

47
5. Introduction to competitive assessment
5.1 This chapter sets out some aspects of our treatment of the evidence
considered in our assessment of the potential competitive effects of the
Merger, including a summary of the evidence base itself. It then sets out our
findings with respect to the market definitions relevant for the competitive
assessment.

Treatment of evidence

5.2 In the following paragraphs, we briefly set out some of the sources of
evidence we have gathered during our investigation. We then set out some
aspects of our approach to the assessment of certain types of evidence – in
particular, internal documents, views submitted by market participants and by
the public, and economic modelling. We discuss the remedy consideration
process separately in Chapter 11.

Evidence gathered in this investigation

5.3 In assessing this Merger, we have looked at a wide range of evidence that we
considered in the round to reach our decision. The evidence we have
gathered has been tested rigorously, and we considered the context in which
the evidence was produced when deciding how much weight to give it.

5.4 We received a significant volume of evidence from the Parties. In response to

targeted information requests, we received over three million internal business
documents from Microsoft and Activision, including key strategy documents
and email communications among senior staff. These documents which, for
the most part, were created in the ordinary course of business, set out the
Parties’ views of the console and cloud gaming markets, as well as their
future commercial strategy.

5.5 The Parties also had several opportunities to make submissions and
comment on our emerging thinking throughout the investigation. In October
2021, the Parties submitted a response to our phase 1 decision. They
subsequently submitted a response to our Issues Statement, where we set
out the theories of harm on which we planned to focus our phase 2
investigation. We held a site visit with each of the Parties, where their senior
business staff gave us several presentations on the nature of their
businesses, the rationale for the Merger, and answered our questions relating
to our investigation. We subsequently produced working papers and an
annotated Issues Statement with our emerging thinking, and the Parties
submitted their views, in writing, on those materials. We held formal hearings

48
 with each of the Parties, in which we spoke to the Parties’ senior management
about topics that we were exploring in our investigation. The Parties made
submissions in response to our Provisional Findings, and Microsoft also
submitted a response to the Addendum to our Provisional Findings. In
accordance with our standard practice, we continued to collect and analyse
evidence relevant to the investigation of the Merger after publication of our
Provisional Findings and the Addendum to the Provisional Findings. 87 In the
interests of transparency and to ensure that we had the Parties’ submissions
on relevant additional evidence gathered and analysed in the period following
the Provisional Findings, we sent a short supplementary paper summarising
this additional evidence to the Parties. The Parties submitted a response to
this paper, which we have taken into account. In addition, the Parties made a
number of other submissions setting out their views on our theories of harm
and evidence base at different points in our investigation.

5.6 We gathered evidence from other gaming console providers, game

publishers, and cloud gaming service providers. We sent out around 90
requests for information, held a number of calls and meetings, and gathered
hundreds of internal documents from these third parties. In our calls, we
spoke to senior staff and business experts across the industry to gain a better
understanding of the competitive landscape and likely future developments in
these markets.

5.7 We engaged an independent market research company to conduct an online

survey. The survey polled a random sample of PlayStation CoD gamers—
defined for the purposes of the survey as those who played at least 10 hours
or spent at least $100 on the game between July 2021 and June 2022—to
better understand how important this game franchise is to them, and what
they might do if it became partially or totally exclusive to Xbox after the
Merger.

5.8 While there are no pre-defined measures for assessing whether a merger
may be expected to result in an SLC, market shares are a commonly used
measure in merger control cases. There is a high degree of product
differentiation in some of the markets in which Microsoft and Activision
operate, which means that market shares may not be the best indicator, in
this case, of how closely businesses compete with each other. As such, when
assessing the impact of the Merger on competition, we have considered the
evidence on market shares alongside other evidence on how closely the
Parties compete with rivals (either currently or in the future). As well as the
Parties’ market shares, our assessment has taken account of the type of

87 See Mergers: Guidance on the CMA’s jurisdiction and procedure (CMA2), December 2020, paragraph 10.9.

49
 games that Activision offers, of the technical specifications of different
consoles (and the types of games that users play on them), and of Microsoft’s
potential strengths in cloud gaming arising from its broader multi-product
ecosystem. We have also taken account of the strength of competitive
constraints on the Parties, and the extent of past entry and exit from the
relevant markets.

Internal documents

5.9 Internal documents can be a useful source of information in merger

investigations. Documents produced in the ordinary course of business
provide evidence on the perspectives of market participants beyond their
direct submissions to the CMA, often from before the merger under
investigation was in contemplation. In some cases, they speak directly to
questions we seek to answer in our investigations, including for example
questions on what businesses can do, what businesses may find it in their
interest to do, sources of competition that businesses monitor or react to, or
industry trends that provide context for our analysis.

5.10 Among the large number of documents that we obtained from the Parties and
third parties, we identified a range of documents containing information and
discussions relating to a range of themes that are of importance to our
investigation. By way of summary, the evidence base we have drawn from
internal documents includes evidence on:

(a) commercial discussions relating to consumer behaviour and market

mechanics, including drivers of console choice, the presence and extent
of network effects, multi-homing, and other aspects of how competition
works in the gaming industry;

(b) expectations in relation to industry trends (such as subscriptions and

cloud gaming), including quantitative forecasts of revenue growth,
reflections of senior staff on the importance of those trends in the wider
industry, and details of firms’ strategic responses to those trends;

(c) consideration of whether content should be included on subscription

and/or cloud services, including senior staff views on the merits of doing
so, discussions between publishers and platforms relating to supplying
such content, and actual decisions on whether content should be listed on
platforms (both in general and specifically in relation to supplying content
to platforms that are rivals);

(d) the importance of large game franchises (including but not limited to CoD
and WoW), including in relation to console, PC and cloud gaming

50
 services, and the possible ramifications of content exclusivity for
commercial success;

(e) the role of operating systems in cloud gaming, including the cost to serve
when using Windows or Linux, and the impact of choice of operating
system on access to content;

(f) the role of cloud infrastructure in cloud gaming, strategies and plans for
such infrastructure, and forward-looking assessments of the expected
evolution of firms’ cloud infrastructure costs over time; and

(g) internal discussions of the relative strengths of rivals and potential

entrants in cloud gaming services, including in relation to advantages from
existing scale, access to content, operational capabilities and cloud
infrastructure.

5.11 In our review of these internal documents, we took care to interpret them in
their context. In deciding what weight to attach to them we considered
information such as the identity and role of the staff that prepared, sent or
received them. In line with our guidance, where internal documents support
claims being made by merger firms or third parties, we considered whether
those documents were generated prior to the period in which the Parties were
contemplating the Merger, and the period in which third parties were aware of
the Merger. We also attached weight to whether documentary evidence was
consistent with other evidence.

5.12 Where we received submissions suggesting that our interpretation of

documents was incorrect, we assessed those claims. We applied our
judgement to the credibility of those claims, accounting for the nature of the
statements made in documents, the interests of the relevant party in
suggesting an alternative interpretation, and the consistency of any alternative
interpretations with other documents and evidence.

Views of market participants

5.13 During our phase 2 investigation, we engaged with a range of market

participants, including through written questionnaires, data requests,
meetings, as well as having received and considered submissions from some
third-party market participants.

5.14 Views of market participants are an important source of evidence in CMA

merger investigations. That said, the outcomes of a merger investigation can
have a direct financial or strategic impact on market participants, and we
therefore—as with internal documents—considered the interests and
incentives of the Parties, market participants that oppose the Merger and

51
 other third parties, as well as the extent to which such claims are consistent
with other evidence, when assessing what weight to attach to those views.

5.15 In addition to views provided by firms, we also heard from the public directly,
both through a consumer survey, and from the large volume of emails
received from members of the public during the period for response to the
CMA’s Issues Statement, as well as further emails received during the period
for response to the CMA’s Provisional Findings and the Addendum to the
Provisional Findings. 88

5.16 Details of our customer survey are provided in Appendix D. As set out in that
Appendix, our customer survey (the CMA Survey) was commissioned by an
independent specialist agency under controlled conditions, with questions
specifically tailored to bring out objective responses, and accounting for
comments provided by the Parties and by SIE. We are of the view that the
CMA Survey has been designed, conducted and analysed robustly in
accordance with survey good practice.

5.17 In relation to views received by email from individual members of the public, it
was necessary for us to apply caution in interpreting these. There is no
practical way for us to verify that views provided through an open channel are
submitted by genuine consumers, or that those submitting views are
representative of all relevant consumers (and in particular UK consumers). In
relation to representativeness in particular, open calls for comment may
attract greater attention from particular cohorts of the market than others,
either through random chance, or as a result of varying degrees of awareness
across different cohorts (which can be affected by press coverage, internet
commentary and social media coverage—including coverage by interested
parties). As a result, we were unable to place any particular weight on these
email submissions, or to use them to make inferences about ‘average views’
of consumers. However, we note that many respondents raised issues and
concerns that were aligned with those under consideration in the inquiry (both
in terms of the impact of the Merger on competition and any potential benefits
for consumers arising from the Merger). These issues are considered further
in Chapters 7 and 8 below.

88The summary of Issues Statement responses received from members of the public is available on the case
page: Issues statement: Summary of responses. See Appendix H for a summary of responses received from
members of the public to the CMA’s Provisional Findings and the Addendum to the Provisional Findings. The
summary also includes responses received from members of the public to the CMA’s Remedies Notice.

52
 Role of quantitative evidence

5.18 Economic modelling and quantitative evidence—including shares of supply,

financial modelling, and analysis of business data—can be helpful in
assessing a range of questions relevant to merger investigations, which is
how they have been used in this case.

5.19 Quantitative analysis has strengths but can also have limitations. On one
hand, attempting to quantify certain variables can help by giving a sense of
degree when an aspect of the investigation is not clear cut—for example: the
credibility of a substitute, the strength of a firm, the importance of an input, or
the profitability of a strategy. On the other hand, quantitative evidence can
vary depending on the assumptions applied in the methodology.
Quantifications do not capture all relevant considerations. Shares of supply
may not capture the extent to which firms may be closer (or less close)
alternatives to each other because of product differentiation. Financial data
may fail to capture economic gains that are not typically captured in
accounting terms but are nevertheless implicitly or explicitly accorded weight
by senior management, such as increased future opportunities, greater
strategic flexibility, long-run payoffs in terms of scale or network effects, etc.

5.20 In our assessment, we have considered a range of both quantitative and

qualitative evidence, and also considered carefully the relative weight to
attach to each. Where both types of evidence have been available, we have
sought to interpret quantitative evidence alongside qualitative evidence.

Market definition

Framework

5.21 Where the CMA makes an SLC finding, this must be ‘within any market or
markets in the United Kingdom for goods or services’. 89 An SLC can affect the
whole or part of a market or markets.

5.22 Market definition provides a framework for assessing the competitive effects
of a merger. The assessment of the relevant market is an analytical tool that
forms part of the analysis of the competitive effects of the merger and should
not be viewed as a separate exercise. 90 The boundaries of the market do not
determine the outcome of the analysis of the competitive effects of the
merger, as it is recognised that there can be constraints on merging parties
from outside the relevant market, segmentation within the relevant market, or

89 Section 35(1)(b) of the Act.

90 Merger Assessment Guidelines (CMA129), March 2021, paragraph 9.1.

53
 other ways in which some constraints are more important than others. We
take these factors into account in the competitive assessment.

5.23 In this case, as in other digital markets, the relevant products are complex and
have been—and will continue to be—subject to change and development over
time. Demand for gaming products can vary considerably between gamers,
and there are complexities in how customers make decisions. The choices
available to customers depend on whether they already own a gaming device
or if they are planning on buying one. In addition, different customer
characteristics mean that some customers may consider a broader range of
choices than others. For example, some gamers may prefer complex games
that require considerable time and skill (and have historically been played on
consoles or high-specification gaming PCs). Others may prefer simpler games
that can be played casually for short periods of time on a range of devices
including mobiles, whilst another group may prefer a mix of both. This choice
also extends to the supply side, where some devices support games that
require high-powered graphical processing units (eg consoles), while others
do not. Where possible, we have accounted for these factors in the discussion
of market definition which follows. However, we note that a single market
definition may not always capture the true competitive interactions between
different providers and, where this is the case, these are discussed in the
competitive assessment. 91

5.24 The potential issues under analysis in this case relate in various ways to how
competition between the Parties and their rivals will dynamically evolve over
time, in particular in relation to MGS and cloud gaming services, and also in
relation to the next generation of consoles. This chapter does not attempt to
predict the direction in which these markets, and competition in these
markets, will evolve in the future (our views on these issues are set out in the
chapters that follow). In these circumstances, we will place more emphasis on
the competitive assessment than on static market definition. In the
assessment of the impact of the Merger on competition, we will consider
evidence on concentration measures alongside evidence of closeness of
competition. This involves assessing the strength of the current and likely
future constraints between the products of the Parties and their rivals.
Evidence on concentration and on closeness of competition can be
interpreted and taken into account without the need for a precise definition of
the relevant markets. 92

91 CMA 129, paragraph 9.4.

92 CMA 129, paragraph 9.3.

54
5.25 Accordingly, and in particular in relation to those areas involving evolving
conditions of competition (including the emergence of MGS and cloud gaming
services), our analysis does not seek to conclude on a bright-line definition of
the relevant markets, but instead describes the competitive framework within
which the Parties and their rivals operate. 93

5.26 In this case, we consider that gaming platforms are two-sided, with users on
one side and content providers on the other. In defining the market, we have
assessed each side of the market separately, focusing primarily on the user
side of the market, where the potential competitive concerns in this Merger
arise. We have considered both sides of the market in the competitive
assessment, including the impact of direct and indirect network effects. 94

5.27 The starting point for our assessment is the relevant services provided by the
Parties in:

(a) the supply of gaming hardware, and the associated game distribution
(where Microsoft is currently active);

(b) the supply of cloud gaming services (where Microsoft is currently active);
and

(c) the supply of game publishing services (where both Microsoft and
Activision are currently active).

5.28 We consider each of these in turn below.

The supply of gaming hardware (and associated gaming distribution)

5.29 In this section, we discuss the product and geographic market definition
relating to services involved in the supply of gaming hardware. As explained
in Chapter 4, the hardware that gamers typically use to consume gaming
content includes PCs (including desktops, laptops and gaming-specific PCs),
dedicated gaming consoles and mobile devices. As such, our analysis
focuses on the supply of PCs, consoles and mobile devices.

5.30 Microsoft manufactures and sells dedicated gaming consoles under its Xbox
brand. Microsoft also sells PCs (such as its Surface series of computers) and
mobile devices (such the Microsoft Surface Duo), which can also be used to
play video games.

93 CMA 129, paragraphs 9.4-9.5.

94 CMA 129, paragraph 4.22.

55
5.31 Activision does not manufacture or sell any gaming hardware including PCs,
consoles, or mobile devices. 95

Product market definition: supply of gaming hardware (and associated gaming

distribution)

5.32 In this section we consider the product market definition relating to the supply
of gaming hardware, ie PCs, consoles and mobile devices. We also consider
the extent to which game distribution is integrated with this hardware (in
particular, for gaming consoles), and therefore how this also fits with the
product market definition.

Substitutability of mobile devices with consoles and PCs

5.33 First, we consider whether mobile devices should be included in the same
product market as PCs and consoles.

• Parties’ views

5.34 The Parties submitted that there are separate markets for the manufacture
and supply of (i) PCs and consoles and (ii) mobile devices. 96 In particular, the
Parties explained that mobile devices lack the technical capabilities to play
most PC and console games, and only a small proportion of PC and console
games are also available as native mobile apps. 97

• Our assessment

5.35 From a demand side substitutability perspective, the available evidence

indicates that there is limited substitution of a mobile with consoles/PC for
gaming. Third party reports indicate that gamers use mobile devices in a
complementary manner to PC or console; for example, with one third-party
report indicating that a majority of gamers ([]) play games on their mobiles
and at least one other platform. 98

5.36 There are also differences in the price and user interface of mobiles as
compared to PC and consoles. Mobiles tend to be less expensive than
consoles or high-end PCs (though we recognise that mobile devices are
available at a wide range of price points). Mobiles may be purchased for a
number of different uses as opposed to consoles or gaming PCs. Additionally,

95 Parties FMN.
96 Parties FMN.
97 Microsoft estimates that only approximately []% of consoles games and []% of PC games were available

on mobile devices in 2020. See Parties FMN.

98 Microsoft Internal Document.

56
 users tend to interact with mobiles via a touch screen rather than a controller,
mouse or keyboard.

5.37 From a supply side substitutability perspective, the evidence indicates that
there are significant differences in the technical capabilities of a mobile versus
PC/console.

(a) Third party views highlight differences in technical capabilities between

mobile devices and other types of hardware. For example, one publisher
[] stated that some of its games are available across all platforms
except mobile due to the level of porting and optimisation required. 99
Another publisher [] stated that even if the technology to play console
titles on mobile does become available, the mechanics required to play a
game on console may not translate well to a mobile screen. 100 A
competitor [] noted that most third parties publish their content on both
PCs and console. Mobile games are more removed from PC and console
games. 101

(b) Microsoft’s internal documents also discuss the technical differences

between mobiles and consoles. One transcript of a call states that []. 102
Another document summarises internal research into mobile gamers. The
document states that []. 103

5.38 More generally, we note that mobiles are considered separately to PCs and
consoles in Microsoft’s internal documents ([]). 104 The CMA has also seen
third-party reports that consider mobile gaming separately to PC and console
gaming. 105

5.39 Testimony of Microsoft’s executives in US court proceedings further

substantiates that Xbox does not view mobile devices as their competitors.
Speaking in relation to the competitors of Xbox for hardware sales, Microsoft’s
vice-president of business development for gaming, media, and entertainment
in 2021 stated that Microsoft ‘certainly don’t view iPhone as a competing
device’. 106

5.40 Based on the above evidence, we do not include mobile devices in the same
product market as consoles.

99 [] call note.

100 [] call note.
101 [] call note.
102 Microsoft Internal Document.
103 Microsoft Internal Document.
104 Microsoft Internal Document; Microsoft Internal Document; and Microsoft Internal Document.
105 Microsoft Internal Document; and Microsoft Internal Document.
106 Testimony of [], ‘Epic Games, Inc., v. Apple, Inc.’, 5 May 2021, [].

57
 Substitutability of PCs with consoles

5.41 Next, we consider whether PCs and consoles should be included in the same
product market.

• Parties’ views

5.42 In relation to the substitutability between PCs and consoles, the Parties
submitted that PCs and consoles should be considered part of the same
market because (i) PC graphics technology has caught up; (ii) the gap in
prices has narrowed substantially; (iii) performance and characteristics of
consoles and PCs have converged. 107

• Our assessment

5.43 From a demand side substitutability perspective, third party evidence

suggests a slightly higher degree of multi-homing of gamers between
consoles and PC than between different consoles, indicating that PCs are
considered less substitutable with a console by gamers. For example,
external research from Game Track in Q4 2021 submitted by a third party []
showed that a greater percentage of users of one console played on a PC
[] as opposed to playing on another console ([] and [] on each other
console). 108 However, we note that many games are available on both PCs
and consoles, and gamers can play the same game across these devices.

5.44 In terms of price differences, one third party submitted that gaming PCs are
more expensive than PlayStation/Xbox, typically costing $800-1,200+ as
opposed to $400-500 for consoles at first release. 109 This means that in
response to a general increase in the price of consoles, a substantial price
difference would persist between gaming PCs and consoles (given they
currently cost at least twice as much). We consider this likely to limit the
extent of substitution between them.

5.45 From a supply-side substitutability perspective, third party evidence indicates

that PCs are still considerably different to consoles in terms of their technical
specifications. One third party console supplier [] submitted that PCs are
general purpose computing devices that are not specifically dedicated to
video gaming. The third party stated that while consoles are designed to be
used out of the box and require minimal technical knowledge, many gaming

107 Parties FMN.

108 [] response to the CMA’s RFI.
109 [] response to the CMA RFI.

58
 PC users build or customise their device with components from a variety of
manufacturers.

5.46 Evidence from Microsoft’s internal documents also indicates that there are
some technical challenges to porting a PC game to console. For example,
[]. The document goes on to explain that []. 110 Another draft internal
document []. 111

5.47 Finally, evidence indicates that the competitive landscape is different on PCs
and consoles, though this may evolve in the future.

(a) Third parties noted the differences in the gaming ecosystems and
demand of PCs and consoles. One third party publisher [] told the CMA
that certain device categories (ie PC, console, and mobile) are better
suited to certain types of games. 112 Another third party [] considered
that the PC gaming ecosystem has always been open and generally not
as commonly driven by exclusive titles as consoles; which suggests the
competitive landscape in game publishing is different for consoles and
PC. 113 Additionally, another third party [] stated that, unlike consoles,
which are updated uniformly and generationally, users of gaming PCs
generally determine for themselves when and how upgrades are
required. 114

(b) As in the case of mobiles above, internal documents and third-party

reports submitted by the Parties also consider PC and console gaming
separately in their discussion of competitive landscapes. 115 However,
some third-party reports discuss the increasing similarity between PCs
and the latest generation of consoles; for example, one report noted that
console manufacturers are moving away from the more static 8-10 year
console cycle, and instead adopting behaviour more akin to that of PC,
where owners upgrade to a new machine every 3-4 years. 116 On the other
hand, another third party report notes differences, arguing that []. 117

5.48 Given the supply-side and demand-side evidence, together with the mixed
evidence on the evolving competitive landscape between consoles and PCs,
for the purposes of the competitive assessment, we do not include PCs in the

110 Microsoft Internal Document.

111 Microsoft Internal Document.
112 [] call note.
113 [] call note.
114 [] response to the CMA RFI.
115 Microsoft Internal Document; Microsoft Internal Document; Microsoft Internal Document; Microsoft Internal

Document; Microsoft Internal Document.

116 Microsoft Internal Document.
117 Microsoft Internal Document.

59
 product market with consoles; nevertheless, we consider any constraint from
PC in the competitive assessment.

Substitutability of different console types

5.49 Finally, we have considered whether the product market should be further
segmented by console type. The Parties submitted that static consoles and
handheld consoles should be considered part of the same market because
there is no longer a marked distinction between these different types of
consoles; and handheld consoles also support technologically advanced
games (eg Nintendo Switch, Valve’s Steam Deck). 118

5.50 We consider evidence on the similarity between the Xbox and PlayStation
(static consoles) and Nintendo Switch (which is both a static and handheld
console) in the following chapters. The evidence broadly shows that Xbox,
PlayStation and Nintendo all compete in the console market—from a demand-
side perspective, they are all used almost exclusively for gaming and enable
publishers to offer a variety of games on their hardware. From a supply side,
they offer consoles with broadly similar technical specifications and prices that
are comparable (particularly Xbox and PlayStation, and to a lesser extent,
Nintendo). 119 As such, we do not consider it appropriate to further segment by
console type.

5.51 However, whilst we recognise that all gaming consoles exert some
competitive constraint on each other, we also consider any differences
between consoles that may make them less substitutable with each other.
This is particularly true for Nintendo consoles—the latest Switch consoles are
handheld with lower technical specifications (eg in terms of RAM, GPU,
memory) and offer content that is more family/child friendly. This is discussed
further in the competitive assessment.

Conclusion on product market definition for gaming hardware

5.52 Based on this evidence, we consider that there are significant differences
between PCs, consoles, and mobile devices as gaming hardware, and that it
is appropriate to distinguish between each of them for our analysis of the
Merger. We do not consider it appropriate to further segment the console
market by console type, but we assess the different closeness of competition

118Parties FMN.
119In relation to the recently released Steam Deck – a portable PC - we note that it allows gamers to play PC
games natively (or via a cloud platform, including console games) on a handheld device. While we note that the
Steam Deck is more similar to the Nintendo Switch than static-consoles such as Xbox or PlayStation in its
technical specifications, we do not assess this further given it is a portable PC and has a relatively insignificant
share of the gaming hardware market.

60
 between rivals within the competitive assessment. We cannot rule out that
over the longer-term PCs and consoles may become increasingly
substitutable, and the distinctions between different hardware types may
become less relevant in a world where cloud gaming has become more
important. We consider any dynamic elements of competition as they are
relevant in our competitive assessment.

Other considerations in the supply of gaming consoles

5.53 In addition to our assessment above that consoles, PC and mobile devices
are in separate product markets, we set out below two additional
considerations regarding the gaming console market relevant to our
competitive assessment.

Console hardware and distribution

5.54 Digital storefronts are integrated with the console hardware and are the
primary means by which gamers purchase B2P games and subscription
services for that specific console. It is not possible to buy or access console
games in digital form other than from the digital storefront associated with the
specific console. For example, for Xbox, digital games are purchased via the
Xbox Store; Xbox’s MGS, Game Pass, is also available through the Xbox
Store (and likewise for PlayStation and the PlayStation Store, and Nintendo
and the Nintendo Store). While some games (albeit a small and declining
proportion) are also still purchased in physical form, which can be via an
intermediary as well as direct from the console manufacturer, physical
console games are also tied to a particular console. 120 As such, we consider
that competition happens at a platform level from both a supply and demand-
side (ie the console and associated console game distribution combined). We
therefore consider it appropriate to treat console hardware and associated
game distribution as part of the same product market, and consider these
collectively in the assessment. This includes MGS services as well as B2P
console games for the reasons discussed further below.

5.55 For PCs, on the other hand, we note that the storefronts are distinct to
hardware, with different firms and competitive dynamics and with no
tied/bundled relationship between the hardware and storefronts. For example,
there are various storefronts available for purchasing PC games which are not

120For example, you would need to purchase an Xbox version of a game to play that game on an Xbox console.
Further, many purchases of physical games are made at the time as the console and are sold as part of a
bundle.

61
 inherently tied to the particular manufacturer’s hardware and offer games for
multiple operating systems.

5.56 For the reasons set out above, we do not consider game distribution for either
mobile or PC in our competitive assessment and do not assess these further.

Segmentation between MGS and B2P services

5.57 In this section, we discuss evidence related to the product market for MGS
services, and whether we should assess it separately to other B2P console
games.

• Parties’ views

5.58 Microsoft submitted that MGS services were a means of payment – not a
market. It further submitted that the gaming content that is available on MGS
services was also available to purchase on a standalone B2P basis and there
is no additional or new content offered via MGS. Microsoft submitted that this
evidence was consistent with gamers perceiving subscription and B2P purely
as alternative payment methods to access the same content. 121

5.59 Microsoft also submitted that Xbox telemetry data shows that gamers
frequently switch between the two payment options. In particular, Microsoft
submitted that based on this data:

(a) Between [] of Xbox gamers continue playing games purchased on a

B2P basis in the 12 months after unsubscribing from Game Pass.
Between [] of gamers purchase a game on a B2P basis within a year
after unsubscribing from Game Pass. 122

(b) Gamers that subscribe to Game Pass “mix and match” across games
irrespective of how they paid for them. Microsoft submitted figures on the
gamers who spent at least []% of their total game-time on Halo Infinite
in the month following release. According to Microsoft, these figures show
that while these gamers’ total game-time increases as a result of their
engagement with the new Halo release, both B2P game-time on other
games and Game Pass game-time fall. Microsoft submitted similar figures
for CoD: Modern Warfare. According to Microsoft, these figures show that
gamers who play that game increase their overall game-time, however
their Game Pass and B2P game-time on other games declines. 123

121 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 1.8 (a).
122 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.18.
123 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.19.

62
 (c) []% of Game Pass customers who had been subscribed to Game Pass
for six months or more had bought games on a B2P basis during that
period. 124

5.60 Microsoft also submitted that its internal analysis shows a []% decline in
base game sales twelve months following their addition on Game Pass. 125

• Our assessment

5.61 We consider that Microsoft’s telemetry data and internal analysis presented
above have some limitations. The time period over which Microsoft appears to
infer that a B2P purchase constitutes substitution is quite long (up to one
year). Moreover, the observation that Game Pass customers continue to
purchase B2P games during the period is more consistent with
complementarity than substitutability. Given B2P purchases continue even
during the subscription period, it is unclear from Microsoft’s submission
whether and to what extent the proportion of consumers that make B2P
purchases within a year of closing their subscription is different from the
proportion of consumers making B2P purchases that never had a subscription
in the first place.

5.62 However, notwithstanding these limitations, the telemetry data suggests that,
overall and at least at a point in time while MGS is expanding, there is some
substitution from B2P to MGS services. It also suggests that there is possibly
some diversion from MGS to B2P services, but the data is more unclear in
that respect as the magnitude of any substitution effect is lower and more
difficult to detect by just looking at the data.

5.63 Microsoft internal documents recognise that adding titles to Game Pass would
lead to cannibalisation of B2P sales, []:

(a) One internal email exchange stated that []. 126

(b) One internal document stated that []. 127

(c) One internal document states that []. 128

5.64 We asked third parties (competitors and publishers) to provide their views on
the level of substitutability between B2P and MGS services within gaming.
Some responses were fairly generic (eg explaining that games are available

124 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.20.
125 Microsoft response to the phase 1 Issues Letter.
126 Microsoft Internal Document.
127 Microsoft Internal Document.
128 Microsoft Internal Document.

63
 on both MGS and B2P), and were not informative on the relevant question.
However, the other responses in the round suggest that MGS services place
some constraint on B2P services and vice-versa. 129

• Conclusion on segmentation by MGS and B2P

5.65 In the long-term, as the console gaming market matures, we consider that
diversion may be stronger between competing MGS services than between
MGS and B2P services.

5.66 This might be, for instance, because MGS and B2P have different
characteristics. Under an MGS service, gamers can access a bundle of
games during a period of time for a fixed fee. The drivers of demand in that
context are the price of the bundle, the range of content, and the quality of
individual titles. By contrast, when accessing games through digital
storefronts, gamers typically pay a higher up-front price for a single game.
Although stores can monetise games using a range of methods (eg in-app
purchases), gamers typically have unlimited access to the game after the
purchase. The drivers of demand in that context are the price and quality of
the individual game.

5.67 However, we have seen no evidence that consumers choose between

different MGS in isolation or the rate of switching between different MGS.
Rather, we see that Xbox and SIE market their MGS as part of their overall
console package, and these are therefore competitive parameters of their
offering. 130

5.68 Therefore, we conclude that console B2P and MGS are within the same
product market, but recognise there is some differentiation between the two
(which we discuss in more detail in the competitive assessment, as relevant).

5.69 We refer to this product market as ‘console gaming services’ which, for the
reasons explained, incorporates the console hardware and integrated digital
storefront, including for both B2P games and MGS.

129 [] response to CMA RFI; [] response to CMA RFI; [] response to CMA RFI; [] response to CMA RFI;
[] response to the CMA’s section 109 notice (s109 notice).
130 See for instance Microsoft’s webpage advertising/explaining its consoles where it describes the speed and

performance of the Xbox Series X/S, the expanding game library, Xbox Game Pass, and Smart delivery. See
https://www.xbox.com/en-US/consoles.

64
Geographic market definition: supply of console gaming services

Parties’ views

5.70 The Parties submitted that their products and services are available on a
global basis, including in the UK. In relation to PCs and consoles, the Parties
submitted that the geographic market is at least wider than the UK (including
the EEA) if not worldwide, as consoles supplied across the UK/EEA are
technically equivalent and there are no impediments to cross-border sales of
PCs/consoles within the UK and EEA. 131

Our assessment

5.71 We note that Microsoft and its console gaming service rivals (ie SIE and
Nintendo) are active across a broader geographic region than the UK and are
often active globally.

5.72 However, various third-party internal documents discuss the competitive

landscape in consoles at a regional/national level. For example, a third party’s
[] internal document considers pricing, the available game library,
awareness and interest of gamers and supply of consoles by country, for this
third party’s main sales countries. 132 The document also analyses and
indicates differences in the breakdown of new console adoption rates by
different customer segments at a national level. 133 Another document of the
same third party [] analyses platform preference and brand momentum at a
country level. 134 Various other third party documents also survey users by
region and discuss findings and competition by country. 135

5.73 We found additional evidence suggesting that console gaming services may
have a national dimension:

(a) The market shares provided by the Parties show material differences
between the UK and global shares, suggesting differences in the
competitive landscape by geography. In particular, the Merged Entity
generally has higher shares of supply in the UK as compared to
worldwide. For example, in 2021, the Parties estimate Microsoft’s share of

131
Parties FMN.
132 [] Internal Document.
133 [] Internal Document.
134 [] Internal Document.
135 See for example, [] Internal Document; and [] Internal Document.

65
 supply of console hardware by sales volume to be [10-20]% globally but
[20-30]% in the UK. 136

(b) In addition, the CMA has seen some evidence of differential availability of
services in local areas, including of Microsoft’s Game Pass offering. 137

(c) Further, in relation to Xbox Live Gold, for example, Microsoft’s website
explains that ‘[t]o ensure that pricing for the Xbox Live Gold subscription
service reflects the local market economies, Xbox Live Gold subscription
cards are only redeemable in the country purchased and cannot be
redeemed in any other country’. 138

5.74 As such, while recognising that there are multi-national aspects to competition
in this market, there is also evidence of regional and national variations in
supply and demand. We therefore consider it is appropriate to assess the
impact of the Merger in console gaming services in the UK, but taking into
account the multi-national context and evidence as relevant and appropriate
in the competitive assessment. We consider much of the evidence available
which is not limited to the UK is highly relevant to our assessment, and we
also do not consider our competitive assessment or conclusions would
materially change if we were to consider a wider geographic frame of
reference (ie regional or global). For example, foreclosure strategies may be
implemented on a multi-national or global level, and we take this into
consideration where appropriate.

The supply of cloud gaming services

Product market definition: supply of cloud gaming services

Parties’ views

5.75 Microsoft submitted that cloud gaming is a means of accessing games, not a
separate market. It described cloud gaming as a feature that provides an
alternative way for gamers to access content that is not tied to a specific
device and should not be considered separately from other ways of accessing
and playing games, at least on console and PC. 139

136 Parties FMN.

137 For example, see ‘Xbox Supported Countries/Regions’, accessed by the CMA on 30 August 2022.
138 ‘Making sure your Xbox digital subscription is valid for your country | Xbox Support’, accessed by the CMA on

4 August 2022.
139 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.7. See also Microsoft

response to the Provisional Findings, 2 March 2023, paragraph 3.26.

66
5.76 Microsoft stated that cloud gaming was []. In support of this it stated that the
[]% of xCloud use in 2022 was on console, and that telemetry data
indicates that []% of XPGU usage (in terms of active users) was on xCloud
in September 2022. 140

5.77 Microsoft further submitted that this was particularly clear on console, where
cloud gaming could not be divorced from the console itself, and is no more
than another parameter on which consoles compete against each other. It
submitted in this context that cloud was not even a standalone ‘product', let
alone a separate market. 141 It also submitted that on Xbox Cloud Gaming,
around []% of the usage on console is simply to try a game before
downloading it, []. 142 Microsoft also submitted that feedback which Microsoft
receives from Xbox Cloud Gaming users shows that []% of users have used
cloud gaming to try a game before installing it on a console or PC. 143

5.78 Microsoft submitted that gamers who use cloud gaming features on their
console are not the intended customers for cloud gaming service providers
(and [] of Xbox Cloud Gaming customers access the service on console)
given that cloud gaming is primarily aimed at enabling consumers to play high
performance games on a range of lower-powered devices, and are as such
not relevant to the CMA’s theory of harm. 144

5.79 Microsoft further stated that gamers choose a gaming experience based on
whether it provides enjoyment at an attractive price point and are not
motivated by the location of the content or the means of delivery. Cloud
gaming services therefore need to compete effectively with downloadable
game options if they are to grow. 145

5.80 Microsoft also stated that, whilst it agrees that in the long-term cloud gaming
may make hardware distinctions less important, it remains unproven as a
customer proposition and the available evidence did not indicate in any
manner that this is likely to change anytime soon. 146 We discuss the future of
cloud gaming separately in Chapter 8.

5.81 The Parties however acknowledged that cloud gaming services could be
attractive to a different pool of customers that does not have access to the

140 Microsoft response to working papers.

141 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.27.
142 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.27.
143 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.27.
144 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.28, 3.29.
145 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.8.
146 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.9.

67
 current hardware required for playing more complicated games and will lower
the barriers for certain gamers to access titles. 147

Our assessment

5.82 On the demand side, cloud gaming allows consumers to play high
performance games that were previously only available on consoles or high-
end PCs, on a range of lower-powered devices including mobiles, smart TVs,
and low-end PCs. 148 This means consumers do not need to pay the
substantial upfront cost associated with consoles and high-end PCs, therefore
making cloud gaming attractive to a new pool of consumers. An internal
document from one competitor [] describes how cloud gaming will grow the
total addressable number of gamers playing console-quality games. 149
Another competitor [] similarly described how cloud gaming could make
interactive 3D gaming more accessible to billions of gamers. 150

5.83 An internal document from Microsoft []. 151 The responses show that of
consumers who were aware cloud gaming was available on XGPU, 152 most
were [] 153 []. 154 In our view this indicates that consumers consider
different factors important for cloud gaming as compared to consoles or PCs.
This [] was at the end of 2020. Over time we may expect behaviour to
change as the installed base of devices age and gamers have an alternative
to buying a new machine when the time comes.

5.84 The evidence submitted by Microsoft in paragraph 5.77 suggests that

currently trying games before downloading them is [] with xCloud users.
However the [] noted in the previous paragraph also showed that only []
of consumers who were aware cloud gaming was available on XGPU []. 155
We assume respondents mentioning [] would include those who value
being able to try games before downloading it. 156 We therefore consider this
evidence suggests it is unlikely that trying games before downloading is the
primary reason people are interested in cloud gaming, despite it being [].
Nor do we think that this would be the case as cloud gaming develops. We

147 Submission made during the phase 1 Issues Meeting.

148 Cloud gaming can also be used for lower performance games, but as described below, this is not the focus of
our competitive assessment.
149 [] Internal Document.
150 [] response to the CMA’s RFI.
151 Microsoft Internal Document.
152 In general, we consider the views of those Game Pass customers who were aware cloud gaming was

available to be more informative in terms of why they are interested in cloud gaming.
153 [].
154 [].
155 This figure [], among customers who were not aware that cloud gaming was available on XGPU [].

Microsoft Internal Document.

156 Indeed, the relevant Microsoft Internal Document [].

68
 further note that a memo from September 2022 describing priorities for Fiscal
Year 2023 noted in FY22 [], 157 suggesting xCloud [], and is not merely a
console feature.

5.85 Furthermore, in contrast to Microsoft’s submission that cloud is just an

additional feature, a note from one of its senior employees ([]) in 2020 (with
comments from []) discusses their view of cloud gaming and how streaming
is not a feature, but rather something that fundamentally changes the way and
type of games played. 158

5.86 In any case, even if some customers continue to use cloud gaming to try a
game before downloading it, this would suggest that it is complementary to
offline console play, and not a substitute.

5.87 Whilst Microsoft submitted that [] xCloud usage is currently on console, and
a [] of customers try a game before downloading it, in our view this may be
influenced by Microsoft tying its own cloud gaming offering to Game Pass
Ultimate and using its existing consumer base to grow its cloud gaming
service, and therefore not representative of the cloud gaming market more
generally.

5.88 We also expect this to change in the future. First, we have received no
evidence from Microsoft that it thinks the majority of future cloud gaming will
be on console. In contrast, a number of documents discussing Microsoft’s
expectations for cloud gaming in the future suggest it will become increasingly
important relative to playing natively on console. One document from 2019 in
particular discusses strategy and sequencing of expanding cloud streaming
noting that Microsoft first plans to ‘[]’. 159 In an email exchange with [],
[]. 160 Ultimately a view that xCloud usage will remain primarily attached to
the latest Microsoft console is inconsistent with evidence on Microsoft’s (and
competitors’) view that an important aspect of cloud gaming is being device
agnostic and bringing new customers into gaming.

5.89 Second, evidence from Microsoft’s internal documents also suggest it is still
serious in pursuing a [], as discussed further in Chapter 8 below.

157 Microsoft Internal Document.

158 []. Microsoft Internal Document.
159 Microsoft Internal Document. In addition this document states that ‘cloud streaming’ represents an []’. We

note that this document [].

160 Microsoft Internal Document. This is discussed in more detail in Chapter 8.

69
5.90 Third, we consider evidence that Microsoft [] 161 is inconsistent with the idea
that streaming is merely a way to try a game before you download it to
console given, [].

5.91 Fourth, we also think a future strategy of cloud being primarily a feature of
consoles is inconsistent with Microsoft’s [], as discussed further in Chapter
8 below.

5.92 Finally, the pattern of use across devices is already different for other
competitors. Evidence from one competitor [] shows that around []. 162
Furthermore, this competitor stated that the vast majority of its users were not
using high-end PCs and were new to gaming. 163

5.93 In relation to Microsoft’s submission that gamers who use cloud gaming
features on their console are not the intended customers for cloud gaming
service providers, we do not think there are different markets (or market
segments) based on the end-device:

(a) First, given that cloud gaming is device agnostic, we do not think that a
service developed for those using a console as an end-device is
fundamentally different from a service developed for users of other end-
devices, just as its development is not determined by the end-device.

(b) Second, based on the evidence discussed in the above paragraphs,

Microsoft’s strategy, investment and the growth Microsoft has made in
cloud gaming, we do not think it likely Microsoft has invested in cloud
gaming, and will continue to do so in future [], only so that customers
can try games before they download them (ie we do not think it likely that
is the sole reason). We further note in this context that Microsoft
described the acquisition as supporting the opportunity to ‘pivot away from
the device centric strategy of the past’ and ‘focus instead on a consumer-
centric, cross-platform approach that will allow gamers to play wherever
and on whatever device they wish.’ 164

(c) Third, many customers already use xCloud off-console, and [] as
described above.

(d) Fourth, cloud gaming offers various benefits that are not limited to access
by a particular device. For instance, being able to try a game before
downloading it, or indeed, avoiding the need to download at all, apply to

161 See further Chapter 8 below.

162 [] Internal Document.
163 [] call note.
164 See paragraph 2.27.

70
 other end devices, not just consoles. The evidence above does not
support a customer segmentation on the basis of any particular use case.

(e) Finally, as consoles age, customers accessing cloud gaming via a

console may derive greater benefit from the technological advantages of
cloud gaming, which could include the option of cloud as a way to avoid a
console upgrade. In particular those customers might well decide not to
spend a significant sum of money on a new console and so only consider
cloud options at that point.

5.94 On the supply side, cloud gaming services are very different from console
gaming. To offer cloud gaming services, a provider needs access to cloud
infrastructure that is capable of running high performance games and offering
a low-latency gaming experience. Console gaming, by contrast, requires the
manufacture, distribution, and ongoing support of physical devices.

5.95 Certain internal documents from Microsoft show that []. 165 These
documents are discussed in more detail in Chapter 8. Microsoft gaming’s
CEO publicly stated in 2020 that it sees Amazon and Google as its main
competitors in gaming going forward, rather than the traditional gaming
companies SIE and Nintendo which are not well positioned for cloud
gaming. 166 In our view, this suggests that Microsoft considers cloud gaming
as distinct to console gaming (and that it expects cloud gaming to be more
important, which is covered in more detail in the ‘future development of cloud
gaming’ section).

5.96 Similarly, internal documents from another competitor [], discuss cloud
gaming separately from console. 167

5.97 As the market is still developing, and our assessment is forward looking, we
do not have real world evidence of consumer diversion in response to a
change in price or quality. However, in our view, the above evidence suggests
customers are likely to have specific reasons for choosing to cloud game
(including not being tied to a particular piece of hardware), Microsoft and other
cloud gaming service providers monitor cloud gaming competitors separately
from console competitors, and on the supply side the required inputs to
compete are very different. Therefore, while we expect at least some

165 For example, see Microsoft Internal Document; Microsoft Internal Document; Microsoft Internal Document;
Microsoft Internal Document; and Microsoft Internal Document.
166 ‘Games Industry.biz - "We see Amazon and Google as the main competitors"’, accessed by the CMA on 30

January 2023.
167 [] Internal Document; and [] Internal Document.

71
 constraint from console and PC, we find that cloud gaming services should be
considered as a distinct market to console and PC.

Other considerations in the supply of cloud gaming services

5.98 As described above, we have found that, when it comes to gaming hardware,
mobile devices are in a different market to consoles, partly because of
technical differences and the comparative limitations of mobile devices.
Similarly, we do not think that firms that are solely focused on bringing mobile-
quality games to cloud are a meaningful constraint on cloud gaming services
that provide console-quality and/or gaming PC-quality games (high-
performance games). Throughout this Final Report, where we refer to high-
performance games, we mean games that would not typically run natively on
mobile. From the evidence we have reviewed in Chapter 8, we consider that
these games are the focus of competition within cloud gaming services, and
are therefore the focus of our competitive assessment.

5.99 As the market is still developing, cloud gaming service providers are testing a
range of different business models to monetise their service. Different
services monetise their gaming content in different ways, including (i) the
traditional B2P model, whereby users must purchase a game through the
cloud provider’s storefront in order to play it (eg, Google Stadia), 168 (ii) BYOG
model, whereby users pay a regular subscription fee for access to cloud
gaming servers and can play games bought in third-party storefronts (eg,
NVIDIA GeForce Now (NVIDIA GFN)), (iii) free-to-play offerings paid through
advertising revenue and in-game purchases, and (iv) MGS services, whereby
users pay a subscription fee for access to gaming servers and a catalogue of
games through the cloud provider’s storefront (eg, Amazon Luna and Xbox
Game Pass Ultimate). As with consoles, we consider it appropriate to treat
cloud gaming services and their associated game distribution as part of the
same product market, and consider these collectively in the competitive
assessment.

5.100 The evidence we have reviewed shows that these business models are not
fixed, and that cloud gaming service providers are open to exploring different
ways of monetising their services, and a single participant may have more
than one way of monetising content. For example, Amazon has implemented
a B2P and BYOG option in Luna for Ubisoft games, []. 169 An internal
document from Microsoft []. 170

168 This service was shut down in January 2023.

169 ‘Amazon Luna Ubisoft Store', accessed by the CMA 17 January 2022; and [] call note.
170 Microsoft Internal Document.

72
5.101 As noted above, given the market is still developing, we also do not have real
world evidence of consumer diversion between different options. However, we
would expect the evidence on diversion and segmentation between B2P and
MGS on console presented above to apply to cloud gaming too, given the
point about cannibalising sales holds for cloud gaming too, with the additional
option of BYOG—which in practice is similar to B2P for the customer,
although with the flexibility to use the game across compatible platforms–not
changing that assessment.

5.102 We therefore consider it appropriate to assess the impact of the Merger within
a product market for cloud gaming services that includes all business models
(including B2P, BYOG and MGS). We have however taken the differences
between these business models into account in our competitive assessment,
including our assessment of the relevant counterfactual, which is discussed
further in Chapter 8.

Geographic market definition: supply of cloud gaming services

Parties’ views

5.103 The Parties submitted the market for cloud gaming services is worldwide. 171

Our assessment

5.104 Customers can only use cloud gaming services that are available in the
country in which they are located, as providers can typically restrict usage
based on IP address. 172 There are currently substantial differences in the
availability of services across countries, for example:

(a) Amazon Luna is currently only available in the US, Canada, Germany and
the UK, having launched in the UK very recently. 173

(b) NVIDIA GFN is available only in selected countries, excluding for example
China and India. 174175

171 Parties FMN.

172 For instance, Amazon’s terms of service for Luna note that ‘Amazon will use technologies to verify your
geographic location. You may not use any technology or technique to obscure or disguise your location.’ Amazon
Luna Terms of Use, accessed by the CMA on 3 April 2023. In addition, GFN uses ‘Alliance Partners’ in some
countries, where customers sign up directly with the Alliance Partner in the given country.
173 Amazon Luna accessed by the CMA on 3 April 2023.
174 ‘NVIDIA GeForce Now supported locations’, accessed by the CMA on 26 January 2023.
175 For instance, NVIDIA GFN notes ‘Not all membership plans or promotions may be available in your location or

language’ GeForce Now Membership Terms, accessed by the CMA on 10 April 2023.

73
 (c) xCloud is only available in selected countries, excluding for example
China, India, and Indonesia. 176

5.105 Whilst we might expect these differences in availability to reduce over time as
services are expanded to more countries, providers may still price differentiate
based on customer location, since customers are unable to easily switch to
cloud gaming services operating in other countries if the price of the UK
service increases.

5.106 In addition, on the supply side, to provide a low latency gaming experience a
cloud gaming service provider must have servers located in a data centre
close to the consumer. 177 This creates a barrier to expanding to new
geographies as a provider must first invest in or gain access to national or
regional data centres.

5.107 Providers, however, are multi-national. The content available on different

services is generally the same across countries, and the cloud infrastructure
technology used is also generally the same, subject to being rolled out to
regional data centres.

5.108 As such, while recognising that there are multi-national features to this
market, there is also evidence of regional and national variations in supply
and demand. We therefore consider it is appropriate to assess the impact of
the Merger in the market for cloud gaming services in the UK, but taking
account of the multi-national aspects of competition where relevant. We
consider much of the evidence available which is not limited to the UK is
highly relevant to our assessment, in part because (i) some business
decisions are taken at a global level and (ii) understanding global trends and
developments may help to inform how we expect the market to develop in the
UK. We also do not consider our competitive assessment or conclusions
would materially change if we were to consider a wider geographic frame of
reference (ie regional or global). For example, foreclosure strategies may be
implemented on a multi-national or global level, and we take this into
consideration where appropriate.

The supply of game publishing services

5.109 Game development refers to the creation of a game (including the design, art,
programming and testing), using software development tools (including
software development kits (SDKs) and ‘game engines’). Game publishing

‘Game Pass supported countries/regions’, accessed by the CMA on 26 January 2023.

176

For example, providers are likely to require servers in a UK or Western European data centre to provide cloud
177

gaming in the UK.

74
 refers to subsequently making available a game to the public, for sale or free
of charge. For the purposes of this Final Report, and in line with the Parties’
submissions, we have considered development and publishing activities
jointly (under the single term, publishing). 178

5.110 The Parties are both active in the publishing of games for PCs, consoles, and
mobile devices. 179

Product market definition: supply of game publishing services

5.111 In this section, we consider the product market for game publishing services.
There is a wide variety in the types of games that are published, with
significant differences between them, including based on the device on which
they are published and their content/target audiences. In this section, we
therefore assess whether game publishing should be segmented into markets
by device types (ie PC, consoles and mobile) and/or by game genre.

Product market definition by device type

5.112 First, we consider whether games published on mobiles, PCs and consoles
should be included in the same product market.

• Parties’ views

5.113 The Parties submitted that the CMA should consider an overall market for
game publishing.

5.114 The Parties submitted that there are differences between PC/console games
and native mobile games. 180 They explained that technical differences
between mobile and PCs/consoles lead to the development of different
gaming content in each. The Parties explained that PC/console games tend to
be more complex and narrative-driven than native mobile games and,
therefore, cater to different audiences. 181

5.115 The Parties also submitted that the distinction between PC and console
games has blurred as game publishers are increasingly launching games for
both PCs and consoles. The Parties explained that this trend has been
reinforced by the launch of next generation consoles that are becoming
similar to PCs, and by the emergence of cross-play and cross-progression

178 Parties FMN.

179 Parties FMN.
180 Parties FMN.
181 Parties FMN.

75
 (ie the ability to save game progress on one device and access it on another
device) which enables gamers to play across devices. 182

• Our assessment

5.116 We have assessed whether game publishing should be segmented by the

type of device on which the game is published.

5.117 Mobile vs PC/consoles: Third party evidence supports the view that there
are material differences between publishing games on mobile devices relative
to PCs/consoles, due to their technical differentiation, differentiated
monetisation models and type of games:

(a) One third party [] stated that mobile devices have physical limitations in
terms of power, battery life, screen size, etc. As such, mobile games are
generally lower quality, simple in design and concept, and frequently
played to kill time, eg while commuting. The third party explained that
mobile games tend to be free-to-play, with revenues driven almost
exclusively by in-app purchases. 183 The competitor [] noted that most
third parties put their content on PCs and consoles, but that mobile
content is further removed from PC and console content. 184

(b) A publisher [] stated that some of its games are available across all
platforms except mobile due to the level of porting and optimisation
required. 185

(c) Another publisher [] stated that even if the technology to play console
titles on mobile does become available, the mechanics required to play a
game on console may not translate well to a mobile screen. 186

5.118 The Parties’ internal documents also support the view that there is relatively
low substitutability from the publishers’ side between mobile and
PCs/consoles. For example, one Microsoft document notes that []. 187 An
Activision strategy document noted that publishing companies are []
between PC/console and mobile. 188 Another Activision document noted that

182 Parties FMN.

183 [] response to the CMA’s RFI.
184 [] call note.
185 [] call note.
186 [] call note.
187 Microsoft Internal Document.
188 Activision Internal Document.

76
 games such as [] and [] either chose not to pursue a cross-platform
game that include mobiles due to [] and for []; or chose to []. 189

5.119 From a demand side perspective, evidence from the Parties’ internal
documents indicated that the target audiences for mobile are different to
consoles and PC games. One Activision internal document highlighted that by
[], Activision has been able []. 190 Another Activision internal document
notes that only approximately []% of [], and points to the differences in
[] as an area for investigation. 191 One Microsoft internal document looks at
[]. 192 An Activision document looks at [], finding that they differ in areas
such as the length of playtime and preferred type of gameplay. 193

5.120 PCs vs consoles: Third party views similarly also indicated that there are
hardware and software differences between PCs and consoles which limits
their ability to publish games on both PC and consoles without additional cost
and effort:

(a) A number of third parties told the CMA that they require SDKs from
console original equipment manufacturers to publish games on console,
which is not required for publishing PC games. 194

(b) One third party explained that one console has hardware limitations that
other consoles and PCs do not, which can present difficulties in publishing
PC games on that console. 195

5.121 We note that many games, including the Parties’, are often published on only
one form of hardware (for example, Activision’s WoW games are only
playable online on PCs, and have not been released on the latest generation
consoles). 196

5.122 As discussed above, internal documents and third-party reports submitted by

the Parties also consider mobile, PC and console gaming separately in their
discussion of competitive landscapes in publishing. 197 However, some third-
party reports discuss the increasing similarity between PCs and the latest
generation of consoles; for example, one report noted that trends in cross-
play and cross progression will eventually lead to consolidation as consumers

189 Activision Internal Document.

190 Activision Internal Document.
191 Activision Internal Document.
192 Microsoft Internal Document.
193 Activision Internal Document.
194 [] call note; and [] response to the CMA questionnaire.
195 [] call note.
196 Parties FMN.
197 Microsoft Internal Document; Microsoft Internal Document; Microsoft Internal Document; Microsoft Internal

Document; and Microsoft Internal Document.

77
 []. 198 We note that this overlaps with the developments in cloud, ie over
time gaming could become more device agnostic (at least for devices capable
of running high-performance games), such that over the longer term we can
expect these high-performance games for PC and console to increasingly
overlap.

5.123 Finally, we note that (i) Microsoft’s stated rationale for the Merger is in part to
improve its presence and expertise in mobile gaming; and (ii) the Parties’
shares of supply vary between publishing games for each of PCs, consoles,
and mobile devices, with the competitor sets varying between category of
gaming hardware too. 199 We consider this further supports a conclusion of
different markets by device type.

• Conclusion on segmentation by device type

5.124 Based on the above, we consider that there are differences between games
published on PC, console and mobile, and that it is appropriate to treat them
as separate game publishing markets for our analysis of the Merger. We also
note that over the longer term there is evidence that PC and console games
may become increasingly substitutable, especially with the increased
relevance of cloud gaming that may impact preferred game development
formats in future. We will consider any constraint from PC games or console
games where relevant in the competitive assessment, and consider the
expansion and entry of games more generally in Chapter 9 on countervailing
factors.

Product market definition by genre

5.125 In this section, we assess whether it is appropriate to consider separate

markets by genre within game publishing.

• Parties’ views

5.126 In relation to segmenting the market by game genre, the Parties submitted
that, whilst games are sometimes categorised by reference to their genre (eg
role-playing games, shooters, puzzle games, etc), there is no need to further
segment on that basis as many games combine aspects of multiple genres,
and gamers generally just play whatever game they like best, irrespective of
what genre that game may be. 200

198 Microsoft Internal Document.

199 Parties FMN.
200 Parties FMN.

78
5.127 Microsoft further submitted that Xbox telemetry data shows that gamers play
multiple genres: []% of gamers play three or more genres and []% of
gametime on Xbox was accounted for by gamers playing at least three
different genres. 201 Microsoft also submitted that evidence from actual gamer
behaviour on Xbox also shows that gamers switch between genres in
response to new releases. 202

• Our assessment

5.128 Games are generally classified, discussed and listed for sale in digital
storefronts by reference to their ‘genre’, which can include action, adventure,
role-playing games, shooter, sport, strategy, etc. Some of these genres can
be further segmented into sub-genres—for example, shooter games can be
further classified into first-person shooters, third-person shooters, etc. We
also note that some games can belong to multiple genres.

5.129 As context, we note that two of Activision’s main games on console, CoD and
Overwatch, belong to the shooter genre. The shooter genre constitutes a
significant share of the publishing market, particularly on consoles. Evidence
from the Parties shows that shooters account for [] of publishing revenues
across all platforms. 203 Evidence submitted by third parties also indicates that
shooter is an important game category on consoles. For example, data from
[] indicates that shooters constituted []% of PlayStation’s revenue and
40% of PlayStation’s playtime globally. 204 [] also submitted that in 2021,
‘first person shooter’ games represented[]% of gameplay hours amongst
PlayStation’s top 100 titles. 205

5.130 The Parties’ internal documents indicate that the Parties categorise,
distinguish, and compare games from different publishers according to genre.
For example, a third-party report submitted by Microsoft analyses trends and
market shares in the game publishing market by genres, within different
geographies. 206 The Parties also advertise games by genre in their digital
storefronts. 207

5.131 Some third parties explained that certain game genres (such as shooter
games) compete more closely with each other than with games from different

201 Microsoft response to working papers.

202 Microsoft response to working papers.
203 Parties Internal Document.
204 [] Internal Document; and [] Internal Document.
205 [], submission to the CMA.
206 Microsoft Internal Document.
207 See ‘Xbox Games Catalog’, accessed by the CMA on 24 August 2022; and ‘Battle.net’, accessed by the CMA

on 10 January 2022.

79
 genres. 208 Specifically in relation to the shooter genre and CoD, []
publishers that responded with CoD’s closest competitors listed other shooter
games such as EA’s Battlefield. 209 Publishers also listed: Halo, Overwatch,
Apex Legends, Rainbow 6 Siege, Far Cry, Destiny, PlayerUnknown’s
Battlegrounds (PUBG) and Hell Let Loose.

5.132 In relation to the shooter genre, various Activision internal documents list and
track [] as competitive constraints to CoD (and Overwatch), including
comparing []. 210 The main competitors tracked include [].

(a) In particular, multiple documents track [] as a close competitor to

CoD. 211 For example, one Activision email chain refers to a competitor
analysis produced by Activision's EMEA insights team: it compares the
performance of [] to the one of 'Call of Duty Vanguard' in the first week
after its launch since CoD is a 'competitor’. 212

(b) Various Activision documents also track [] closely as a competitor to

CoD, 213 including comparing the effect on engagement on CoD during a
[]. 214

(c) Other documents also analyse the impact of [] on CoD. 215

5.133 In addition to the Activision documents listed above, one internal Microsoft
presentation analyses the impact of the introduction of [] in the market on
gameplay time and spending on []; and notes that []. 216

5.134 On the other hand, we received evidence suggesting that the importance of
genre as a means of segmenting games has decreased in recent years. One
third party publisher [] stated that popular games have adopted elements of
multiple genres, and games often compete across genres for the attention of
gamers. 217 Other third party publishers [] stated that factors beyond genre
may lead to titles competing closely, such as the quality of the franchise,

208 For example, see [], submission to the CMA; and [] call note.
209 [] response to the CMA’s RFI; and [] response to the CMA’s RFI.
210 See for example, Activision Internal Document.
211 See for example, Activision Internal Document; Activision Internal Document.
212 Activision Internal Document.
213 Activision Internal Document; Activision Internal Document; and Activision Internal Document.
214 Activision Internal Document.
215 Activision Internal Documents; Activision Internal Document; Activision Internal Document; and Activision

Internal Document.
216 Microsoft Internal Document. [].
217 [] call note.

80
 release date or quality of the gaming franchise (leading to its overall
popularity). 218

5.135 Third parties’ internal documents suggest that CoD’s closest competitors are
titles of other large franchises across genres, along with shooters. For
example, one third party console gaming service competitor’s [] internal
document from February 2022 compared CoD’s total gameplay hours per title
on the competitor’s two most recent console generations with other titles like
Fortnite, FIFA, Apex Legends and GTA V (out of which Fortnite and Apex
Legends are shooters). 219

5.136 Activision’s internal documents also indicate that it monitors titles that release
in the same [], rather than by genre. One Activision internal document
analyses specific []. 220 Another Activision internal document analyses CoD
Vanguard’s competitors in its launch window. The document states that []
was the top competitor in the launch window, with [] the top competitors in
the shooter genre []. The document later states that outside of [] are CoD
Vanguard’s top competitors for overall share of gameplay time. 221

• Conclusion on segmentation by genre

5.137 Based on the above evidence, we consider that games compete most
strongly with other games within the same genre; however, they also compete
with games outside the genre (particularly if they are large multi-player
franchises with gameplay elements and social aspects similar to other large
multi-player franchises in other genres).

5.138 In addition, we note that defining genres, and whether or not a game belongs
to a particular genre, can be difficult and somewhat arbitrary, and also subject
to change over time (eg games can belong to multiple genres, and games are
sometimes classified under different genres by different publishers/platforms).

5.139 We therefore consider it is appropriate to assess the impact of the Merger in

the overall game publishing market, rather than defining separate markets by
genre. However, given the strong evidence above on the extent to which there
is similarity within genres, and differentiation across them, we consider any
relevant differences regarding competitive constraints within genres in our
competitive assessment. In particular, we consider the particular competitive
conditions within the shooter segment in our competitive assessment.

218 [] call note; and [] call note.

219 [] Internal Document.
220 Activision Internal Document.
221 Activision Internal Document.

81
Geographic market definition: supply of game publishing services

Parties’ views

5.140 The Parties submitted that they consider the relevant geographic market for
game publishing to be worldwide. They stated that this is supported by the
fact that many publishers typically produce one version of a video game for
distribution worldwide (subject to possible localizations), without significant
price differences. 222

Our assessment

5.141 The demand for game publishing services for both PC and console is both
from the distributors of the game (direct) and the end-gamers (indirect, for
example because game publishers, particularly on console, have a certain
degree of control in suggesting the retail price).

5.142 From a demand-side perspective of the distributor, we note that contracts are
generally negotiated and agreed upon globally.

5.143 From a demand-side perspective of the consumer, we note that there do not
seem to be significant differences in price across jurisdictions (though there
are some variations based on exchange rates, VAT and other taxes
applicable different regions). 223

5.144 However, evidence also points to some regional variations across games,
especially on PC:

(a) Market shares in publishing vary significantly by geography, indicating

that there are likely regional variations in consumer demand. Some
games (such as League of Legends, Genshin Impact, and Crossfire) are
significantly more popular in Asian countries than in the UK. For example,
one Microsoft internal document contains the top genres for PC gaming
by hours played split by region []. 224 [] 225, [].

(b) Games are not always released at the same time across all countries (or
at all). There are also differential requirements in some countries in

222 Parties FMN.

223 See for example, Steam All Region Price Checker (steamregionalprices.com), accessed by the CMA on 3
February 2023.
224 Microsoft Internal Document.
225 We note that the document indicates that Chinese data may be less reliable due to the high level of regional

localization of titles, however we believe that it is still broadly representative of preferences.

82
 relation to age restrictions, depiction of violence, etc which require games
to be tailored accordingly.

(c) Matching of gamers in multiplayer games places a degree of weight on

players’ relative locations when forming a lobby, because geographical
proximity both to the hosting server and to other players is an important
determinant of the latency experienced by gamers when playing a
match. 226

5.145 As such, our view is that it is appropriate to assess the effects of the Merger
on the supply of console game publishing and PC game publishing services
globally, while taking account of any regional or UK-specific differences where
relevant in the competitive assessment. In any event, we consider that our
competitive assessment and its conclusions would not materially change if we
were to consider a narrower geographic frame of reference (eg UK or EEA
countries).

226 Activision response to the CMA’s RFI.

83
6. Counterfactual

Framework for assessment of the counterfactual

6.1 The counterfactual is an analytical tool used to help answer the question of
whether a merger gives rise to an SLC. 227 It does this by providing the basis
for a comparison of the competitive situation with the merger against the
competitive situation absent the merger. 228

6.2 The counterfactual is not, however, intended to be a detailed description of

those conditions of competition that would have prevailed absent the
merger. 229 The CMA’s assessment of those conditions is better considered in
the competitive assessment. 230 The CMA also seeks to avoid predicting the
precise details or circumstances that would have arisen absent the merger. 231

6.3 At phase 2, the CMA will select the most likely conditions of competition as its
counterfactual against which to assess the merger. 232 For anticipated
mergers, the counterfactual may consist of the prevailing conditions of
competition, or conditions of competition that involve stronger or weaker
competition between the merger firms than under the prevailing conditions of
competition. 233

6.4 In its assessment of the counterfactual, the CMA may need to consider
multiple possible scenarios, before identifying the relevant counterfactual. 234
As part of this assessment, the CMA will take into account whether any of the
possible scenarios make a significant difference to the conditions of
competition, 235 and if they do, the CMA will ultimately select the most likely
conditions of competition absent the merger as the relevant counterfactual. 236
Counterfactual assessments will often focus on significant changes affecting
competition between merger firms, such as entry into new markets in
competition with each other, significant expansion by the merger firms in
markets where they are both present, or, exit by one of the merger firms. 237

6.5 We may examine several possible scenarios to determine the appropriate

counterfactual, one of which may be the continuation of the prevailing

227 CMA129, paragraph 3.1.

228 CMA129, paragraph 3.1.
229 CMA129, paragraph 3.7.
230 CMA129, paragraph 3.7.
231 CMA129, paragraph 3.11.
232 CMA129, paragraph 3.13.
233 CMA129, paragraph 3.2.
234 CMA129, paragraph 3.13.
235 CMA129, paragraph 3.13.
236 CMA129, paragraph 3.13.
237 CMA129, paragraph 3.8.

84
 conditions of competition. An example of a situation where the CMA may
select a counterfactual different from the prevailing conditions of competition
is where the target is likely to exit the market absent the transaction under
review. Another scenario in which the CMA may consider an alternative
counterfactual to the prevailing conditions of competition is where one of the
merging parties would have entered or materially expanded its presence in a
market absent the transaction.

6.6 Further, the time horizon considered by the CMA in its assessment of the
counterfactual will depend on the context and will be consistent with the time
horizon used in the competitive assessment. 238

The Parties’ submissions on the relevant counterfactual

6.7 The Parties submitted that the relevant counterfactual against which to assess
the Merger is the prevailing conditions of competition. They submitted that: (i)
there are no competing bids or parallel transactions; (ii) neither Party can be
considered a failing firm; and (iii) the Merger does not result in the loss of a
potential entrant that would make a material difference to the competitive
assessment. 239

6.8 Microsoft submitted that there is no counterfactual available in which

Activision content (specifically CoD) would be widely available via multiple
subscription services. 240 It told us that Activision currently only makes limited
back-catalogue titles available on subscription, and only on a limited time-
period basis. 241 Microsoft submitted that Activision has never published newer
content on multi-game subscription services and has no intention to do so in
the future, 242 which it submitted is reflected in statements made by Activision’s
senior leadership and in internal documents. 243

6.9 Microsoft also noted Activision’s existing agreements with SIE, []. Microsoft
submitted that [] SIE. 244

6.10 The Parties also submitted that Activision games would not have been
available to cloud gaming services absent the Merger. 245

238 CMA129, paragraph 3.15.

239 Parties FMN.
240 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.39.
241 Microsoft response to the phase 1 decision, 11 October 2022, paragraph 4.12.
242 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.35.
243 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.36-4.37.
244 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.38.
245 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 1.16. See also Activision response

to the Provisional findings, 2 March 2023.

85
Our assessment

6.11 In making our counterfactual assessment, we do not seek to ossify the market
at a particular point in time. 246 Our assessment can reflect that, absent the
merger, a merger firm would have continued making investments in
improvements, innovations or new products. 247

6.12 We agree with the Parties’ view that the prevailing conditions of competition is
the correct framework under which to assess the Merger. The Parties both
operate in dynamic markets which are growing and evolving quickly, and in
which competition revolves around bringing new and innovative products to
market (see Chapter 4, Industry Background). On this basis, we consider the
prevailing conditions of competition in this Merger include changes or
developments in strategy to reflect changing market dynamics, and
innovations, including the continued growth and development of multi-game
subscription and cloud gaming services.

6.13 We are of the view that the assessment of the likelihood of Activision’s
content becoming available on MGS services or cloud gaming services is best
carried out within the competitive assessment. For the reasons set out in our
competitive assessment, we consider that absent the Merger, in the
foreseeable future, Activision ‘day and date’ content would become available
on cloud gaming services, but not on MGS services on gaming consoles, at
least for Activision’s most valuable games. However, we consider that
Activision would likely place increasingly valuable parts of its gaming
catalogue on MGS services as these services continue to grow.

246 CMA129, paragraph 3.3.

247 CMA129, paragraph 3.3.

86
7. Theory of harm 1: Vertical effects in console gaming

Introduction

7.1 This chapter analyses the Merger’s effects on competition in the supply of
console gaming services (ie their hardware, associated digital storefronts and
subscription services) arising from input foreclosure.

7.2 Microsoft currently competes in the supply of console gaming services via its
Xbox, and Activision provides inputs to Microsoft and other console providers
in the form of videogames, including CoD.

7.3 The concern under this input foreclosure theory of harm is that the Merger
may lead to the Merged Entity using CoD and other Activision content to
foreclose rival providers of console gaming services (which we refer to as
‘gaming consoles’ in this Final Report). Specifically, we consider whether
Microsoft could harm its rivals’ competitiveness by ceasing to supply CoD and
other Activision content (total foreclosure), or by worsening the terms or
quality of the content provided to rivals (partial foreclosure).

7.4 This chapter is structured as follows. First, we assess the competitive

landscape in the market for console gaming services. Second, we assess the
possibility of Activision’s content being available on gaming consoles’ MGS
services. Third, we present our framework for the assessment of this theory of
harm. Fourth, we assess the Merged Entity’s ability and incentive to
implement a foreclosure strategy, and the effect that this would have on
competition in the supply of console gaming services.

7.5 For the reasons set out below, our conclusion is that the Merged Entity has
the ability to foreclose downstream console gaming rivals, particularly
PlayStation. However, we found that the Merged Entity would not have the
incentive to do so.

Competitive landscape in console gaming services

Introduction

7.6 Before considering the vertical foreclosure framework and assessment, we

provide an overview of the competitive landscape of the console gaming
services market.

7.7 In this section, we consider:

87
 (a) Features of the market that are relevant to our assessment (network
effects and multi-homing).

(b) Shares of supply in console gaming services, ie looking at gaming

hardware, distribution and subscription services.

(c) The closeness of competition between Xbox, PlayStation and Nintendo’s

consoles.

Features of the console gaming market

Network effects

7.8 Gaming platforms are two-sided, with users on one side and content providers
on the other. Two-sided markets are often characterised by network effects,
where the value of the product for customers on one side of the platform
depends on the volume of users either on the same side (direct network
effects) or on the other side (indirect network effects). 248

7.9 Network effects in gaming can be direct and indirect:

(a) Direct network effects are likely to arise particularly in multiplayer games,
where users like playing the same game with their friends (at the same
time), and where players get ‘matched’ with other players more effectively
as the number of players on the platform increases. This often requires
having the same manufacturer’s consoles (except for games that allow
cross-play across different consoles). As such, an increase in the number
of users that play a game in a particular platform would increase the
attractiveness of that game and platform, thereby drawing additional
users.

(b) Indirect network effects arise because game publishers are more likely to
develop content for a platform with a significant user base and, in turn, a
strong content library attracts more users to the platform.

7.10 In this section, we assess whether evidence supports the existence of such
direct and indirect network effects at game and platform level. This is
important because (i) the existence of direct network effects amplifies the
effect of any foreclosure strategy, with a greater number of users switching
away to continue playing the same game with their friends, and (ii) indirect
network effects incentivise publishers to focus their efforts on making games

248 CMA129, paragraph 4.22.

88
 for platforms with a significant user base (which in turn attracts more
customers to that platform).

Parties’ views

7.11 In relation to direct network effects, Microsoft explained that gamers like to be
on the same platform as their friends to play multiplayer games. 249 It also
submitted that gamers may benefit from a broad gamer base with multiplayer
games on account of better ‘matchmaking’ across players when there is a
larger pool of players available to be matched. Microsoft stated that this
therefore implies that direct network effects operate primarily at the game
level rather than at a platform level. 250

7.12 Microsoft further submitted that direct network effects are too weak for
Microsoft to foreclose rival consoles because (i) gamers make console
decisions based on the utility they derive from the library of games on a
console rather than a single game (and therefore network effects on one
game cannot foreclose rival consoles); (ii) the co-existence of Nintendo,
PlayStation and Xbox over the past 20 years does not support the existence
of strong network effects; and (iii) more multiplayer games can be played
online across devices, due to cross-play functionality. 251 Microsoft submitted
that an increasing number of games have cross-play functionality, which
means that network effects operate at the market level (rather than the
platform level) and publishers have an increased incentive to publish on
multiple consoles. 252 Evidence submitted by Microsoft showed that the
gametime share of Xbox’s top 20 games with cross-play available as an
option was [] in 2021, up from [] in 2020. 253

7.13 In terms of indirect network effects, Microsoft explained that game publishers
are more likely to develop content for a platform with a significant user base
and that, in turn, a strong content library attracts more users to a platform. 254
Microsoft also submitted that evidence from the past 20 years has shown that
indirect network effects in the gaming industry are too weak to foreclose rivals
because (i) SIE, despite having ‘double the user base of Microsoft’, has not
been able to foreclose Microsoft or stop it from attracting content from

249 Microsoft response to the phase 1 Issues Letter.

250 Microsoft response to theory of harm (TOH) 1 working papers.
251 Microsoft response to TOH 1 working papers.
252 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.56(d).
253 Microsoft response to TOH 1 working papers.
254 Microsoft response to the phase 1 Issues Letter.

89
 publishers, and (ii) publishers multi-home across platforms, which is
consistent with weak indirect network effects. 255

7.14 In respect of Microsoft’s argument that the continued co-existence of

Nintendo, PlayStation and Xbox does not support the existence of strong
network effects, we consider that network effects can be strong even in a
scenario where they are not sufficiently strong to have caused a market to ‘tip’
to a single supplier. We acknowledge that cross-play can, in principle,
dampen network effects at the console level, and the extent to which this will
be the case will tend to be reflected in the evidence we consider in our
assessment below. The following sections set out our assessment of the
evidence on the strength of network effects, which includes third parties’
views and internal documents.

Our assessment

• Third parties’ views

7.15 Third parties have noted the existence of direct and indirect network effects in
gaming, with direct network effects being particularly strong for large multi-
player social franchises such as CoD:

(a) A third party gaming console supplier [] submitted to the CMA that
network effects are a prominent feature in console video gaming and are
directly related to the user base and number of content providers. The
supplier stated that this therefore would magnify the impact of any action
that led to a change in sales, as it would have wider effects beyond the
impact on direct sales. It described both direct and indirect network effects
as being prevalent in gaming. 256

(b) A cloud gaming provider [] explained that leading AAA games had
multiplayer functionality and had become social media platforms. As a
result, established AAA games benefited from significant network effects,
which raised barriers to entry for new games and developers. 257 It also
stated that, as gamers play games with their friends, they created
something akin to a social network devoted to specific games, which they
said created large switching costs when transitioning to other games,
even in the same genre. These switching costs could result in franchises

255 Microsoft response to TOH 1 working papers.

256 [] response to the CMA’s RFI.
257 [] response to the CMA questionnaire.

90
 being popular for years, as gamers are so deeply invested in terms of
time, money and friends. 258

(c) The PC storefront Steam announced that it changed its revenue share
structure to reward the positive network effects that developers of large
games create, stating that successful games and their large audiences
had a material impact on those network effects. 259

7.16 We note that some of the evidence above is not directly related to consoles
and comes from rivals in cloud gaming and PC storefronts. However, we
consider that network effects, particularly at the game-level, work similarly
across different gaming platforms.

7.17 We consider other evidence relevant to assessing the strength of direct and
indirect network effects in gaming below, such as (i) the costs of making a
game available on multiple consoles; (ii) the prevalence of cross-play and
cross-progression; and (iii) the views of publishers regarding making games
exclusive to one platform. These factors determine the stickiness of gamers
and/or publishers to consoles, in turn affecting the strength of network effects.

7.18 Third parties noted that the costs of developing a game for different consoles
or porting a game across consoles were not too high. One publisher []
stated that developing games for different consoles requires the use of
different software development kits, but this does not result in a material
increase in development difficulty, time or costs. 260 Another publisher []
stated that there are minor hardware differences between PlayStation and
Xbox, such as GPU and driver speeds, which require work to account for the
hardware differences. However, the publisher also noted that in general, the
game development platforms for PlayStation and Xbox are similar (eg both
use C++ coding language). 261 Another publisher [] noted that tools like the
Unreal Engine empower developers to create games that work on multiple
platforms including PC and consoles. 262

7.19 In relation to the prevalence of cross-play and cross-progression, nearly all

publishers that responded to our phase 2 questionnaire stated that there was

258 [] response to the CMA questionnaire.

259 ‘New Revenue Share Tiers and other updates to the Steam Distribution Agreement (steamcommunity.com)’,
accessed by the CMA on 13 December 2022.
260 [] response to the CMA’s RFI.
261 [] response to the CMA’s s109 notice.
262 [] response to the CMA’s RFI.

91
 an increase in the prevalence of cross-play and cross-progression, particularly
in multi-player games like CoD. 263

7.20 Further, publishers stated that they would in general not be in favour of
making games exclusive to a platform, unless there were strong commercial
or strategic incentives to do so. For example, all of the third party publishers
that we have obtained evidence from noted that making a game exclusive to a
platform prevents publishers from reaching a large part of players, who are on
other formats/hardware, therefore forgoing a part of potential revenues. 264
One publisher [] noted that exclusivity to a console platform favours certain
players and upsets others, which can be risky in terms of community
management and/or public relations given the current installed base for each
console. 265

7.21 However, various publishers also stated that they have made games either
fully or partially exclusive to a certain platform for commercial or strategic
reasons. These included obtaining access to the first-party IP of a particular
platform and/or increasing interest in a particular game from users of a
platform. 266 For example, two publishers [] and [] stated that platform
exclusivity can aid in gaining deeper market penetration on a given
platform. 267

• Evidence from the Parties’ internal documents

7.22 Various internal documents of the Parties note that direct network effects are
prevalent in gaming and are associated with both (i) ecosystems/platforms
that facilitate gaming and (ii) large social multiplayer games like CoD.

7.23 In relation to direct network effects at the platform-level:

(a) One Microsoft internal document from September 2020 looks at platform
network effects on Xbox. A regression analysis using Xbox data shows
[]. 268

(b) Another Microsoft internal document from June 2022 states that []. 269

263 [] response to the CMA’s RFI; [] response to the CMA’s RFI; [] response to the CMA’s RFI; and []
response to the CMA’s RFI.
264 [] responses to the CMA’s RFI.
265 [] response to the CMA’s RFI.
266 [] response to the CMA’s RFI; and [] response to the CMA’s RFI.
267 [] response to the CMA’s RFI; and [] response to the CMA’s RFI.
268 Microsoft Internal Document.
269 Microsoft Internal Document.

92
 (c) A Microsoft internal document from April 2020 containing a review of
Microsoft’s strategy states that []. 270

7.24 In relation to direct network effects at the game-level:

(a) An Activision internal document providing an overview of the game

streaming industry states that multiplayer games have []. The
document explains that [] as the gaming audience widens. 271

(b) An Activision presentation notes that ‘while the concept of enduring, multi-
decade franchises has long been seen across media and entertainment,
gaming franchises enjoy attributes that are unique to the sector’. These
include []. 272 Activision submitted that this document was a deck
prepared by Activision to share with []. Activision also explained that
the purpose of this particular slide was to explain how consumer interest
in video game franchises is []. 273

(c) Another Activision document discussing industry trends notes that ‘top
performing games are []. The biggest franchises [] are massively
accessible, deeply social and engaging (often linked to real-world pop
culture), []’. The document also noted that []. The document further
notes that consumers are increasingly demanding an []. 274

7.25 We note the Parties’ submission that some of the internal documents referred
to above are consistent with network effects occurring at the market/game
(rather than the platform) level and were drafted prior to [] cross-play games
[]. 275 However, we note that (i) more recent documents discuss the
existence of strong network effects at the platform level; and (ii) the
prevalence of cross-play has increased, but not to the extent of eliminating
platform-level network effects.

7.26 Microsoft’s internal documents also discuss the presence of indirect network
effects in gaming, noting the presence of a virtuous cycle where more games
attract more content, which in turn attracts more gamers.

(a) One Microsoft internal document, for example, explains []. 276

270 Microsoft Internal Document.

271 Activision Internal Document.
272 Activision Internal Document.
273 Activision response to TOH 1 Ability working paper.
274 Activision Internal Document.
275 Microsoft, email to the CMA.
276 Microsoft Internal Document.

93
 (b) Another Microsoft internal document explains that Microsoft’s gaming
ecosystem creates []. The document goes on to illustrate a flywheel that
shows [], 277 []. 278

Conclusion on network effects

7.27 The evidence presented above indicates that direct network effects are
prevalent in the gaming industry at both a game-level and platform-level.
There is evidence that network effects are stronger for large multiplayer social
franchises like CoD, with gamers wanting to play with their network of friends.

7.28 Indirect network effects also exist, with publishers wanting to publish their
games on the platform with the most users. However, the evidence shows
that most publishers choose to make their games available on as many
platforms as possible. It indicates the additional cost of developing a game for
both Xbox and PlayStation—as opposed to developing it for a single
console—is relatively low. Developments in technology (eg Unreal Engine)
have reduced these costs further. Therefore, we consider that Xbox and
PlayStation both have a sufficiently large user base to attract good content
from game publishers. Although we cannot rule out that indirect network
effects could magnify the effectiveness of a foreclosure strategy (due to
publishers choosing to not make their games available on the foreclosed
console, especially if any such strategy succeeds in significantly reducing a
console’s user base), we consider that at present indirect network effects do
not represent an obstacle to attracting content for either Xbox or PlayStation.

7.29 We also found that the prevalence of cross-play has increased in recent
years, based on data from the Parties and third-party evidence. In general,
the increased prevalence of cross-play makes it possible to play with gamers
on different consoles, along with increasing the total user base available for
matchmaking (though we note that it is still possible to have differences in
gameplay features and certain content while still enabling cross-play). This, in
turn, reduces the strength of direct network effects at console level (but does
not reduce the strength of direct network effects at game level). We discuss
the impact of network effects and cross-play on incentives to foreclose in our
competitive assessment.

277 [].
278 Microsoft Internal Document [].

94
Multi-homing across consoles

7.30 In this section, we consider evidence on the extent to which users own more
than one type of console (multi-homing). This is important to the assessment
for the following reasons:

7.31 First, the extent of multi-homing across various consoles indicates the degree
to which gamers consider them to be substitutable, ie a higher level of multi-
homing indicates that gamers view the consoles as complements rather than
substitutes.

7.32 Second, to the extent that gamers are already multi-homing across Xbox and
PlayStation, a hypothetical withholding of CoD and other Activision content
from PlayStation is likely to (i) involve lower switching costs for these gamers
and (ii) allow Xbox to recapture a smaller amount of the sales lost on
PlayStation, which we consider in our assessment of incentives.

Parties’ views

7.33 Microsoft submitted that gamers often multi-home across gaming consoles. In
relation to multi-homing across the PlayStation and Xbox consoles (as well as
PC), it submitted that at least [20-30%] of PlayStation 5 users also owned an
Xbox Series X/S and [] also gamed on a PC, based on third party data for
the USA procured by Microsoft from a third-party data provider (NPD).
Microsoft also submitted that these metrics are as of December 2020 and are
likely to have increased materially at the time of its submission (31 October
2022). 279

Our assessment

• Third parties’ views

7.34 In relation to multi-homing across PlayStation and Xbox, third party evidence
indicated that a small proportion of PlayStation gamers also owned an Xbox.

7.35 Evidence [] submitted by a third party [] indicated that []. 280

7.36 In relation to multi-homing across other consoles, third party evidence

indicated that a greater proportion of users owned a Nintendo Switch along

279 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.78.
280 [] Internal Document.

95
 with either Xbox or PlayStation, as compared to multi-homing between Xbox
and PlayStation of the same generation.

7.37 Evidence [] submitted by a third party [] indicated that []. 281

7.38 Survey results submitted by a third party [] in 2021 indicated that a high
proportion of UK users of a Nintendo console ([]%) owned an Xbox or a
PlayStation. 282

7.39 Evidence based on third-party NPD data submitted by the Parties showed that
in Q4 2020, []% of PS5 owners owned an Xbox whereas []% of PS5
users owned a Nintendo Switch console. 283

• Evidence from the CMA survey

7.40 Our survey of a sample of PlayStation CoD gamers in the UK found that
levels of multi-homing of the surveyed PlayStation CoD gamers with either
Xbox or Nintendo were similar. 284 In particular, alongside a PlayStation, 15%
of respondents also owned an Xbox, 18% also owned a Switch, and 13% also
owned both an Xbox and a Switch. This therefore indicates that 28% of
respondents owned both a PlayStation and an Xbox, whereas 31% of
respondents owned a PlayStation and a Nintendo console.

7.41 We note that the above statistics do not separate out by the console
generation and therefore over-estimate the extent of multi-homing within a
generation. The CMA survey found that among those gamers owning both
PlayStation and Xbox, 61% owned consoles of the same generation whilst the
remaining 39% owned consoles only across generations.

7.42 However, most of these respondents use only one device as their main
gaming console despite owning multiple devices. In particular, among those
owning multiple gaming devices, 92% of them use one device almost all of the
time or most of the time.

Conclusion on multi-homing across consoles

7.43 We note that different sources of evidence point to different levels of multi-
homing across consoles. This may be driven by differences in the generation

281 [] Internal Document.

282 [] response to the CMA’s RFI.
283 The Parties’ submission.
284 The CMA survey refers to an online survey of a sample of SIE’s UK customers conducted by DJS Research

Ltd. Our sample frame included primary accounts used to play CoD between July 2021 to June 2022 for 10 hours
or more and/or to have spent at least $100 in the period. Please refer to Appendix D for further detail on the CMA
survey.

96
 of console owned, timing of the survey and the size, geography and other
characteristics of the underlying sample of respondents being considered.
The evidence nevertheless shows that:

(a) less than 28% of PlayStation users already own an Xbox; and

(b) some evidence suggests that a greater fraction of users owns the
Nintendo Switch along with either Xbox or PlayStation, as compared to
multi-ownership between Xbox and PlayStation. 285

Shares of supply in the downstream market

7.44 Shares of supply in the downstream console gaming market can provide an
indication of the relative size of console manufacturers and can be used to
assess the competitive constraints faced by Microsoft in the console gaming
market. They can also be informative in analysing the effect of any foreclosure
strategy on downstream rivals.

7.45 However, while shares of supply are a useful indicator when firms provide
similar products, they are less useful if their products are differentiated. As
noted by the CMA’s Merger Assessment Guidelines (MAGs), ‘the CMA may
calculate concentration measures on multiple different bases, including and
excluding different firms, depending on which firms the CMA wishes to
compare. The CMA may then attach greater weight to concentration
measures that include firms whose products are more substitutable, and less
weight to concentration measures that include firms whose products are less
substitutable.’ 286

Parties’ views

7.46 The Parties submitted that Microsoft’s sales of console hardware have lagged
behind SIE’s and Nintendo’s on a worldwide basis over the last three years: 287

(a) Globally, the Parties submitted that Xbox’s share of console hardware in
2021 was [10-20%] by sales value. 288 They also submitted that the
corresponding shares for PlayStation and Nintendo were [40-50%] and
[30-40%] respectively. 289

285 Please refer to paragraphs above discussing information on multi-homing submitted by Microsoft and [].
286 CMA129, paragraph 9.3.
287 Microsoft, email to the CMA.
288 The corresponding shares by sales volume, ie number of console units sold, were [10-20%] for Xbox, [30-

40%] for PlayStation and [40-50%] for Nintendo.

289 Microsoft, email to the CMA.

97
 (b) In the UK, the Parties submitted that Xbox’s share of console hardware in
2021 was [20-30%] by sales value. The Parties stated that the larger
shares of Nintendo ([20-30%]) and PlayStation ([40-50%]) in the UK are
indicative of their strength in the console segment. 290 In addition, the
Parties submitted that PlayStation’s shares increased significantly relative
to Nintendo between 2020 and 2021 following the release of the new
console generation both globally and in the UK. 291

7.47 Microsoft also submitted that PlayStation was twice the size of Xbox in terms
of the installed console base globally in 2021: 292

(a) Globally, Microsoft submitted that PlayStation had the largest share [40-
50%], followed by Nintendo [30-40%] and Xbox [20-30%]. 293

(b) In the UK, Microsoft submitted that Xbox had a share of [30-40%] by
installed console base, while PlayStation and Nintendo had a share of
[40-50%] and [20-30%] respectively. 294

7.48 Microsoft further submitted that PlayStation had the average MAUs of Xbox
globally. 295

7.49 In console digital game distribution (B2P), 296 the Parties submitted that Xbox
has a UK revenue share of [40-50%] in 2021, whereas PlayStation and
Nintendo have a share of [40-50%] and [5-10%] respectively. At the global
level, the Parties say that Xbox’s share is [20-30%], while PlayStation has a
share of [50-60%] and Nintendo [10-20%]. 297 The Parties noted that these
shares reflect the fact that Nintendo sells more of its games in physical (rather
than digital) form. 298

7.50 In MGS services, the Parties submitted that Xbox had a UK revenue share of
[50-60%] in 2021, whereas PlayStation had [30-40%]. At the global level the
Parties stated that Xbox’s share was [30-40%], while PlayStation had a share
of [30-40%]. 299 These shares included subscription services on all platforms.
The Parties also noted that in console MGS services that Xbox’s share is not
materially different to its overall share in digital distribution, estimating Xbox to
have a [50-60%] share in the UK and a [30-40%] share worldwide in 2020.

290 Microsoft, email to the CMA. The corresponding shares by sales volume, ie number of console units sold,
were [20-30%] for Xbox, [30-40%] for PlayStation and [30-40%] for Nintendo.
291 Parties FMN.
292 Microsoft response to working papers.
293 Microsoft response hearing opening statement.
294 Microsoft response hearing opening statement.
295 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.28.
296 This includes distribution revenue from all content – including in-game and add-ons for free-to-play games.
297 Parties Internal Document.
298 Microsoft, email to the CMA.
299 Parties FMN.

98
 The Parties said that PlayStation had higher shares in console MGS globally
in 2020. 300

Our assessment

7.51 We used data from the Parties and third parties to estimate market shares for
different gaming consoles. We found that, depending on how these shares
are measured (ie, by total gaming console revenues, installed base or yearly
active users), Microsoft is the second-largest gaming console provider in the
UK, and the second or third largest gaming console provider globally. As set
out below, Microsoft’s global share of supply ranges between [20-30%] to [30-
40%].

7.52 We also found that Microsoft’s shares by gaming console revenue and
volume of console units were higher than estimated by the Parties. 301 While
still smaller than SIE, the difference between the shares of Microsoft and SIE,
both in the UK and globally, is significantly smaller than suggested by the
Parties. We consider that the discrepancy arises because the Parties do not
have access to third party data, which shows that []. 302

7.53 The table below summarises the shares of supply we estimated using the
Parties’ and third parties’ data in 2021 (with further detail provided in
Appendix C). 303 We consider all gaming console revenues collectively (ie
hardware, B2P distribution and subscription) based on our assessment of
market definition above.

300 Parties FMN.

301 The Parties’ estimated shares presented above indicate that PlayStation was [] times the size of Microsoft
in 2021 by console hardware revenue and [] times by console units. We found that the PlayStation was []
times larger than Microsoft, even when just considering console hardware revenue (ie. excluding distribution and
subscription revenue).
302 [] response to the CMA’s RFI; [] response to the CMA’s RFI.
303 Microsoft’s shares are not significantly different in 2019 and 2020, but SIE’s share in 2020 is c. []% lower by

console hardware revenues, with Nintendo gaining that lost share.

99
Table 7.1: Shares of supply in 2021
%

Console revenues Installed console base Yearly Active Users

Global
Microsoft [20-30] [20-30] [30-40]
SIE [40-50] [40-50] [40-50]
Nintendo [20-30] [30-40] [10-20]
Total global 100 100 100

UK
Microsoft [30-40] NA NA
SIE [40-50] NA NA
Nintendo [10-20] NA NA
Total UK 100 NA NA

Source: CMA’s calculations.

Note: The installed console base was computed as the cumulative number of console hardware sales over the period 2013-
2021.Console revenues include revenues from hardware, B2P digital distribution and subscription revenues

7.54 In 2021, depending how market shares are calculated:

(a) PlayStation was [] times the size of Microsoft globally and about []
times Microsoft’s size in the UK.

(b) Nintendo was [] times the size of Microsoft globally, but [] times
Microsoft’s size in the UK.

Conclusion on shares on supply in the downstream market

7.55 We therefore consider SIE to be the largest supplier of console gaming

services both globally and in the UK. Microsoft is smaller than SIE, but the
difference is smaller than suggested by the Parties, particularly in the UK.
Nintendo also has a significant share in the console gaming market globally.
However, Nintendo has lower shares in the UK relative to both PlayStation
and Xbox in console gaming revenue.

Competition among gaming consoles

7.56 In this section, we assess the closeness of competition between Nintendo’s

Switch, Microsoft’s Xbox and SIE’s PlayStation.

Parties’ views

7.57 Microsoft submitted that the suggestion that Xbox, PlayStation and Nintendo
are not close competitors is incorrect. 304

7.58 Microsoft submitted that there are many ways that SIE markets the
PlayStation ‘reflecting the fact that there are many dimensions of competition

304 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.46.

100
 in the console market’. Microsoft submitted that these included (i) the
technical capabilities and price of the console itself, (ii) SIE’s own exclusive
content; and (iii) third party content, including Call of Duty and other titles such
as FIFA and Grand Theft Auto. 305

7.59 In relation to Nintendo specifically, Microsoft submitted that Nintendo is a

strong competitor in consoles and the Nintendo Switch is one of the most
successful consoles of all time. 306 Microsoft further submitted that while the
Nintendo Switch is []. 307 []. 308

7.60 In response to the suggestion that Nintendo’s offering is likely different to

other consoles by virtue of its target audience, Microsoft submitted that
Nintendo does not just target a family-friendly audience, and instead offers a
broader range of ‘mature’ content than Xbox which are also actively
marketed. 309 Microsoft stated that several of Nintendo Switch’s latest
exclusives have been non ‘family-friendly’ games and have received mature
ratings. 310 Microsoft also submitted data showing that the distribution of
gamers by age on all three consoles is similar, to further substantiate that
Nintendo caters to a similar audience as Xbox and PlayStation. 311 It also
submitted that, according to the same data, on average the Switch has a
lower proportion of young gamers [] than SIE and Microsoft’s consoles. 312

7.61 Microsoft also submitted that the portability of the Nintendo Switch offers a
significant advantage over PlayStation and Xbox, as users have the ability to
play games away from their home base. 313

Our assessment

7.62 In assessing the closeness of competition between the Xbox, PlayStation and
Nintendo consoles, we first note the following preliminary points in relation to
the competitive landscape between the three consoles:

7.63 First, Microsoft’s internal documents track PlayStation more closely than
Nintendo, with Nintendo often being absent from internal competitive

305 Microsoft response to TOH 1 working papers.

306 Microsoft response to TOH 1 working papers.
307 Microsoft, response to TOH 1 working papers.
308 Microsoft, email to the CMA.
309 Microsoft cited examples of games rated ‘Mature’ by the Entertainment Software Rating Board (“ESRB”) for

violence, blood and gore, partial nudity and strong language, such as Bayonetta 3 and Megami Tensei V.
310 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.46b.
311 Microsoft response to TOH 1 working papers; Microsoft response to the CMA supplementary evidence paper.
312 Microsoft response to the CMA supplementary evidence paper.
313 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.25.

101
 assessments. 314 One Microsoft internal document assesses []. 315 Another
email exchange between []. 316

7.64 Second, previously expressed views of Microsoft show that the Parties
consider Nintendo to compete less closely with Xbox compared to the
PlayStation. For example, Microsoft’s vice-president of business development
for gaming, media, and entertainment in 2021 stated in the US Epic Games v.
Apple case that ‘the most direct competitor for hardware sales would be the
SIE PlayStation’; and when asked if there were any others, stated that the
Nintendo Switch is also a competitor, but ‘to a much lesser extent’. 317

7.65 In addition to the above evidence, we have considered the following aspects
of competition in the rest of this section:

(a) Technical specifications and price

(b) Content type and target audience

(c) Consumer preferences

Technical specifications and price

7.66 We first assess the closeness between the three consoles in terms of their
technical specifications. This is important because gamers require high-
performance devices (eg in terms of memory, processing speed) to play
graphically intensive games such as CoD. To the extent that a console does
not meet these technical requirements, it is likely to compete less closely with
other consoles for those games.

7.67 Xbox and PlayStation are similar in terms of their technical specifications,
while Nintendo’s technical specifications differ significantly from either of the
two. As can be seen from the table below, the Nintendo Switch has
significantly lower energy usage, fewer cores in the CPU, a lower clock
speed, a significantly lower value of graphics processing power and
performance and a smaller RAM as compared to PlayStation 5 or Xbox Series
X. 318 By contrast, the PlayStation 5 and Xbox Series X are very similar to
each other in terms of these parameters.

314 Parties Internal Document; Parties Internal Document; Parties Internal Document.
315 Parties Internal Document.
316 Parties Internal Document.
317 Testimony of [], Epic Games, Inc., v. Apple, Inc., 5 May 2021.
318 This is measured in TFLOPS, or Tera floating point operations per second.

102
Table 7.2: Comparison of technical characteristics of the Nintendo Switch, Xbox X and
PlayStation 5
Specification Xbox Series X PlayStation 5 Nintendo Switch (TV mode)

Launch Date 10 November 2020 12 November 2020 8 October 2021

Energy Usage 153W 208.8W 7W
CPU: number of cores 8 8 4
CPU: clock speed 3.8 GHz 3.5 GHz 1 GHz
GPU: performance 12 TFLOPS 10.3 TFLOPS ~0.77 TFLOPS
RAM 16 GB 16 GB 4 GB

Sources: ‘Unveiling New Details of PlayStation 5 Hardware Technical Specs’ accessed by the CMA on 17 January 2023;
‘Energy Efficiency: Active Power Consumption’ accessed by the CMA on 17 January 2023; ‘Xbox Series X’ accessed by the
CMA on 17 January 2023; ‘About power options on Xbox One and Xbox Series X|S’ accessed by the CMA on 17 January
2023; https://www.nintendo.com/switch/tech-specs/ accessed by the CMA on 17 January 2023; [] response to the CMA’s
RFI; ‘Nintendo Switch OLED Model’, accessed by the CMA on 24 January 2023; and ‘PlayStation 5 Release Date, Pricing, and
Launch Games’, accessed by the CMA on 24 January 2023.

7.68 The Nintendo Switch is also different in terms of its portability. Unlike Xbox or
PlayStation, which need to be connected to a screen to function, the Nintendo
Switch is a portable device that contains its own screen. The Switch is also
significantly cheaper (currently at most £310) than the flagship Xbox Series X
and PlayStation 5 consoles (sold at £450 and £480, respectively), and is
closer in price to the technically less capable Xbox Series S console (sold at
£250).

7.69 Activision’s internal documents note the technical limitations of the Nintendo
Switch console. For example, one Activision document notes in an early-stage
assessment that, to produce a CoD title on the Nintendo Switch, the CoD
game would need [] 319 (whereas most current CoD titles require from 125-
175GB of storage on console or PC). 320 The document also refers to Apex
Legends’s []. 321 Another Activision document analysing potential studios
[] CoD assesses the additional work required [] and notes technical
issues in other games []. 322

7.70 We have also seen evidence that large shooter games do not run as well on
Nintendo’s consoles due to its technical differentiation. One third party
submitted that graphically intensive shooters may often be targeted originally
at PlayStation and Xbox due to the specific characteristics of their console
performance, and that porting to the Nintendo Switch may require financial
investment and compromises on graphical quality, or the use of cloud-gaming
solutions. 323

319 Activision Internal Document.

320 See for example, Minimum and Recommended System Requirements for Call of Duty: Warzone Caldera on
PC (activision.com) and Minimum and Recommended System Requirements for Call of Duty: Modern Warfare on
PC (activision.com), accessed by the CMA on 17 January 2023.
321 Parties Internal Document.
322 Parties Internal Document.
323 [] response to the CMA’s.

103
7.71 Publishers’ views similarly indicated that developing a game for Switch is a
significantly different task relative to doing it for Xbox and PS due to its
technical differences. One publisher stated that it encountered technical
difficulties when bringing a game to Nintendo Switch but no difficulty in bring
the same game to Xbox or PlayStation. The publisher noted that the Switch’s
limited graphics and storage are technical limitations that affect the
performance of competitive games more than that of game(s) brought to Xbox
or PlayStation. 324 Another publisher stated that several of its games are not
available on Nintendo as Nintendo has different capabilities from PlayStation
and Xbox. 325

7.72 We note the Parties’ submission that challenges with porting a game to
Nintendo Switch has not impacted Nintendo’s ability to compete on the
downstream console market, as it offers more games than Xbox and
PlayStation, including major games such as Apex Legends, Fortnite and
Doom Eternal. 326 However, we consider the evidence above shows that,
relative to the Xbox and PlayStation, the Nintendo Switch (i) does not
currently offer the same suite of graphically intensive games that PlayStation
and Xbox compete on (with the exception of a few games such as Fortnite
and Apex Legends), (ii) may not be capable of offering certain graphically
intensive multiplayer games (such as CoD), and (iii) does not offer a similar
user experience (eg, in terms of storage, graphics, and framerate).

7.73 We also note the Parties’ submission that Nintendo’s partnership with Ubitus’
cloud streaming technology to release Resident Evil Village on Switch
enables gameplay with levels of graphical fidelity comparable to that found on
a high-level PC, PlayStation 5 or Xbox Series X. 327 However, as discussed
above, we consider that there are currently significant differences between
cloud gaming and gaming on consoles (eg, the need for an internet
connection to stream games from cloud gaming services). Also, the ability of
the Switch to connect to a third party cloud gaming service provider would not
make it a closer competitor to Xbox and PlayStation in the console gaming
market.

7.74 Overall, the evidence shows that the product characteristics of Nintendo
Switch are significantly different from those of Xbox and PlayStation, including
its technical specifications, capability to host graphically intensive games and
prices. Xbox and PlayStation are more similar in this respect.

324 [] call note.

325 [] response to the CMA’s RFI.
326 Microsoft, response to working papers; and Microsoft response to the phase 2 Issues Statement, 31 October

2022, paragraph 3.46c.

327 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.46c.

104
 Content type and target audience

7.75 Next, we have considered the target audience and content type of Xbox,
PlayStation and Switch.

7.76 Microsoft’s internal []. One Microsoft internal document points to the
differences in the technical strategies of Xbox/PlayStation and Nintendo
consoles []. 328 This internal document considers that []. 329

7.77 Third parties noted the differences in Nintendo’s business strategy in terms of
developing own content for its consoles—making its proposition sufficiently
differentiated from other gaming ecosystems; and less dependent on the
availability of specific third party games. Two third parties submitted that
Switch’s most popular games are largely developed and published by
Nintendo and are based on Nintendo’s owned intellectual property. 330

7.78 Third parties also noted the differences in the type of content and target
audience of Nintendo, compared to PlayStation or Xbox. Two third parties
submitted that Nintendo content typically focuses on its family-based target
audience; 331 and that its games tend to be family-friendly (eg, Mario Kart and
Super Mario). 332

7.79 We also assessed data from the Parties and third parties on the most popular
games in the UK on Nintendo Switch, SIE PlayStation and Microsoft Xbox,
ranked by revenue, gametime and MAUs, including their Pan European Game
Information (PEGI) content age rating. 333

7.80 This data shows that a [] of titles in the top 50 titles on the Switch were
rated PEGI 3 or PEGI 7, than the equivalent lists on PlayStation and Xbox
(see Table 7.3 below). 334 We consider that this shows that the most popular
content on the Switch generally falls under a different category (ie, more
family-friendly) from the most popular content on PlayStation and Xbox.

Table 7.3: Number of top 50 titles on console with PEGI 3 or PEGI 7 ratings in 2021
Ranked by Revenue Ranked by Gametime Ranked by MAUs

PlayStation [] [] []

Xbox [] [] []
Switch [] [] []

328 Parties Internal Document.

329 Parties Internal Document.
330 [] call note; and [] response to the CMA’s RFI.
331 [] response to the CMA’s RFI; and [] call note.
332 [] response to the CMA’s RFI; and [] response to the CMA’s RFI.
333 [] response to the CMA’s RFI; [], Annex to response to the CMA’s RFI; Microsoft, Annex to response to

the CMA’s RFI; and ‘PEGI age ratings’, accessed by the CMA on 20/03/2023.
334 PEGI 3 titles are suitable for all ages, while PEGI 7 titles have content with scenes or sounds that can possibly

be frightening to younger children.

105
Source: CMA calculations

7.81 Using the same data, 335 we analysed the extent to which there is an overlap
in the top 50 titles across the Switch, PlayStation, and Xbox in the UK for
2020 and 2021 (see Table 7.4 below). This data shows that the overlap in the
top 50 titles between Xbox and PlayStation is [] than that of the Switch with
either PlayStation or Xbox. For example, when ranked by revenues in 2021,
the number of top titles overlapping between PlayStation and Xbox was []
than the number of titles overlapping between the Switch and either Xbox or
PlayStation. From the titles that overlap between PlayStation and Xbox in
2021 (and are therefore not exclusive titles), [] ([] by revenue, gametime,
and MAUs, respectively) are not available to be played on the Switch. Most of
the remaining titles, despite being available on the Switch, [].

Table 7.4: Number of UK top 50 Titles by console that overlap by year

Year Revenue Gametime MAUs

PlayStation/Switch 2021 [] [] []

Xbox/Switch 2021 [] [] []
Xbox/PlayStation 2021 [] [] []
PlayStation/Switch 2020 [] [] []
Xbox/Switch 2020 [] [] []
Xbox/PlayStation 2020 [] [] []

Source: CMA calculations

7.82 The above evidence suggests that the Switch is differentiated from
PlayStation and Xbox. We note the Parties’ submission above that the
distribution of gamers by age on the Switch is not very different to Xbox or
PlayStation. We also note that the Switch offers some content for a mature
audience. However, the technical differences noted above mean that users do
not generally play certain types of games on the Switch which are more
popular on Xbox and PlayStation. The evidence on the extent of overlap
between the titles that are popular on each of the consoles also shows that
the Switch’s most popular titles tend to be targeted at a family audience. We
consider that family-friendly content is differentiated and may be more
complementary to the other consoles’ content. The same evidence also
shows that the overlap between the Switch’s most popular titles and the most
popular titles on each of PlayStation and Xbox is low compared to the
corresponding overlap between PlayStation and Xbox. Therefore, we consider
that while the Switch may be a substitute to either PlayStation or Xbox for
some gamers, overall it is likely to be less so.

335 [] response to the CMA’s RFI, question 1; [], Annex to response to the CMA’s RFI; and Microsoft, Annex
to response to the CMA’s RFI.

106
 Consumer preferences

7.83 Consumer behaviour in relation to multi-homing across consoles is likely to

indicate closeness of competition to a certain degree, with consumers being
more likely to own another console if they consider it to be sufficiently
differentiated to their current console.

7.84 We described the evidence on multi-homing across consoles above. While

the exact figures by which gamers multi-home across consoles varied based
on the timing of the analysis, the sample considered and the console
generation, overall, most of the evidence suggests that people tend to multi-
home more often between the Switch and PlayStation/Xbox than between
PlayStation and Xbox (within the same generation). This supports the view
that the Switch is less substitutable to the PlayStation or Xbox.

7.85 On the other hand, evidence showed that the extent of multi-homing between
the PlayStation and Xbox was fairly low, indicating that these are closer
substitutes.

Conclusion on closeness of competition among consoles

7.86 The evidence shows that Xbox, PlayStation and Nintendo all compete in the
console market on the basis of technical specifications, content (including
range and type), and price.

7.87 The evidence also shows, however, that whilst Xbox and PlayStation compete
closely with each other, Nintendo competes less closely with Xbox and
PlayStation:

(a) Microsoft and SIE offer non-portable consoles with similar technical
specifications. Nintendo’s flagship console, the Nintendo Switch, offers a
portable console for a lower price and with significantly lower technical
specifications.

(b) Whilst Nintendo offers some content for a mature audience, its most
popular titles are targeted at a family audience.

(c) The overlap between the Switch’s and either of PlayStation’s or Xbox’s
most popular titles is low compared to the corresponding overlap between
PlayStation and Xbox.

(d) Gamers more often own a Nintendo console and an Xbox or PlayStation
than own an Xbox and a PlayStation. This suggests that Nintendo is less
substitutable with Xbox or PlayStation than the latter are with each other.

107
Availability of Activision’s content on MGS services

7.88 As previously explained, MGS services have become a feature in the offering
of gaming console providers and an additional lever for these providers to
compete in the gaming consoles market.

7.89 Activision has previously made some of its content available on console MGS
services. In that context, this section focuses on the following question: to
what extent would Activision’s content be made available on console MGS
services absent the Merger?

7.90 The greater the extent to which Activision’s content would be made available
on these services, the more relevant these services would be for foreclosure
of gaming console rivals. In other words, if absent the Merger, Activision’s
games would be available on rivals’ console MGS services (eg PS+), the
Merged Entity could use that as an additional lever to foreclose those rivals.
For example, the Merged Entity could do that by making either current and/or
future Activision content unavailable on rival console MGS services (ie
exclusive to XGP). If, instead, Activision’s games would not have been
available on those services at all absent the Merger, the Merged Entity would
be unable to foreclose rival gaming consoles using content on MGS services
as a foreclosure mechanism.

7.91 In this and the following sections, we focus on console MGS services and use
the phrase ‘MGS services’ to mean console MGS services, unless specified
otherwise.

Parties’ submissions

7.92 Microsoft submitted that subscription services are expected to coexist and
compete alongside other payment models, whilst remaining only a limited
segment of the broader gaming industry. 336 Microsoft submitted that [].
Microsoft also submitted that, although MGS services were growing, the vast
majority of gaming revenues would still come from gamers purchasing
individual games. 337

7.93 The Parties submitted that Activision currently makes only limited back-
catalogue titles available on subscription, and only on a limited time-period
basis. 338 Microsoft also submitted that Activision has never published any
newer content on MGS services and has no intention to do so in the future. 339

336 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.25.
337 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.9.
338 CMA, Phase 1 Decision, 12 October 2022, paragraph 212(e).
339 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.35.

108
 The Parties also submitted that Activision does not currently offer and has no
current plans to offer an MGS service similar to XGP. 340

7.94 Microsoft submitted that an important reason for Activision not offering current
content on MGS is because MGS sales would cannibalise Activision’s B2P
sales, which are a key source of and driver of revenue for Activision. It
submitted that the []. According to Microsoft, in the rare cases where
Activision has contemplated placing its content on MGS services, its ordinary
course documents consistently reflect Activision’s view that MGS services,
regardless of platform, severely cannibalise B2P sales, particularly in the case
of newer releases. 341

7.95 Activision submitted that [] for several other reasons, including that it would
[] which, according to Activision, is important to ensure a high-quality
experience in its games. 342 Activision submitted that these considerations
apply even if [], and that Activision’s incentives as an independent game
developer and publisher will remain the same. 343

7.96 Activision also submitted that Activision's senior leadership has never
supported []. 344

7.97 Microsoft also submitted that []. 345

Our assessment

Future growth of MGS services

7.98 In this section, we assess the growth prospects of MGS services based on
evidence from independent third party forecasts, the Parties’ forecasts and
internal documents, and views and evidence provided by third parties during
our investigation.

Forecasts

7.99 Microsoft provided independent third party forecasts relating to the growth
prospects of MGS services. One of these is a report by IDG (an industry
analysis firm), which estimated that the share of MGS services relative to B2P

340 CMA, Phase 1 Decision, 12 October 2022, paragraph 210.

341 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.28.
342 Activision response to the working paper.
343 Activision response to the working paper.
344 Activision response to working paper.
345 Microsoft response to working papers.

109
 would grow from []% in 2021 to []% in 2026. 346 We note that these IDG
estimates might include any subscription services, including single-game
subscriptions and services like PS+ Essential, which mainly offer access to
online multiplayer and only a few monthly games. We do not consider these
services to fall into our definition of MGS services for the purpose of our
assessment (ie services which offer a relatively wide range of games through
a subscription). Therefore, in the context of our assessment the IDG
estimates might overstate the share of MGS relative to B2P.

7.100 Microsoft submitted its own estimates showing that MGS services are likely to
show [] growth relative to B2P over the next few years. It estimates that
MGS services as a proportion of all transactional revenue—including B2P and
in-game transactions—will grow from []% in 2022 to []% in 2030. 347

7.101 We note, however, that Microsoft’s estimate may not accurately capture the
growth of MGS services relative to B2P. A more accurate comparison would
either exclude in-game transactions on B2P or include them for both B2P and
MGS Services. We also note that in the same estimate, Microsoft predicts the
number of XGP subscribers will grow from [] million to [] million over the
period 2022-2030. Microsoft submitted that actual realised figures for XGP
subscribers demonstrated that []. 348

7.102 Taken together, these forecasts suggest some growth in MGS services
relative to B2P in the next 5-8 years. Although it is difficult to arrive at a
precise estimate, these suggest that MGS services will continue to grow and
represent a higher proportion of the overall console gaming market over time.
However, these forecasts also suggest that MGS will remain smaller than B2P
over the same time period.

Parties’ internal documents

7.103 Several of the Parties’ internal documents suggest that, along with the
traditional B2P model, the importance of MGS services in the gaming industry
is increasing:

(a) One Microsoft internal email from October 2018 states that []. 349 This
document suggests that []. This is in line with the IDG forecast

346 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.24; and [] Internal
Document; these estimates are for console gaming only, ie they exclude PC, mobile, and cloud gaming; we
calculated the share of MGS relative to B2P by dividing ‘subscription’ by the sum of ‘subscription’, ‘digital full
game’, and ‘physical SW’ (physical software).
347 Parties FMN; Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.24; and

Microsoft Internal Document.

348 Microsoft, email to CMA.
349 Microsoft Internal Document.

110
 discussed above, although the email is relatively dated compared to other
documents presented in this section.

(b) One Activision internal document from 2020 states that []. The
document adds that MGS services are increasing on console and mobile
and that the free to play, premium and subscription business models will
coexist at scale. 350

(c) A document from late 2021 submitted by Microsoft states that []. 351

(d) Another Microsoft document from 2020 explains that []. 352

7.104 Microsoft submitted that these documents are []. Microsoft submitted its
more recent Gaming Strategy documents show that []citing an internal
document which states that ‘[]’. 353 We consider that these documents
submitted by Microsoft do not contradict the other documents presented
above.

7.105 According to Activision, the document in paragraph 7.103(b) above shows

that Activision [] for gaming content. According to Activision, the document
does not suggest that []. 354

7.106 Third party reports submitted by the Parties also suggest MGS services will
be increasingly important:

(a) A third party report from July 2020 held by Activision and shared with the
CMA states that the expansion of new business models has boosted
monetisation in the gaming industry, one of those models being MGS
services. 355

(b) Another third party report from 2022 provided to the CMA by Activision
states that MGS services are one ‘key development’ for the gaming
industry. The document explains that the gaming industry is going through
a transition towards digital services, particularly MGS services, and that
gamers will be attracted to these services because they would have a
lower cost access point and a large game library. 356

350 Activision Internal Document.

351 Microsoft Internal Document.
352 Microsoft Internal Document.
353 Microsoft response to working papers; Microsoft Internal Document.
354 Activision response to TOH 2&3 Subscription or Cloud working paper.
355 Activision Internal Document.
356 Activision Internal Document.

111
 (c) A third party report from October 2021 submitted by Microsoft states that
cloud gaming and subscriptions will become an increasingly important
part of the games market. 357

7.107 Activision submitted that the third party reports it provided above [] – and in
any case any such assertion has not been borne out in practice, where in fact,
free-to-play games account for the vast majority of growth in the gaming
industry today. 358 We note that the relevant question in this section is whether
MGS is growing compared to B2P, not F2P. In particular, adding a F2P game
to an MGS service does not lead to cannibalisation of upfront sales because
by definition there are no upfront sales on F2P games.

7.108 Some of Microsoft’s internal documents also talk about the future importance
of XGP to its business, which is relevant to the potential prospects of growth
in MGS services generally. For example:

(a) One Microsoft internal document suggests that []. 359

(b) Another Microsoft document from August-November 2021 indicates

[].’ 360 This suggests that [].

7.109 These documents show that Microsoft was confident in significant growth in
MGS services and XGP in the future.

Third party evidence

7.110 We received third party evidence showing that subscription-based gaming is

expected to grow relative to B2P, even though the precise extent and pace of
such expected growth was unclear from the evidence:

(a) A publisher [] submitted that it believed that in the future consumers
would still want to purchase individual games but noted that the
percentage of consumers purchasing subscription services was
increasing. It added that subscription services would become a sizeable
part of the game industry ([]) and that it wanted to be present in it. It
also explained that, although some players spent time on a limited
number of games per year, the subscription model was likely to become
increasingly attractive as the publishers’ first-party catalogues strengthen

357 Microsoft Internal Document.

358 Activision response working paper.
359 Microsoft Internal Document; [].
360 Microsoft Internal Document.

112
 each year. It stated that even casual gamers would find the subscription
model interesting. 361

(b) One internal document from a gaming platform supplier [] shows that
this competitor is expecting the MGS market to grow and, as a
consequence, is investing in its subscription services. The document
notes that []. 362

(c) One competitor [] speculated that the gaming sector had already seen
a recent shift from purchasing individual games (either digitally or
physically) towards MGS services, and it thought that this was as a result
of consumers’ desire to access a wide variety of content. 363

(d) One publisher [] told the CMA that MGS services are still nascent and
that it believes that they may become a more notable way for players and
consumers to experience new ways of accessing more games. 364

(e) One competitor [] explained that the gaming industry was gradually
moving towards MGS service business models. 365

Conclusion on future growth of MGS services

7.111 We consider that the evidence from industry analysts, Microsoft’s forecasts,
the Parties’ internal documents and most third party evidence indicates that
MGS services are growing in importance within gaming, and will become a
sizeable part of the gaming industry. However, we note that the available
estimates of MGS growth vary significantly depending on the source.

7.112 The evidence also suggests that, while MGS is expected to grow relative to
B2P, MGS is likely to remain a smaller part of the console gaming services
market in the foreseeable future.

7.113 This expected growth could still make placing games on MGS services more
attractive, as it would reduce the rate of cannibalisation of B2P sales (as the
proportion of B2P sales shrinks). However, whether this would be enough to
change Activision’s incentives depends on the starting point, ie Activision’s
current incentives and its current stance and behaviour towards MGS. This is
assessed below.

361 [] response to the CMA’s RFI.

362 [] Internal Document.
363 [] call note.
364 [] call note.
365 [] call note.

113
Publishers’ behaviour on adding content to MGS services

7.114 The behaviour of rival publishers in relation to the supply of content, including
their AAA titles, is a potential indicator of Activision’s incentives to supply CoD
and other Activision content to MGS services.

7.115 Rival publishers such as Ubisoft and Electronic Arts already offer a selected
portfolio of their games as part of their own MGS service on PC, console or
cloud, either on a standalone basis or as part of third party MGS services. In
particular:

(a) Electronic Arts currently offers its EA Play MGS service as part of XGP,
including some latest releases and AAA game franchises like FIFA and
Battlefield. However, we note that the latest version of FIFA is not
available on the service. 366 We also note Electronic Arts’ submission that
the games it offered as part of EA Play on XGP are added to EA Play
around 8-12 months after their release. 367

(b) Ubisoft currently offers some of its AAA games as part of XGP – these are
mostly older releases with the exception of Tom Clancy's Rainbow Six
Extraction, although this was not released day and date on XGP. 368

(c) Ubisoft also offers its Ubisoft+ MGS service as part of Amazon Luna,
including some AAA game franchises and latest releases like Assassin’s
Creed Valhalla, Far Cry 6, and Tom Clancy's Rainbow Six Extraction. 369
We note this example is not relevant to console, but rather to cloud
gaming services.

7.116 Microsoft submitted that, unlike Ubisoft and Electronic Arts, Activision does
not have a wide portfolio of games to leverage. Microsoft also submitted that
Activision cannot use its main franchise on console to attract gamers that it
can monetise on other titles, simply because it does not have other titles to
benefit from. 370

7.117 Activision submitted excerpts from public interviews by the CEOs of SIE and
Take-Two, which expressed reservations about adding their major titles day-
and-date to MGS services. 371

366 See EA Play | Xbox, accessed by the CMA on 16 January 2023.

367 [] response to the CMA’s s109 notice.
368 See Xbox Game Pass Games Library | Xbox, accessed by the CMA on 16 January 2023.
369 See Amazon.com: Play Ubisoft+, accessed by the CMA on 16 January 2023.
370 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 4.31-4.32
371 Activision response to the Activision Titles on Subscription or Cloud Gaming working paper.

114
7.118 We consider that the evidence from publishers’ behaviour is mixed. On one
hand, some large publishers like Electronic Arts and Ubisoft are adding their
games, including AAA content and some latest releases to third party MGS
services. On the other hand, these games were almost never added to the
service on the date of their release and, in the case of some big franchises
(eg FIFA), they are older releases. We also note based on public sources that
some publishers (eg Take-Two) seem to be more cautious in their approach
towards MGS services.

Parties’ internal documents

7.119 Activision’s internal documents suggest that Activision was. For example:

(a) One document dated May 2020 discusses []. 372 The document
assumes []. 373

(b) Another document dated May 2020 discusses similar []. The document
notes that []. However, the document suggests that []. 374

(c) In a document dated August 2021, Activision was at that time []. The
document also explains that []. 375

7.120 One internal Activision email from [] suggests that Activision saw []. The
same employee also notes that third party publishers putting their games on
XGP []. 376 Activision submitted that this document is written by []. 377

7.121 Despite this [], Activision’s internal documents show that it []. These []
included []. However, we note that the most advanced negotiations []. In
particular:

(a) One internal document shows that Activision had presented to []. 378

(b) One document dated April 2020 shows that []. 379

(c) One Activision document dated October 2021 shows that []. This
document states that []. 380

372 [].
373 Activision Internal Document.
374 Activision Internal Document.
375 Activision Internal Document.
376 Activision Internal Document.
377 Activision response to TOH 2&3 Subscription or Cloud working paper.
378 Activision Internal Document.
379 Activision Internal Document.
380 Activision Internal Document.

115
 (d) Another document dated December 2021 discusses []. The document
shows that []. The document notes that []. The document also notes
that []. 381 Activision submitted that this document presents []. 382

(e) [] – one Activision document [] shows that []. The document also
notes that []. 383 A [] document from May 2022 confirmed that [].
However, the document also notes that []. 384

7.122 It is not entirely clear why []. The documents suggest that the main reason
was that Activision [] and that [].

7.123 The documents also show that Activision has [] and might have []. []
Activision has submitted that the position of Activision's senior leadership,
including its CEO, has always been [] these internal documents [].

7.124 Microsoft submitted that []. 385 In support of this statement, Microsoft
submitted a document dated March 2020 that []. The document shows that
[]. 386

7.125 This was not, however, []. Subsequent internal documents from Microsoft
[]. For example, one document [] shows that []. The document notes
that []. The same document shows that, []. 387

7.126 Activision’s internal documents also show that it considered []:

(a) One internal document shows that Activision []. 388

(b) Another document presents a detailed analysis of a possible []. The

document notes that []. 389

(c) One Activision document []. 390

7.127 Activision submitted that the documents identified by the CMA are isolated
examples of businesspeople discussing potential ideas for the business, and
that they []. Activision also submitted that no version of []. 391

381 Activision Internal Document.

382 Activision response to working paper.
383 Activision Internal Document.
384 Microsoft Internal Document.
385 Microsoft response to working papers.
386 Microsoft Internal Document.
387 Microsoft Internal Document.
388 Activision Internal Document.
389 Activision Internal Document.
390 Activision Internal Document.
391 Activision response to working paper.

116
7.128 These documents show, however, that Activision was actively considering
[]. Its discussions concerning MGS services []. The documents also show
that senior employees within Activision, [].

Conclusion on availability of Activision content on MGS services

7.129 This evidence suggests that Activision is open to making at least some of its
back catalogue and older games available on MGS services. [].

7.130 The evidence also shows, however, that Activision is concerned about the
potential impact on its B2P revenues of placing any games – especially newer
games and day and date releases – on MGS services. []. Up until now,
Activision has made its content available on subscription only to a limited
extent, both in terms of the type of content available – namely, older titles –
and for a limited time period.

7.131 We consider it relevant context that other AAA games (facing similar
cannibalisation effects) already make some of their content available on
console MGS services (eg XGP). However, we note this is almost always
limited to back catalogue or older games. We also note that some large
publishers do not currently add their games to MGS services.

7.132 Based on this evidence, we consider it unlikely that Activision would make its
most valuable games – such as CoD – available on MGS services on the date
of release in the foreseeable future absent the Merger. We believe the
evidence indicates, however, that Activision would likely place increasingly
valuable parts of its gaming catalogue on MGS services as these services
continue to grow. This would likely include back-catalogue games, as well as
Activision’s latest releases, although some time after they are released (ie, not
on the date of release).

7.133 This therefore implies that the Merged Entity has the option of denying some
or all of Activision’s gaming catalogue on MGS services to rival gaming
consoles. Further, the likelihood of more valuable content appearing on MGS
services could increase, over time, any ability and incentive of the Merged
Entity to foreclose rivals. We assess the significance and relevance of any
such content on MGS services in relation to the Merged Entity’s ability and
incentive to foreclose rival consoles collectively alongside B2P content in the
sections that follow.

Framework for assessment

7.134 Vertical theories of harm involve the merged entity harming the ability of its
rivals to compete post-Merger, for example through input foreclosure; raising

117
 effective prices or worsening non-price aspects, such as quality, of an input to
its rivals (partial input foreclosure); or refusing to supply them completely (total
input foreclosure). Such actions may harm the ability of the merged entity’s
rivals to provide a competitive constraint on the merged entity, and thereby
lead to higher prices, worse quality or less choice for consumers. The CMA
only views foreclosure as anticompetitive where its effect is to reduce
competition sufficiently to give rise to an SLC in the affected market, not
where it merely disadvantages one or more competitors.

7.135 The concern under an input foreclosure theory of harm is that the Merger may
lead to Microsoft using Activision’s games to foreclose rival console providers,
harming their current and future ability to compete in the supply of consoles;
thereby leading to lower quality, higher prices or less choice for consumers.
Specifically, we consider whether Microsoft could harm its rivals’
competitiveness and thus lessen current and future competition in console
gaming services through the following foreclosure strategies (which could be
used in isolation or combination):

(a) total foreclosure: making either current and/or future Activision content
unavailable on rival consoles (ie exclusive to Xbox);

(b) partial foreclosure, which includes several potential strategies such as:

(i) making Activision content available for release on rival console

gaming platforms at a later date compared to Xbox (ie timed
exclusivity);

(ii) degrading the graphical quality (eg resolution and framerate) of

Activision gaming content available to rival console gaming platforms;

(iii) making features or upgrades of Activision games unavailable to other

console gaming platforms (ie content exclusivity); and/or

(iv) raising the wholesale price of Activision content on rival console

gaming platforms.

(v) We are assessing overall whether the Merged Entity would have the
ability and incentive to engage in partial and/or total foreclosure
strategies and the effect of these strategies. In our main assessment,
we do not presuppose any specific foreclosure strategy. The Merged
Entity may focus first on partial foreclosure strategies with the
expectation of shifting users to its platform over a period of time,
gradually shifting to a total foreclosure strategy. This assessment
does not assume any particular combination or order of foreclosure
strategies but considers more generally whether the Merged Entity

118
 would have the ability and incentive to engage in these strategies,
and how that would affect competition. However, we also assess
whether the Merged Entity would have the ability and incentive to
foreclose PlayStation post-Merger through partial foreclosure
strategies alone. In this additional assessment, we focus on those
partial foreclosure strategies that would stop short of total foreclosure.

7.136 Our assessment below focuses on the strategies that the Merged Entity could
engage in using CoD, as CoD constitutes []% of Activision revenue on
consoles. 392 We note, however, that Activision makes other games available
on PlayStation and other consoles (such as Overwatch, another shooter
game and Diablo, an action role-playing game). To the extent that the Merged
Entity could use these other Activision games to foreclose rivals, this could
worsen any effect on competition that the Merger could have in the market for
console gaming services, as assessed below.

7.137 We consider that any foreclosure of Microsoft’s console rivals would affect
console hardware and the related storefronts (including subscription services).
This is because gaming consoles are integrated in terms of their offer of
hardware and digital storefronts (as discussed in the market definition section
in Chapter 5).

7.138 CoD is not available on Nintendo consoles, nor have we seen evidence to
suggest it would become available on Nintendo absent the Merger. As noted
above, we consider that Nintendo competes less closely with Xbox as
compared to PlayStation. Therefore, we consider that SIE is currently the
strongest competitive constraint on Microsoft, and to assess whether
competition is affected as a result of the Merger, we are assessing whether
this closest rival in the console gaming market would be foreclosed. This
theory of harm, therefore, focuses on the potential impact of the Merger on
SIE. In our assessment of whether Microsoft may harm SIE’s ability to
compete in consoles by denying or worsening its access to Activision’s
games, we follow the framework set out in the MAGs for assessing input
foreclosure theories of harm. 393 We consider whether three cumulative
conditions are satisfied: 394

(a) Would the Merged Entity have the ability to use its control of inputs to
harm the competitiveness of SIE?

392 Parties, submission to the CMA.

393 CMA129, paragraphs 7.9-7.22.
394 CMA129, paragraph 7.9.

119
 (b) Would it have the incentive to actually do so, ie would it be profitable in
the short-term and/or long-term?

(c) Effects of foreclosure: would the foreclosure of SIE substantially lessen

overall competition between console gaming providers?

Ability to foreclose

7.139 In order to assess whether the Merged Entity would have the ability to
foreclose rivals in the downstream market for console gaming services, we
have considered evidence on the following factors:

(a) the importance of content to a gaming platform such as a console,

including SIE’s PlayStation;

(b) the importance of Activision content, particularly CoD, to SIE; and

(c) the extent to which rival gaming consoles, particularly SIE, can substitute
CoD with effective alternatives.

7.140 The MAGs explain that the CMA will typically focus on two issues when
assessing whether the Merged Entity will have the ability to foreclose its
rivals, namely: (i) market power upstream (looking at inter alia the structure of
the upstream market and ability of downstream rivals to switch to a range of
effective alternatives); and (ii) the importance of the input (assessing whether
the input the merged entity will supply plays an important role in shaping
downstream competition). In this case, this involves assessing whether
Activision’s content gives it market power in the upstream market for console
game publishing, and whether Activision content is an important input to
PlayStation in particular. We explore these questions below through the three
factors set out above.

7.141 The three factors above allow us to assess both the upstream market power
of Activision stemming from its content, in particular CoD, within the console
game publishing market, and the importance of Activision’s content to
PlayStation in particular in the downstream market for console gaming
services. After assessing these factors, we assess contractual arrangements
in relation to the Merged Entity’s ability to foreclose. Although most of the
assessment in this section is focused on total foreclosure, the evidence we
used is largely relevant to standalone partial foreclosure strategies as well.
We have also assessed specific evidence and arguments on partial
foreclosure strategies.

120
Importance of content

Parties’ views

7.142 The Parties submitted, based on a detailed conjoint analysis conducted in the
ordinary course of business, that the most important factor in the choice of
console is the price of the console. That is followed by the type of console
which the gamer currently owns (ie, whether the owner already has an Xbox
or PlayStation), with content offered by the new console being further down
the list. 395 The Parties, however, also acknowledge in other submissions that
content plays a role in the success of a gaming platform like a console. For
example, in response to the CMA’s issues statement, Microsoft refers to the
gaming industry as ‘content-focused’ while discussing content exclusivity. 396

Our assessment

Third parties’ views and internal documents

7.143 Overall, third parties have stated that content is one of the main reasons for
consumers’ choice of platform:

(a) One third party [] told the CMA that the key driver for consumers is
content. The same company also added that if a user is going to switch to
another platform, content will generally be the motivation for that. The
third party explained that it attracts and retains new customers by
marketing to consumers that it offers the best range of content. 397

(b) Another gaming platform supplier [] also explained that the quality and
number of games as well as the franchises offered are the most important
factors driving consumers' decisions. 398 In particular, this competitor
stated that “quality of games” is the single most influential driver of
engagement for a platform, above other attributes including graphics,
interface, and hardware. It submitted that, for most gamers, popular
content is more important than any customer loyalty towards a particular
platform. 399

7.144 Various internal documents from a third party [] further show the importance
of content, [].

395 Microsoft response to working papers.

396 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.66.
397 [] call note.
398 [] submission to the CMA.
399 [] response to the CMA’s RFI.

121
 (a) An internal document dated [] emphasises that the game library
available on a platform is an important element of the gaming experience
and thus it is highly important to consumers. The document says that the
size [], and this is also not likely to change following Xbox’s significant
investments. 400

(b) The same document emphasises the importance of exclusives,

mentioning that for Xbox to increase its software strength it needs to
[]. 401

7.145 Another internal document produced by the same third party [] also
discusses how the quality and variety of content is a key factor in consumers’
satisfaction with their console. It assesses the factors that contribute to user
satisfaction [] and shows that ‘quality of games’ is the most important factor
contributing to gamers’ satisfaction [], which is closely followed by ‘Variety
of games’. 402

Evidence from documents

7.146 Various internal documents from Microsoft, as well as statements made by its
staff, indicate that the company considers content (especially large game
franchises and exclusive content) to be an important factor affecting both
customers’ choice of console and the attractiveness of Xbox’s subscription
service:

(a) For example, in an interview with the US's Federal Trade Commission
(FTC), Microsoft gaming’s CEO []. In the same interview he added that,
[].

(b) An internal email exchange between members of Microsoft’s PR team

that discusses Microsoft’s strategy for the next generation of consoles
notes that []. The document states that []. 403

(c) One Microsoft internal document highlights []; 404 []. 405

(d) A report prepared by IDG Consulting and submitted by the Parties

confirmed that []. The report also outlined that []. 406

400 [] Internal Document.

401 [] Internal Document.
402 [] Internal Document.
403 Microsoft Internal Document.
404 [].
405 Parties Internal Document.
406 Parties Internal Document.

122
 (e) A Microsoft internal document explains that []. 407

(f) Finally, in reference to the importance of content to Microsoft's future

gaming offerings, one Microsoft internal document explains that []. 408
We note the Parties’ submission that []. 409 However, we consider that
franchises with annual releases of sophisticated graphically-intensive
multiplayer titles (such as CoD) are well placed to offer [] given their
strong brand recognition.

7.147 We also reviewed documents which suggest content, especially AAA day and
date releases, drives the success of MGS services (including on consoles).
These documents suggest that acquiring a big publisher would have a severe
impact on the ability of rival MGS services to succeed:

(a) One email from []. 410

(b) One Microsoft email dated January 2022 contains []. 411

7.148 Internal documents from Activision also emphasise the importance of large
game franchises in driving consumer engagement and spend. For instance,
an Activision strategy document, in discussing the competitive landscape in
publishing, notes that []. The document also noted that developers are
increasingly []. 412

Evidence from the CMA survey

7.149 The CMA survey responses provided some information on the importance of
the content available on a console as a choice factor when purchasing a
console. A total of 89% of respondents to the CMA survey said that the
available games were quite important (32%) or very important (57%) as a
choice factor.

7.150 The other factors that respondents considered quite important or very
important in their choice of purchasing a PlayStation were:

(a) the possibility of playing with friends and/or family members who also own
a PlayStation (87%) and with other PlayStation gamers in general (67%);

407 Microsoft Internal Document.

408 Microsoft Internal Document.
409 Microsoft, response to working papers.
410 Microsoft Internal Document.
411 Microsoft Internal Document.
412 Activision Internal Document.

123
 (b) PlayStation’s superior technical specifications, performance, visual appeal
and features (89%); and

(c) the better value for money of PlayStation’s offer compared to other
consoles (75%).

7.151 We note that the CMA survey was targeted at CoD gamers and therefore may
not be representative of console gamers more broadly—we consider this
evidence in the round along with the other evidence on the importance of
content.

Conclusion on the importance of content

7.152 We consider that the evidence indicates that content is a very important factor
affecting the choice of a console (including its MGS services offering) and
driving consumer engagement. This is supported by evidence from internal
documents from both Parties, third parties’ submissions and evidence from
the CMA survey.

Importance of CoD to SIE

7.153 Having found that content is important for console providers, we now consider
the importance of Activision’s content – and in particular CoD.

7.154 The MAGs explain that, if downstream rivals can easily switch away from the
upstream party to a range of effective alternative suppliers, they will be less
likely to suffer harm than if the merged entity occupies an important position
upstream. 413 The starting point for this assessment will be the structure of the
upstream market. As such, we consider upstream shares of supply in this
section, as well as the alternative content available. The MAGs also explain
that: ‘The merged entity could only harm the competitiveness of its rivals if the
input it supplies plays an important role in shaping downstream competition.
In assessing this the CMA will have regard to all foreclosure mechanisms, so
will consider not only the proportion of rivals’ costs that the input accounts for,
but also for example the role it plays as a determinant of product quality or the
rate of innovation. Its focus will be not on predicting the precise impact of
each possible deterioration on rivals’ businesses, but on the overall question
of whether in aggregate they could be foreclosed.’ 414 We therefore consider in
this section the role Activision’s content, and in particular CoD, plays in

413 CMA129, paragraph 7.14(a).

414 CMA129, paragraph 7.14(b)

124
 shaping downstream competition by looking at the evidence on the
importance of this content to SIE’s PlayStation.

Parties’ views

7.155 Microsoft submitted that the Merged Entity will not have upstream market
power, given Activision’s low shares in terms of various metrics both in overall
console game publishing and within the narrower segment of shooter games
for consoles. 415 Microsoft further submitted that CoD’s popularity varies over
time and does not equate to market power (eg CoD Vanguard saw drops in
purchases and player engagement). 416

7.156 Microsoft submitted that CoD is not an important input to consoles because (i)
gaming platforms like Nintendo and Steam have prospered without access to
CoD; (ii) CoD did not drive platform adoption; (iii) a majority of Xbox users did
not play CoD; (iv) for the vast majority of gamers, CoD was a small
component of their overall gaming consumption; 417 and (v) gamers play
multiple games and were not likely to be influenced to switch by a single
game. 418

7.157 Microsoft further submitted that SIE had overstated the importance of CoD to
its viability because (i) CoD represented a small share of PlayStation’s digital
sales and total MAUs worldwide in 2021; 419 and (ii) CoD’s significance to SIE
was fuelled by SIE’s marketing efforts and that it would be misleading to
compare CoD with games that SIE does not market. 420

7.158 The Parties also commissioned YouGov to carry out an online survey of
console gamers in the UK from the agency’s online panel. A total of []
gamers responded, of whom [] both used a PlayStation as their main
gaming device and identified CoD as their favourite or second-favourite
game. 421 Microsoft submitted that the aim of its survey was expanding the
scope of the CMA’s survey and assess the impact of any game on gamers’
choice of console. 422 Microsoft submitted that the YouGov survey found that:

415 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 1.5. See also Microsoft

response to the CMA supplementary evidence paper.

416 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.27.
417 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 1.7.
418 Parties, submission to the CMA.
419 Microsoft response to the Phase 1 Decision, 11 October 2022, paragraph 3.17.
420 Microsoft response to the Phase 1 Decision, 11 October 2022, paragraph 3.18.
421 Microsoft, Annex to response to the Provisional Findings, 2 March 2023, paragraph 2.2 and supporting Excel

replication file.
422 Microsoft, Annex to response to the Provisional Findings.

125
 (a) The most popular game on PlayStation is generally [] and that CoD’s
popularity is similar to that found for other games franchises. 423

(b) Estimated levels of diversion are low across all titles, thereby showing that
no single game drives console choice. 424

(c) Xbox gamers systematically reported a higher inclination to switch away

to the other console than PlayStation gamers as a result of their preferred
games becoming unavailable on Xbox, thereby showing that PlayStation
gamers have greater loyalty to their console than Xbox gamers. 425

Our assessment

7.159 In order to assess the importance of CoD to PlayStation, we have assessed

evidence on the following aspects:

(a) First, we have considered evidence on shares of supply, including (i)

upstream game publishing shares of supply across consoles and on
PlayStation; and (ii) shares of supply within the narrower segment of
shooter games on PlayStation.

(b) We then consider other evidence on the importance of CoD to

PlayStation, including (i) the level of engagement and spend by CoD
gamers on PlayStation; (ii) evidence on CoD’s role in console adoption,
(iii) the success and longevity of the CoD franchise; (iv) evidence from
revenue sharing agreements of consoles with various publishers including
Activision; and (v) evidence from the Parties’ valuation model. As part of
our analysis, we look at a range of evidence, including the Parties’ and
third-party views and data, internal documents, and evidence from the
Parties’ and the CMA’s surveys.

Shares of supply

• Introduction

7.160 The share of gamers’ time and money spent on playing CoD is an indicator of
the relative value they derive from playing CoD as compared to other games.
In that sense, the greater CoD’s share of supply within console game

423 Microsoft, Annex to response to the Provisional Findings.

424 Microsoft, Annex to response to the Provisional Findings.
425 Microsoft, Annex to response to the Provisional Findings.

126
 publishing, the more significant CoD is likely to be in influencing gamers’
perception of how good the range of games on PlayStation would be.

7.161 Given the various cuts of shares of supply available, we first assess which
metrics are the most informative of Activision’s upstream market power in
relation to CoD and the importance of CoD to PlayStation:

(a) First, we consider Activision’s share of supply in the overall upstream

console game publishing market.

(b) Second, we consider whether Activision’s shares of supply in publishing

on PlayStation are more relevant to our assessment of the importance of
CoD to PlayStation than the shares in overall game publishing on all
consoles. As noted above, shares of consumers’ time and money spent
on playing CoD on PlayStation would help us understand what proportion
of SIE’s range is currently accounted for by CoD, and therefore the impact
on its competitive offering if it were no longer available or available on
worse terms. While shares across all consoles are indicative of
Activision’s upstream market power in relation to CoD within the console
publishing market, they may understate the importance of CoD to
PlayStation in particular because they will include (i) games that are
popular to gamers that play on other consoles such as Nintendo which, as
set out above, has a differentiated offer and is not as close a competitor to
Xbox and PlayStation, and (ii) first-party content held by Microsoft and
Nintendo that may not be available to PlayStation.

(c) Third, we consider whether it is appropriate to assess shares of supply

within genres (the shooter genre for CoD) or across all genres. As
evidenced above, there is a certain degree of differentiation between
genres in terms of their product offering, with games within a genre being
considered most substitutable than across genres. The evidence also
indicates that, whilst some users play games across genres, others are
likely to prefer shooter games (and therefore may not find other games
substitutable for that proportion of money or time spent). This means that
shares of supply across all game genres might understate the importance
of games that are within a genre (and that matters to a greater extent if
the genre itself is important, which is true for shooters as evidenced
above). 426 Therefore, whilst we consider the shares of supply across
genres, we also look at the shares within the segment of shooter games

426On the other hand, shares of supply within a genre overstate the importance of a game, given material
substitutability across genres.

127
 to understand the extent to which shares across all genres may be an
underestimate of the true importance of CoD to PlayStation.

(d) Fourth, we consider the relevance of shares of supply on a global and UK

basis. As discussed in Chapter 5, we consider the geographic market for
console game publishing to be global, whereas we consider the
geographic market for console gaming services to be the UK (while taking
account of broader evidence where relevant). In assessing whether the
Merged Entity has the ability to foreclose, given our focus on whether
PlayStation could be foreclosed (weakened) as a competitor in the UK,
we consider UK shares to be more relevant. 427 In addition, global figures
may include sales patterns for regions where CoD may be less important.
As such, we discuss UK shares in the below assessment, but we also
consider global shares where they are significantly different.

(e) Finally, we have not considered shares of supply in isolation. In a highly

differentiated market such as console game publishing, some games may
be more or less important than their shares of supply suggest. We
therefore consider other evidence on the importance of CoD before
concluding on the importance of CoD to SIE.

• Shares of supply by consumer spend and gametime

7.162 In this section, we consider the shares of supply by consumer spend and
gameplay time constituted by CoD.

7.163 Share of supply of games on all consoles: First, we consider Activision’s

share of supply in overall upstream console game publishing. The Parties’
submissions and data from third party market testing (for revenues only) show
that Activision’s share in console game publishing in the UK is []% by
revenue, []% by digital downloads and []% by gameplay time. 428 The
other publishers and their shares (by revenue in console game publishing in
the UK) were Electronic Arts ([]%), Nintendo ([]%), Take-Two ([]%),
Ubisoft ([]%), SIE ([]%) and Epic Games ([]%)—we discuss them
further in the section on available alternatives to CoD below.

7.164 However, as discussed above, we are of the view that shares of supply
across all consoles are likely to underestimate Activision’s strength in

427 On the other hand, when we assess the Merged Entity’s incentives to foreclose, our assessment is mainly at a
global level as we recognise that decisions on game exclusivity etc are often made globally.
428 Parties FMN. Globally, Activision’s shares in console game publishing are []% by revenue, []% by digital

downloads and []% by gameplay time.

128
 publishing and its importance to PlayStation. We therefore consider other
relevant metrics of shares of supply below.

7.165 Share of supply within all games to PlayStation: Second, we consider the
share of supply accounted for by Activision content, particularly CoD, as a
proportion of all games currently available on SIE’s PlayStation platform,
without differentiating between genres.

7.166 Activision games, and CoD in particular, constituted a significant share of

PlayStation consumer spend and gameplay time. Data from [] 429 showed
that:

(a) Activision had a share of []% by consumer spend in 2021 globally and
in the UK (with Microsoft representing a further []%). 430 In particular,
CoD constituted []% of PlayStation consumer spend in the UK and
globally. 431

(b) Activision constituted []% of gameplay time on PlayStation, with CoD

alone accounting for []% in the UK. Globally, CoD’s share was []%.

7.167 Additionally, we note that both consumer spend and gameplay time were
concentrated on a few large game franchises including CoD on PlayStation,
with a long tail of other games with lower shares of spend/gameplay time.
Data from [] showed that, in the UK, [] of PlayStation’s consumer spend
and [] of PlayStation’s gameplay time was on three franchises in 2021,
including CoD. There were only two other titles—[] and []—at similar
levels of spend and gameplay time as CoD on PlayStation. 432 We discuss this
further in the section on available alternatives to CoD below.

7.168 We note that the above shares of spend and gameplay time are consistent
with the Parties’ submission that CoD represented []% of SIE’s digital sales
worldwide in 2021. 433

7.169 Share of supply of shooter games to PlayStation: Shares of supply within

the shooter genre indicate that CoD is particularly important to PlayStation in
this genre. Data from [] showed that CoD accounted for a large share of
PlayStation’s revenue from shooter games in the UK, constituting []% of
total spend and []% of gameplay time on PlayStation’s shooter games in

429 [] Internal Document; and [] Internal Document.

430 Consumer spend includes digital spend and in-game purchases on B2P games. While we exclude physical
spend, including it does not make a material difference to the shares of CoD on PlayStation.
431 All Activision games constituted []% of PlayStation spend in the UK and globally, with Microsoft

representing a further []%.

432 [] Internal Document; and [] Internal Document.
433 Microsoft response to the Phase 1 Decision, 11 October 2022, paragraph 3.17.

129
 2021. 434 CoD similarly was important to PlayStation in shooter games
globally, accounting for []% of consumer spend ([]% if we only
considered first-person shooters) and []% of gameplay time ([]% for first-
person shooters) in 2021. 435

7.170 The only other shooter game that was similar to CoD in terms of consumer
spend shares of supply was [] ([]% in the UK), with the next highest
share being that of [] ([]%). 436 As above, we discuss alternative shooters
further below.

• Size and engagement of CoD gamers on PlayStation

7.171 In this section, we assess the number of PlayStation gamers who play CoD,
and the extent of their engagement with the game.

7.172 Microsoft submitted that [] of Xbox gamers did not play CoD in a specific
year. According to Microsoft, amongst those gamers that do play CoD, the
majority only do so for a short period of time. 437 In particular, among gamers
that played at least one hour of CoD on Xbox in 2021, more than []%
played it for less than []% of their total gaming time and []% of CoD
gamers spent less than []% of their gaming time on the game. 438 The
Parties also submitted that gamers play multiple games. More than []% of
gamers on Xbox played at least three games during 2021, with only []% of
total game time accounted for by gamers that played two or fewer games
throughout the year. 439

7.173 [] data shows that CoD has a large and engaged user base on PlayStation
globally:

(a) On average across each month of 2021, []% of PlayStation’s MAUs on

PS4 or PS5 with gameplay played a CoD game in that month. 440

434 [] Internal Document.

435 [] Internal Document.
436 [] Internal Document.
437 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 1.7(c).
438 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 1.7(f).
439 Parties response to RFI. [].
440 [], Annex to response to the CMA’s questions; PlayStation’s MAU on a given month are defined as the de-

duplicated number of PlayStation accounts on the PS4 or PS5 that had any gameplay in that month. CoD’s MAU
on a given month are defined as the de-duplicated number of accounts that played any CoD game on the PS4 or
PS5 in that month.

130
 (b) []% of PlayStation’s Yearly Active Users (YAU) on PS4 or PS5 with
gameplay played a CoD game in 2021. 441 In terms of the number of
PlayStation devices, []% of devices played a CoD game in 2021. 442

(c) In terms of engagement, out of the PlayStation devices that played a CoD
game in 2021, []% played it for all their time, []% played it for more
than 80% of their time, and []% played CoD for a significant amount of
time (>20%). 443

7.174 Microsoft submitted that the figures based on YAU are not informative, as
they are estimated without applying any form of minimum threshold to capture
engagement. 444 Microsoft also submitted that MAU figures are more
informative. 445 It submitted that the Parties’ own estimates of CoD MAU on
PlayStation in 2021 suggests a []% penetration at MAU level. 446

7.175 We acknowledge that the YAU figures may overestimate the level of CoD
engagement and that MAU figures may be more relevant. Nonetheless, we
consider that the CoD MAU figures we estimated and presented above show
CoD captures significant engagement among PlayStation users globally.

7.176 We also consider that the Parties’ own estimates of CoD MAUs on
PlayStation are not accurate and underestimate the importance of CoD on
PlayStation. 447 Even by using Activision’s estimates of CoD MAU on
PlayStation instead of []’s, we obtain that, on average across each month of
2021, []% of PlayStation’s MAU on PS4 or PS5 with gameplay played a
CoD game in that month; 448 we still consider this to be a significant number.

441 [], Annex to response to the CMA’s questions; PlayStation’s YAU are defined as the de-duplicated number
of PlayStation accounts on the PS4 or PS5 that had any gameplay in 2021. CoD’s YAU are defined as the de-
duplicated number of accounts that played any CoD game on the PS4 or PS5 in 2021. The percentage of
PlayStation’s YAU that played CoD across all devices (including PS3 and those with no gameplay and only
multimedia usage) is []%.
442 [], submission to the CMA.
443 [], submission to the CMA.
444 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.40(c).
445 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.40(c).
446 Microsoft response to the CMA supplementary evidence paper,12 April 2023, page 34-35; the Parties

estimated the []% figure by dividing the number of CoD MAU on PlayStation as estimated by Activision ([])
by their estimate of PlayStation MAU for 2021 as based on Sony’s Supplemental Information to its financial report
([]).
447 One of the reasons for the discrepancy between the Parties’ figures and ours is that the Parties’ estimate

appears to be the number of CoD MAU on PlayStation as a proportion of all PlayStation MAU, thereby including
PlayStation users who do not have any gameplay but performed activities like []. ([], Annex to response to
the CMA’s questions). On the other hand, our estimate excludes those activities and focuses only on those
PlayStation users that had any gameplay in 2021 – these are the relevant users to consider when assessing the
importance of a certain game (eg CoD) for PlayStation. Another possible reason why the Parties’ estimate differs
from ours is that Activision and [] might have different methodologies to classify a user as a MAU, as
suggested by Microsoft (Microsoft response to the CMA supplementary evidence paper,12 April 2023, page 34-
35).
448 This figure is obtained by dividing the average CoD MAU in 2021 as estimated by Activision ([]) by the

average PlayStation MAU that had any gameplay in 2021 ([]).

131
7.177 While we acknowledge that the above figures are largely consistent with the
Parties’ submissions that a majority of PlayStation gamers do not play CoD
and some CoD gamers are not highly engaged, we note that the shares of
supply figures that we present above (eg []% by consumer spend, []% by
gametime) already take that into account. Additionally, these shares are
indicative of how much of PlayStation’s ‘range’ of game offering is accounted
for by CoD, and this is relevant to how much scope there is for PlayStation’s
overall range to be affected by CoD. It is not necessary for a majority of
consumers to consume a product in a retailer’s range for that product to
contribute substantially to the overall attractiveness of that range and
therefore to that rival’s competitive offering.

Other evidence on the importance of CoD

7.178 In the previous sections we looked at the share of spend and gameplay time
that CoD accounts for and the size and engagement of the CoD gaming
community. However, as noted above, shares do not capture everything
because (i) there is differentiation even within the segments considered, and
(ii) some games with equal spend or time spent may have a greater or lesser
influence on console choice than others. We therefore consider additional
evidence on the importance of CoD.

• Overall spend and gameplay of CoD gamers on PlayStation

7.179 In this section, we assess whether CoD drives engagement and spend on
other games on PlayStation.

7.180 SIE submitted estimates purporting to show the importance of CoD to their
overall business. In particular, SIE submitted that CoD gamers on PlayStation
generated estimated annual platform spending of around [] billion on
hardware, peripherals, subscriptions, games, and other PlayStation services
and that this represented around []% of total spending on hardware,
peripherals, subscriptions, games, and other PlayStation services. 449

7.181 However, we consider that SIE’s estimate overstates CoD’s importance. The
figure above includes spend of every user who played CoD regardless of the
time spent on it. We consider that it would be more relevant to assess the
importance of CoD by reference to users’ engagement with CoD and the
proportion of overall spend on PlayStation.

449 SIE response to the phase 2 Issues Statement, 28 October 2022, paragraph 12.

132
7.182 As shown in the table below, the evidence provided by [] showed that: 450

(a) Gamers that spent more than 80% of their time on CoD accounted for
[]% of total platform spend (including []% of all B2P game spend,
[]% of hardware spend and []% of subscription spend) and []% of
total gameplay time on PlayStation;

(b) Gamers that spent more than 50% of their time on CoD accounted for
[]% of total platform spend (including []% of all B2P game spend,
[]% of hardware spend and []% of subscription spend) and []% of
total gameplay time on PlayStation; and

(c) Gamers that spent more than 20% of their time on CoD accounted for
[]% of total platform spend (including []% of all B2P game spend,
[]% of hardware spend and []% of subscription spend) and []% of
total gameplay time on PlayStation.

7.183 Therefore, we consider the evidence above shows that customers that spent
a significant amount of time playing CoD accounted for a substantial
proportion of revenue across PlayStation’s business, including customer
spending on other games, hardware, and subscription on PlayStation. These
customers also accounted for a significant fraction of total gameplay time on
PlayStation across all games.

• Diversion of spend and gameplay time based on survey evidence

7.184 Along with the above evidence of overall spend and gametime by CoD
gamers, evidence from the CMA survey is likely to be indicative of the actual
spend and gametime that PlayStation would lose based on stated diversions
of CoD gamers in the UK.

7.185 Our survey asked CoD gamers on PlayStation to think about the time when
they bought their most recent PlayStation. The survey asked what they would
have done if the most recent CoD game they owned at that time had not been
available on PlayStation, but had been available on Xbox and PC.

7.186 Results indicated that 24% of respondents would have bought an Xbox, a PC,
or no gaming device at all instead of a PlayStation. Adjusting this figure to
account for all PlayStation users rather than the respondents considered in

450 [] Internal document.

133
 the survey, we found that withholding CoD from PlayStation would have made
[]% of UK PlayStation users move away from PlayStation. 451

7.187 Adjusting this diversion of PlayStation gamers to diversion of spend/

gametime, we found that the impact on PlayStation amounts to at least []%
of spend and []% of gametime. This is because we found that in the UK, the
average ‘non-casual’ CoD gamer in 2021 spent more than either the average
gamer that played CoD ‘casually’ on PlayStation or the average gamer that
did not play CoD on PlayStation. 452 This difference is even greater when we
look at gametime. 453

7.188 Microsoft submitted that [] – not the other way around. 454 According to
Microsoft, its Xbox data shows that []. 455 Microsoft submitted that, as a
result, the true value of switchers from the CMA survey should be []. 456

7.189 However, our analysis of data provided by [] on PlayStation devices in the
UK shows that PlayStation devices where CoD is not played at all [], and
devices where CoD is the only game played also []. Average spend is
highest among those who []. 457 This pattern is very similar to []. 458 We
consider the data on PlayStation devices to be more accurate for this
analysis, given we are assessing the value of CoD players to PlayStation.

7.190 We also consider it is not necessarily the case that only the most engaged
CoD gamers on PlayStation would switch. The distribution of switchers in
terms of CoD engagement levels is unknown, but it will include some who
play other games and therefore spend only a proportion of their gametime on
CoD. Therefore, we believe that the best estimate of the amount of spend lost
by PlayStation for each switcher as per the survey is the average spend of all
‘non-casual’ CoD gamers. This is the population of gamers we surveyed. This
best estimate forms the basis to our adjustment to the value of diversion
presented above. 459

451 Microsoft submitted an alternative calculation that indicates that less than []% of PlayStation gamers would
move away (Microsoft response to working papers, response to respondent level survey data). However, this is
based on three adjustments we do not consider appropriate: excluding gamers who play only the free Warzone
game, calculating the percentage of respondents who would buy an Xbox instead (excluding those who would
buy a PC instead or no gaming device) and then re-scaling this percentage using published Monthly Average
User statistics rather than annual figures.
452 We classify gamers who spent more than 10 hours gametime or $100 spend in 2021/22 as ‘non-casual' CoD

gamers.
453 CMA analysis of [] data.
454 Microsoft response to Provisional Findings, 2 March 2023, paragraph 2.53; See also Microsoft response to the

CMA supplementary evidence paper.

455 Microsoft response to Provisional Findings, 2 March 2023, paragraph 2.54.
456 Microsoft response to Provisional Findings, 2 March 2023, paragraph 2.54.
457 [], Annex to response to the CMA’s RFI.
458 Microsoft, Annex to response to the CMA’s RFI.
459 The same reasoning applies to our adjustment of diversion by gametime.

134
7.191 In the YouGov survey submitted by the Parties and described above, gamers
were asked how they would have responded if the latest CoD game had not
been available for PlayStation but had been available for Xbox and PC. The
Parties used this data to estimate that only []% of all UK PlayStation
gamers would switch to Xbox in response to total foreclosure of CoD. 460

7.192 We gave this less evidential weight than the estimates derived from the CMA
survey, for the following reasons:

(a) The relevant statistic for assessing ability to foreclose is the proportion of
gamers that would leave PlayStation, not just the proportion that would
switch to Xbox. We are trying to capture the impact on PlayStation’s
competitive offering and all switching away from PlayStation is indicative
of the materiality of taking CoD away.

(b) Our survey good practice guide notes that some customer sources that
are used in commercial research are generally not considered sufficiently
robust by the CMA for merger cases. In particular, it advises against
recruiting customers from panels with non-random samples, because
such panellists may have systematically different attitudes and behaviours
to other customers. 461 The YouGov panel used for the Parties’ survey
does not recruit its panellists in a way that meets the standards set out in
the good practice guide.

(c) [] PlayStation gamers were asked about foreclosure of CoD in the
YouGov survey, compared with 1,397 in the CMA survey.

7.193 There are reasons why our estimate of []% of spend lost to PlayStation
derived from the CMA survey may be an underestimate of the real impact. For
example:

(a) Some of the CoD gamers that decided to stay on PlayStation will likely
reduce their spend and gametime on PlayStation generally in the absence
of CoD. We note that not all gamers will do so but, to the extent that
gamers reduce their spend even by a fraction of their average CoD
spend, there would still be a significant impact on PlayStation’s spend and
gametime.

(b) We are assuming all gamers who are part of the wider PlayStation user
base that were ineligible for, and hence excluded from, our survey sample
(because they do not play CoD or play it ‘casually’) do not move away

460Microsoft, Annex to response to the Provisional Findings, 2 March 2023.

461Good practice in the design and presentation of customer survey evidence in merger cases, May 2018,
paragraph 2.27b.

135
 from PlayStation at all. In fact, some gamers might decide to move away
from PlayStation because, for example, their friends who play CoD
switched reaction to a foreclosure strategy. This will increase the overall
impact of foreclosure on PlayStation.

7.194 On the other hand, multi-homing may decrease the impact that foreclosure
would have on PlayStation. Our survey indicates that, of those who owned
more than one gaming device, 62% spent almost all of their game time on
their main device. For consoles, this multi-homing can be within the same
generation or with consoles across different generations. These gamers are
likely to maintain some of their game time on PlayStation, at least in the short
term. However, we consider it is likely that gamers will progressively switch
more of their game time onto their new device, particularly for gamers that
multi-home across devices of different generations, eg an old PlayStation and
new Xbox.

7.195 Microsoft submitted that the assumption that any gamer who played CoD in a
year, irrespective of the duration, would reduce their engagement level across
PlayStation more broadly to the average gameplay time of gamers who never
played CoD is unjustified and speculative. 462

7.196 As explained above, we acknowledge that not all gamers who remain on
PlayStation would reduce their spend and gametime on PlayStation in the
absence of CoD. However, we think it is reasonable to assume that there
would be at least some further reduction in gametime and spend. We have
found that CoD is a popular game with limited substitutes, so it is likely that
some gamers who stay on PlayStation would not switch to alternative games
in place of CoD. Therefore, the engagement of these gamers on PlayStation
could decline, leading to some reduction in gametime and spend on
PlayStation.

7.197 Results from the CMA survey in relation to partial foreclosure are similar,
albeit slightly lower than the diversion in the total foreclosure scenario
described above. Results indicated 17% of our survey respondents, ie []%
of UK PlayStation users would move away from PlayStation in the event of a
specific type of partial foreclosure strategy (ie content exclusivity). We have
reservations about the results of our survey in relation to partial foreclosure,
and discuss these below when we discuss our modelling of the Parties’
incentive to foreclose.

462 Microsoft response to Provisional Findings, 2 March 2023, paragraph 2.58.

136
7.198 We therefore consider that the above evidence shows that PlayStation would
likely lose a significant share of its spend and gametime based on CoD
gamers switching away from it.

• CoD’s role in console adoption

7.199 Microsoft submitted that CoD does not drive platform adoption. It submitted
that between 2016 and 2022 only []% of gamers played CoD as their first
game on their new Xbox console, and []% of new Xbox gamers never
played or purchased Call of Duty content on Xbox. The Parties stated that []
was the most common first-played game, with [], [] and numerous other
titles being played first. 463

7.200 Evidence from third party internal documents suggests that CoD is a major
driver of console sales.

(a) For example, one [] document prepared in [] stated that over []. 464
Based on third party data, we note that this constitutes at least []% of
PS4’s installed console base. 465 The document also stated that CoD titles
represented [], and that [].

(b) Another [] internal document prepared in [] shows that, based on
[], CoD is the second most important game in the UK for driving
purchase interest []. 466 The same internal document finds that for more
engaged consumers, the importance of CoD is even bigger. []. 467
[]. 468

7.201 We considered additional evidence to understand how relevant CoD is in the

adoption of PlayStation consoles:

(a) First, we considered the proportion of PlayStation gamers globally who

played CoD on the first day of buying the console. Data from [] showed
that []% of PlayStation gamers played CoD on the first day of gameplay
on their new console in 2021, followed by [] ([]%), [] ([]%), []
([]%) and [] ([]%). 469 CoD was also the most played franchise on

463 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.25a.
464 [], submission to the CMA.
465 IDG data submitted by the Parties indicated an installed console base of [] million PlayStation 4 in the US.

Whilst the time period over which the [] figure is measured over is unclear, considering the entire PS4 console
base would be a lower bound of the share of console sales driven by CoD.
466 [] Internal Document.
467 [] Internal Document.
468 [] Internal Document.
469 [] submission to the CMA.

137
 the first day of gameplay in 2020 globally, and ranked 3rd in 2018 and
2019 and 2nd in 2017. 470

(b) Second, we considered the proportion of PlayStation consoles sold in a

year globally that were used to play CoD in the first month following their
purchase. Data from [] showed that []% of PlayStation consoles sold
in 2021 were used to play CoD in the first four weeks of gameplay,
followed by [] ([]%), [] ([]%) and [] ([]%). CoD was also the
most played game in the first four weeks following console purchase in
2020, ranked third in 2018 and 2019 and second in 2017. 471

7.202 The above statistics show that a significant proportion of PlayStation gamers
play CoD as one of their first games on buying a PlayStation, indicating that
CoD is a key driver of platform adoption.

7.203 Evidence from the CMA survey also showed that the availability of CoD was
an important factor in choice of console for survey respondents.

7.204 Among the respondents who stated that the availability of content on a
console is important (89% of our sample), 73% of them said that the
availability of CoD influenced them towards buying a PlayStation.

7.205 At an aggregate level, this was similar to those that were influenced by the
availability of one or more PlayStation exclusive games (69%). However,
individual PlayStation exclusives influenced smaller fractions of
respondents—in particular, respondents were influenced by God of War
(41%), Last of Us (39%), Marvel’s Spider-man (37%) and Ghost of Tsushima
(25.5%).

7.206 A significantly smaller share of respondents picked other games as being

important in their choice of consoles: these games included GTA (45%),
Assassin’s Creed (23%), Fortnite (17%), Star Wars (15%) and Other (15%).

7.207 We note that evidence from the CMA survey in relation to the availability of
games in driving console choice is likely to overstate the importance of CoD
given the CMA survey was targeted at CoD gamers—we therefore place less
weight on this piece of evidence but consider it in the round along with the
other evidence on the importance of CoD.

470 [] Internal Document.

471 [] Internal Document.

138
7.208 Microsoft submitted that the success of platforms that do not offer CoD, such
as Nintendo and Steam (on PC), show that CoD is not important to the
success and adoption of a platform. 472 However, we note that:

(a) Nintendo is not a close competitor to Xbox or PlayStation (as set out
above).

(b) Steam is not a gaming console, and the competitive dynamics of PC are
different to consoles as noted in Chapter 5. PC gamers can access a
range of content from different PC stores, including Steam and others,
which they can then play on the same PC. This is different to the market
for consoles, where most customers typically select one console, and that
choice is driven in large part by the content available on the storefront for
that console.

7.209 Microsoft also submitted that making CoD exclusive on Battle.net on PC since
2018 []. 473 We however note the following points in relation to the above:

(a) The comparison is not analogous because the alternative available to

CoD PC gamers when it was withdrawn from Steam was Battle.net, which
only offered Activision games. On the other hand, in the event of a
withholding from or material degradation of CoD on PlayStation, CoD
gamers on console would have the choice of Xbox, which offers a
significantly larger catalogue of games relative to Battle.net. Therefore, it
is likely to induce greater switching relative to Battle.net.

(b) Notwithstanding the above, making CoD exclusive on Battle.net allowed

Activision to compete with Steam by offering fewer than 10 games, whilst
Steam offered c.8000 games in 2018. 474 The fact [] shows the strength
of the Activision portfolio, particularly CoD. 475

• Success and longevity of the CoD franchise

7.210 In this section, we consider additional evidence on the overall performance of

CoD over time.

7.211 Evidence shows that CoD has been one of a small number of large and
consistently successful game franchises available to gamers for many years.
As can be seen from the figure below, CoD has had a new release every year

472 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.40(a) and (b).
473 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.40(b).
474 statista.com, accessed by the CMA on 19 December 2022.
475 Microsoft response to the phase 2 Issues Statement, 31 October 2022, Figure 31.

139
 since 2012, with each title selling at least [] units worldwide (except CoD
Vanguard, which sold [] units). 476

Figure 7.1: []

[]
Source: Activision site visit.

7.212 Microsoft submitted that CoD’s popularity varies over time and success in the
previous release does not guarantee the success of future titles (eg CoD
Vanguard saw drops in purchases and player engagement). 477 We agree that
there was a drop in the total number of monthly active users across all CoD
titles after the Vanguard release turned out not to be as successful (no
change in the first 3 months after the release of CoD Vanguard, however an
overall decrease of []% in the first ten months of 2022 relative to the first ten
months of 2021). 478 However, given the huge success of the latest release,
CoD Modern Warfare II in November 2022, we infer that the fall in MAUs at
the franchise level was temporary. For instance, Activision stated publicly that
Modern Warfare II ‘has become the #1 top-selling opening weekend ever in
the franchise, delivering more than $800 million worldwide in sell-through
following the first three days from its release on October 28, 2022’. 479

7.213 In relation to PlayStation in particular, CoD has consistently performed well

over the past few years. For example, data from Activision shows that [] in
2020 and 2021 compared to prior years due to the success of Warzone as a
free-to-play title. 480 CoD titles, released annually, []. For example, CoD
World War 2 2017 made []% of its total revenue to date in the [] months
post release, with the figures for Black Ops 4 2018 ([]%) and Cold War
2020 ([]%) being similar. 481 Data also shows that add-on sales are [] part
of CoD sales in recent years, indicating that the game is able to monetise in
different ways. 482

7.214 We further consider evidence from the Parties’ internal documents, third party
views and industry reports on the importance of the CoD franchise below.

7.215 Internal documents: Several internal documents discuss the relative size
and importance of the CoD franchise overall to a platform, including its
consistent and long-lasting success:

476 Prior to 2012, CoD had a release at least once every two years, starting in 2003.
477 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.27.
478 Microsoft Internal Document.
479 ‘investor.activision.com’, accessed by the CMA on 14 December 2022.
480 Activision response to the CMA’s s109 notice.
481 Microsoft response to TOH 1 working papers.
482 Activision response to the CMA’s s109 notice.

140
 (a) A 2019 letter to Activision shareholders noted that the CoD franchise has
been the number one console franchise globally for 10 of the last 11 years
based on third party data and Activision’s estimates of upfront console
sales. 483 Similarly, a 2021 letter to Activision shareholders refers to CoD
as ‘one of the most successful entertainment franchises of all time’, but
also notes that CoD Vanguard ‘didn’t meet our expectations’. 484 We also
note the Parties’ submission that the estimates in the 2019 letter did not
include free-to-play games or post-sale monetisation, many of which are
also popular. 485

(b) One Activision internal document from 2020 shows that Activision had
[] franchises generating revenues greater than USD [] based on
owned IP across platforms and demos (including mobile and PC). Of
these, CoD was the []. The document states that there is ‘[] for
Activision content across new distribution outlets and devices’. 486

(c) One third party consultancy report submitted by the Parties prepared in
2020 show that CoD: Modern Warfare ranked second in ‘Console Top
Titles’ in the UK by physical and digital unit sales in 2019 and 2020. 487 We
note the Parties’ submission that this ranking did not include free-to-play
games or in-game revenues; and included other successful games such
as FIFA, one SIE exclusive and two Nintendo exclusive games. 488

7.216 Whilst we note that the above documents do not provide a comprehensive list
of the importance of every game by every metric, they still point to CoD being
consistently amongst the top franchises on consoles.

7.217 Internal documents also highlight the factors that make a large franchise like
CoD successful, durable, and, therefore, important to a platform. These
include the high level of brand awareness and the loyal user base it
commands. For example, one Activision document prepared in 2020 noted
that the top-grossing game charts are dominated by established franchises,
with multiple factors driving the durability of game franchises. These factors
included [] The document noted that CoD had []; and stated that
Activision’s strengths lie in its []; and []. Other reasons highlighted for
Activision Blizzard’s success were []. 489

483 Activision Blizzard 2019 Annual report, page 5, accessed by the CMA on 17 January 2023.
484 Activision Blizzard 2021 Annual report, page 2, accessed by the CMA on 17 January 2023.
485 Microsoft response to working papers.
486 Activision Internal Document.
487 Microsoft Internal Document; and Parties Internal Document.
488 Microsoft, response to working papers.
489 Activision Internal Document.

141
7.218 In relation to the above document, Activision submitted that, even if big
gaming franchises taken together are important to a console when looked at
in aggregate, an individual franchise may not be as important by itself. 490 We
agree with Activision that it is important for platforms to have access to a suite
of franchises (which both Xbox and PlayStation currently do), and we also
consider that taking away one of those major franchises from a console is
likely to impact its ability to compete as effectively as it would absent such
foreclosure. There are very few franchises with all or most of the features
listed above.

7.219 Internal documents also discuss the importance of CoD to PlayStation in

particular, including the revenue and sales it has brought to PlayStation and
the impact on sales of PlayStation hardware:

(a) One Activision internal document shows that in 2020 the CoD franchise
[] on PlayStation in multiple categories, including full game sales and
standalone add-on revenue. The document also shows that CoD remains
[], with [] games in the Top 10 best-selling games list. 491

(b) Another Activision internal document states that Activision’s content []
and would continue to do so. The document explains that a [] CoD
partnership between SIE and Activision [] since the launch of
PlayStation 4. The document concludes that Activision [] in the next
generation. 492

7.220 Activision submitted that both of the above documents were prepared [],
and that the information presented in these documents are [] and presented
in order to []. 493 Whilst we acknowledge that Activision may have [], we
believe that these documents are nonetheless useful context around the
success of CoD on PlayStation. In particular, we believe [].

7.221 Third party views: Third parties also noted that Activision games, including
CoD, are strong franchises and important to competition in high-end gaming
(whether in console, PC, or cloud gaming services):

(a) One third party [] stated that Activision games are ‘critical to
competition in high-end gaming’; are ‘must have’ games for many
consumers of gaming platforms and have no meaningful substitute; and
that CoD games help drive sales of the gaming platform. 494

490 Activision response to TOH 1 Ability working paper.

491 Activision Internal Document.
492 Parties Internal Document.
493 Activision, response to TOH 1 Ability working paper.
494 [] response to the CMA’s RFI.

142
 (b) Another third party [] stated that successful titles [like CoD] attract
customers to their platform and impact on smaller titles; and that it was
noticeable when new releases of Call of Duty were not available on their
platform. 495

(c) One gaming platform supplier [] submitted that CoD has the highest
awareness and ownership of all third party franchises on its platform. It
argued that having access to the CoD franchise is likely to be a priority for
a larger number of players. 496

7.222 Third party industry reports: Third party analyst and industry reports further
point to the strength and endurance of the CoD franchise:

(a) An analyst report prepared by a third party media and telecommunication

research company in 2021 noted that ‘the Call of Duty franchise could
hardly be stronger, with free-to-play (F2P) entry Call of Duty: Warzone
driving a structural step-up of in-game monetization’. The document also
noted that Call of Duty: Mobile, one of the biggest mobile launches in the
West, should continue to grow. 497

(b) Another public article stated that CoD had the most passionate fan base
amongst top gaming brands in 2019. It explained that CoD’s significance
to entertainment could not be overstated, with it being the only game in
the top 10 of all entertainment brands amongst the most avid gamers. 498

(c) Another external industry report submitted by a third party [] discussed
a research consumer survey dated September 2021, covering a set of
countries (UK, US, Canada, Australia, Germany, France, Sweden, Brazil,
South Korea). The survey found that in Q3 2021, CoD was the most
played game, played by 17% of the sampled consumers. The survey
found that among the top 15 most widely played games in those countries
in Q3 2021, CoD was played by the highest percentage of sampled
consumers (17%). 499

• Evidence from revenue-share agreements

7.223 We have seen evidence that Activision has been able to negotiate [] of B2P
sales on both Xbox and PlayStation storefronts as compared to []:

495 [] call note.

496 [] response to the CMA’s RFI.
497 Activision Internal Document.
498 ‘Report: Call of Duty has the most passionate fan base among 2019’s top gaming brands’, accessed by the

CMA on 3 August 2022.

499 [], Annex to response to the CMA’s RFI.

143
 (a) Activision keeps [] and []% for other third party games on average
(including []). 500

(b) Similarly, Activision has negotiated a revenue share of []% for CoD on
PlayStation, 501 []% for other third party games on average. 502 This
suggests that Xbox and PlayStation are willing to make a more attractive
offer to CoD in order to attract it to their platform.

7.224 Microsoft submitted that publishers’ contractual terms with Xbox include [].
Based on this, Microsoft submitted that Activision’s [] is []%. In
comparison, [] of []. 503 Whilst we acknowledge that an assessment of
bargaining power should consider all aspects of the contractual relationship
with publishers, we note that the above evidence still indicates that Activision
[].

7.225 The greater bargaining power of Activision in relation to CoD was also
confirmed in testimonial evidence by Microsoft’s executives. The Corporate
Vice President of Xbox, stated in an FTC interview that [], and stated that
[]. 504 []. In the same FTC interview, [] also stated that [].’ 505 [] also
stated that []. 506

• Evidence from the valuation model

7.226 In addition to valuing Activision’s existing business at $[], the valuation

model used by the Parties in relation to the Transaction places []. 507

7.227 First, the valuation model includes []. 508 []. 509 We also note that this []
implying that to the extent that adding CoD makes a difference even in the
context of other games being available, taking CoD away from a rival is also
likely to have a substantial impact.

7.228 Microsoft submitted that (i) []; (ii) []; and (iii) []. 510

7.229 In relation to the above, we note that []. 511

500 Parties response to the CMA’s RFI.

501 [] response to the CMA’s RFI.
502 [], submission to the CMA.
503 Microsoft, response to TOH 1 working papers.
504 FTC transcript.
505 FTC transcript.
506 FTC transcript.
507 Parties Internal Document.
508 Specifically, on Game Pass, the model [].
509 Parties Internal Document.
510 Microsoft, response to TOH 1 working papers.
511 The Parties submitted that []. See Parties Internal Document.

144
7.230 Second, []:

(a) [];

(b) []; and

(c) []. 512

7.231 Whilst we note Microsoft’s submission that the increase in ASP is an industry
factor not specific to Activision, 513 we nevertheless consider that
[]demonstrate its continued importance to gaming platforms, including
consoles.

Conclusion on the importance of CoD

7.232 Based on the above evidence, our view is that CoD is strong in the upstream
market for console game publishing and is an important input to PlayStation. It
is one of the three largest franchises on PlayStation. It contributed a
significant share of PlayStation’s consumer spend and gameplay time in
2021.

7.233 There are additional factors that make CoD an important input to PlayStation:

(a) CoD is particularly important to PlayStation within shooter games, an

important segment of the market.

(b) CoD is important in driving console sales and a significant proportion of

PlayStation console sales are associated to sales of CoD. Nearly a
quarter of respondents to the CMA survey stated that they would switch
away from PlayStation if CoD were not available on it – this is equivalent
to []% of PlayStation total users.

(c) Gamers that spend more than a fifth of their gametime on CoD generate a
significant amount of revenue on PlayStation in the form of spend on third
party games. Evidence from our survey indicated that PlayStation could
lose at least []% of its overall consumer spend based on diversion of
CoD gamers.

7.234 There are other features that are unique to CoD (and possibly few other titles,
as discussed further in the section below) that increase this importance:

512 Parties Internal Document.

513 Microsoft, response to TOH 1 working papers.

145
 (a) CoD is one of the largest game franchises available to consoles currently
and has been for a long time. As evidenced above, game franchises are
instrumental in driving demand to a gaming platform like a console.

(b) Data on units sold and revenues show CoD has had consistently
successful annual title releases for at least the past 10 years.

(c) CoD has a high level of awareness and brand loyalty among gamers, and
it is a multiplayer game with a social component which induces network
effects, making it less likely that gamers will unilaterally switch away from
the franchise.

7.235 Overall, we conclude that CoD is an important input in the console gaming
services market.

Alternatives to CoD

7.236 In this section, we assess the alternatives to CoD that are available to
PlayStation, and the implications of these alternatives for the ability of the
Merged Entity to foreclose SIE.

7.237 Before considering evidence on the availability of alternatives to CoD, we

make the following observations relating to how this evidence fits into our
wider assessment:

(a) The loss of CoD would represent a net reduction in SIE’s range. A
foreclosure strategy would move SIE from a scenario where all of its
current games are available to a scenario where one of its most important
games is no longer available.

(b) In that context, the purpose of assessing evidence on the alternative

games and franchises available to SIE is not to examine to what extent
those games could be used to ‘replace’ CoD and maintain a range of
equal quality to the range absent any foreclosure. Rather, the purpose is
twofold:

(i) First, to the extent the range of other games is sufficiently large and
popular, this would reduce the overall significance of CoD to SIE’s
range. The question we are assessing is therefore not whether other
popular games exist, but rather whether the number and popularity of
other games is sufficiently great that the elimination of the CoD
franchise from PlayStation’s range would not represent a significant
reduction in range in relative terms.

146
 (ii) Second, to the extent there is differentiation between games, the
inability of a console supplier to offer more of those differentiated
games will tend to be more significant by contributing to a worse
range and less consumer choice. Conversely, where games are not
significantly differentiated from each other, the removal of any
individual game or franchise would be less likely to make a significant
difference to range or choice.

7.238 We therefore consider below evidence on the alternatives to CoD, as well as

the extent of differentiation between them. This section focuses on the
availability of alternatives that already exist. We consider the scope for entry
and expansion by other alternatives (including SIE’s first-party games) in
response to the Merger in Chapter 9.

Parties’ views

7.239 The Parties submitted that Activision’s shares in the upstream publishing
market indicated that the market was highly fragmented, with gamers playing
the games of many different publishers, and with each of those publishers
having a limited position overall. They submitted that Activision games had a
modest share; that Activision had a share of MAU of []% globally in 2021;
and that other publishers were of equal or comparable scale, including
Electronic Arts (eg FIFA, Apex Legends, Star Wars) at []% globally, Epic
Games (eg Fortnite), with a share of []% globally and Take Two (eg GTA,
Red Dead Redemption) with a share of []% globally. 514 Microsoft submitted
that data from NPD, a market research company, shows that PlayStation
gamers buy games from a large range of publishers to play on their console. It
submitted that, according to this data, [] is the largest supplier of games to
PlayStation, followed by [] and []. [] and [] are listed as the fourth
and fifth largest game publishers respectively. 515 As the Parties noted, this
data does not include F2P games such as Fortnite and CoD Warzone. 516

7.240 The Parties also submitted that there were several other franchises that
gamers play. They stated that whilst Fortnite and CoD account for a
significant proportion of game-time played on Xbox ([]% and []% in 2021,
respectively), gamers also play popular games such as GTA, FIFA, Minecraft,
NBA2K, Tom Clancy, Roblox, Apex Legends and Rocket League, which
account for a significant proportion of the remaining game-time. 517

514 Parties, submission to the CMA.

515 Microsoft response to the Provisional Findings, 2 March 2023, paragraphs 2.33 to 2.36.
516 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.33.
517 Parties, submission to the CMA.

147
7.241 The Parties submitted that Activision’s internal documents also consistently
recognise the strength of other franchises as a competitive constraint on CoD,
reflecting the fact that CoD is not an important input to SIE. Activision’s
documents monitor games in the ‘shooter’ genre, []; along with other games
[]. 518

7.242 The Parties further submitted that SIE had a large portfolio of high-quality
exclusive first-party content that would continue to attract users to its platform.
This included prominent titles such as The Last of Us, Ghost of Tsushima,
God of War, Spider-Man and Demon’s Souls. The Parties said that these
exclusive first-party titles accounted for approximately 17% of consumer
spend on PlayStation over the period 2019-2021. 519

7.243 The Parties also submitted that their YouGov survey shows that [], not
CoD, was the game reporting the highest level of diversion, both in terms of
past console purchases ([]%) and future purchases ([]%). According to
the Parties, the YouGov Survey also confirms that CoD falls squarely in the
middle of the pack of games available on PlayStation that would prompt
diversion, which the Parties submitted in any event is minimal across all
games. 520

7.244 As a preliminary point, as noted above, the availability of other strong

franchises on PlayStation may not be inconsistent with CoD being important
to the range of PlayStation’s offer, particularly because PlayStation gamers
already have access to these other franchises. We have already considered
the limitations of the Parties’ YouGov survey above. We discuss the rest of
the Parties’ arguments in our assessment below.

Our assessment

Third party views and internal documents

7.245 As noted in Chapter 5, we received mixed evidence from third party

publishers on CoD’s closest rivals, indicating that while CoD competed most
closely with other shooter games, it is also constrained by titles from large
franchises across genres such as FIFA, Grand Theft Auto, Assassin’s Creed
etc. We consider the evidence from third parties and their internal documents
on the effectiveness of these competitors in helping PlayStation continue to
compete.

518 Parties submission to the CMA [].

519 Microsoft response to the phase 1 decision, paragraph 3.9.
520 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.69.

148
7.246 Third party views indicate that existing alternatives available to PlayStation,
including other shooters and its first-party content, were not as strong as CoD.
One competitor [] submitted that console gamers are more interested in
purchasing a new CoD title than any other third party franchise. 521 The
competitor also stated that third party developers and/or publishers, who
already have strong incentives to develop an alternative to CoD, have not
managed to do so. The competitor provided the example of Electronic Arts’
Battlefield, stating that despite the apparent similarities between Battlefield
and CoD (eg Battlefield games have been set in both WWII and the modern
era; may contain story-driven campaigns with rich cinematics; and include
large-scale multiplayer modes with similar game types, etc), and despite
developing other successful AAA franchises (such as FIFA, Mass Effect,
Need for Speed, and Star Wars: Battlefront), Electronic Art’s Battlefield had
failed to achieve CoD’s level of user engagement. 522 The competitor
submitted that as of August 2021, more than 400 million CoD games had
been sold, while Battlefield had sold just 88.7 million copies. 523 The
competitor also submitted that [] first-party content was not of the same
scale and importance as CoD. 524

7.247 Third party internal documents however also suggested that titles like [],
[], [] and [] performed similarly to (and sometimes outperformed) CoD
in terms of sales and gameplay time. For example:

(a) One third party internal document suggested that there are alternative
games, such as FIFA, with a similar impact on [] as CoD. 525

(b) Another internal document of the same third party showed that as of
February 2022, only [] attracts more gaming hours than CoD. 526 No
shooter game attracts similar levels of gaming hours to CoD. 527 Other
games that ranked below CoD in terms of hours included [], [] and
[].

(c) Another third party industry report stated that in terms of pure physical
and online sales (with no in-game purchases), [] had the largest sales
in 2021, with CoD games in second and fourth place. CoD is the only
shooter-genre game in the top 10. 528

521 [] submission to the CMA.

522 [] submission to the CMA.
523 [].
524 [] response to the CMA’s RFI.
525 [] Internal Document.
526 [] Internal Document.
527 [] Internal Document.
528 [] Internal Document.

149
 Evidence from the Parties’ internal documents

7.248 As noted above, Activision’s internal documents list and track [] as
competitive constraints to CoD (and Overwatch), before considering other
games. We also discussed in Chapter 5 the evidence as regards closeness of
competition within the shooter segment.

7.249 In terms of the effectiveness of these competitors, the Parties’ documents

show that CoD is one of the top performing games on Xbox, with [], []
being closest to it in terms of revenue and MAU. Other games such as [].
For example:

(a) A Microsoft document analysing sales of third party titles shows that
[]. 529 Multiple documents also showed that []. 530

(b) An Activision document comparing premium and F2P first-person shooter

games shows that []% of revenues of the set of premium first-person
shooter games considered were from CoD in 2020, with Destiny,
Overwatch, PUBG, Rainbow Six and Battlefield each being less than
[]%. In the F2P sector, Fortnite was at []% of revenues of the set of
F2P games considered, with CoD Warzone at []%, and other
competitors like Apex Legends and Valorant were less than []%. 531

7.250 Activision’s internal documents also show that other shooters, such as [],
do not perform as well as CoD in terms of revenues and engagement:

(a) An Activision document assessing the impact of CoD’s expansion on

Steam in 2022 uses the shares of [] on PC as a proxy for CoD,
indicating that these games are considered to be close competitors. 532

(b) Another Activision document comparing the performance of [] and CoD
at the time of their respective season releases shows that CoD had
significantly greater revenues and MAU than []. 533 For example, CoD
(Black Ops Cold War and Warzone) made USD [] million in the first 5
days after the release of its Season 3 (in April 2021) whereas [] made
USD [] million after its Season 8 release (in February 2021). Similarly,
CoD had [] million MAU in the first month while [] had [] million. 534

529 Microsoft Internal document.

530 Microsoft Internal document; and Microsoft Internal document.
531 Activision Internal Document.
532 Activision Internal Document.
533 Multiplayer games have temporary ‘seasons’ wherein the developers can make changes in the game and

introduce new content (such as maps, weapons, etc), and often will cover a few months and/or have special
themes.
534 Activision Internal Document.

150
 Other evidence

7.251 Data from [] showed that the top 100 games on PlayStation by consumer
spend are published by more than 30 publishers globally. However, only four
third party publishers had a share of spend greater than 5% on PlayStation
globally in 2021: [], [], [] and []. In addition, SIE publishes successful
exclusives which collectively account for []% of the total spend on
PlayStation. 535

7.252 We consider this data to be more reliable than the NPD data submitted by the
Parties and described above, as it [] and therefore provides a more
accurate picture of the share of different publishers on PlayStation. In
assessing whether there are any games that are effective alternatives to CoD,
we look at (i) other game franchises across genres available on PlayStation;
(ii) other ‘shooter’ games on PlayStation that are likely to compete with CoD;
and (iii) exclusive games on PlayStation that are likely to draw gamers to the
platform.

7.253 In relation to assessing the extent of differentiation between CoD and these
available alternatives, we note that evidence presented in Chapter 5 shows
that there is a certain differentiation between genres, making shooter games
different to other games such as FIFA, Grand Theft Auto, etc. We therefore do
not reassess the differentiation between CoD and non-shooter games.
However, we consider where relevant the extent to which CoD differs from
other shooters.

• Alternative franchises on PlayStation

7.254 There are two games/franchises that in recent years have performed on the
same scale as CoD in terms of spend and gameplay time on PlayStation.
2021 data from [] indicated that: 536

7.255 In terms of share of consumer spend in the UK, FIFA ([]%) is larger than
CoD ([]%) and Fortnite is of a similar size ([]%).

7.256 In terms of share of gameplay time, FIFA ([]%) is the only game bigger than
CoD ([]%) in 2021 in the UK. Fortnite is the only game bigger than CoD at a
global level ([]%). 537

535 [] Internal Document; and [] Internal Document.

536 [] Internal Document; and [] Internal Document.
537 The shares of consumer spend at a global level were []% for FIFA and []% for Fortnite in 2021. In terms

of gameplay time, the shares of FIFA and Fortnite were []% and []% respectively globally.

151
7.257 Other games/franchises on PlayStation include NBA, Genshin Impact, Apex
Legends, Grand Theft Auto and Assassin’s Creed. However, their consumer
spend and gameplay times are significantly lower than CoD. For example, in
2021, none of these franchises had a share of revenue greater than []% in
the UK; nor a share of gametime greater than []%. 538

7.258 FIFA, Fortnite and CoD also have the highest levels of engagement of gamers
on PlayStation. Account-level data from [] for 2021 showed that: 539

(a) []% of PlayStation gamers played FIFA for more than 40% of their total
gametime, and []% played it for more than 80% of their total gametime;

(b) []% of PlayStation gamers played Fortnite for more than 40% of their
total gametime, and []% played it for more than 80% of their total
gametime.

(c) By comparison, []% of PlayStation gamers played CoD for more than
40% of their total gametime, and []% played it for more than 80% of
their total gametime.

(d) There was a substantial gap with the level of engagement reached by
other games. Apart from GTA (and to a lesser extent Apex Legends and
Minecraft), not more than []% of gamers played any of the other games
available on PlayStation for more than 40% of their total gametime.
Similarly, not more than []% of gamers played any of the other games
for >80% of their total gametime

7.259 Microsoft submitted that CoD is not the most popular franchise; and submitted
data on playtime share and various user rankings (such as on Metacritic,
PlayStation user polls) and game reviews of top video games (such as those
by IGN, Business Insider etc) to evidence that CoD does not feature highly on
these lists, and is consistently outranked by other games. 540 Whilst we have
considered this evidence, we believe that data on actual revenue and
gameplay time on PlayStation provides a better indication of its importance to
SIE and should be given more weight.

7.260 The Parties also submitted that there is a wide range of studios that have
produced a hit title, and that SIE would have the ability to work with these third
party developers. These studios included Epic Games, Electronic Arts, Riot
Games, Konami, TakeTwo Interactive, Ubisoft, etc. 541 The Parties also

538 [] Internal Document; and [] Internal Document.

539 [] Internal Document.
540 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.44.
541 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.43e.

152
 submitted that new titles often reach the top of game-time charts. They stated
that each month, at least [] new titles are released on Xbox, and over the
past five years almost [] of those new titles have risen into the top 25
games, and that []% of those had done so within [] of their release. 542

7.261 We acknowledge that there are a range of studios that have produced hit titles
over the years. However, we consider there are few franchises as enduring
and as significant in terms of PlayStation’s revenue and gameplay time as
CoD. In relation to the new titles that reach the top of game charts, the
Parties’ data above indicates the low chance of success – the data above
shows that []% of titles have made it to the top 25 games over the past five
years.

• Alternative shooters on PlayStation

7.262 Alternative games from the shooter genre are smaller than CoD and/or offer a
differentiated experience. We discuss below three games that could broadly
be considered as shooters and that compete with CoD: Fortnite, Battlefield,
and Apex Legends.

7.263 Fortnite is a free-to-play multiplayer shooter game published by Epic Games,

which contributes significantly to PlayStation’s range of spend and gameplay
time. 543 Data from [] showed that Fortnite represented []% of gameplay
time on PlayStation, and []% in the shooter genre in the UK. This is similar
to CoD, whose gameplay time contribution was []% in the UK (and []%
within shooters). 544 Results from our survey of CoD users on PlayStation
indicate that among those respondents to whom content on a console is
important (89% of our sample), 17% of them would be influenced to buy a
PlayStation by the availability of Fortnite, versus 73% for CoD. However, we
note that this survey was targeted at CoD gamers and therefore will over-
represent console gamers with a preference for CoD. We therefore consider
this evidence in the round along with the other evidence on the importance of
content.

7.264 Moreover, as set out above, to the extent two products are differentiated,
PlayStation’s ability to offer both games will tend to be of value to consumers.

542 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.43f.
543 Fortnite was initially developed as an independent game with a small team and budget and released in 2017;
however, it soon became successful with millions of players playing the game in the first few months, and earning
more than $1 billion in its first year. See for example, How Fortnite became the most successful free-to-play
game ever – The New Economy, accessed by the CMA on 28 December 2022.
544 [] Internal Document.

153
 In this context, we have identified a number of aspects of Fortnite’s gameplay
that are substantially differentiated from CoD. In particular:

(a) Gameplay mode: Fortnite has different modes that are not always first-
person shooter (unlike CoD), and its main mode is battle royale, with CoD
Warzone being the only CoD title with this mode available.

(b) Game features: The overall theme of Fortnite is more casual/cartoon-like,

and has fewer elements of serious combat as in CoD and the additional
build element. The game therefore targets a younger demographic
because of the above—Fortnite’s age rating is 13+, whereas CoD has an
age rating of 18+.

(c) Business model: The game is largely free to play (similar to CoD Warzone
but not other CoD titles), with a single title that gets regularly
updated/expanded.

7.265 Battlefield is a multi-player shooter franchise published by Electronic Arts,

which is significantly behind CoD in terms of user spend and gameplay time.
Data from [] showed that in 2021, Battlefield contributed to less than []%
of overall gameplay time in the UK (CoD at []%) and []% within shooters
(CoD was at []%). 545 We note that though Battlefield releases its titles less
frequently than CoD (normally in 3-year cycles), CoD and Battlefield both
released a title in late 2021—CoD Vanguard and Battlefield 2042 respectively.

7.266 Battlefield however is the closest game to CoD in terms of gameplay features
and settings. For example:

(a) Gameplay mode: Similar to CoD, Battlefield is a first-person shooter

game. In its latest release, Battlefield offers an online multi-player game
mode like CoD.

(b) Game features: Similar to CoD, Battlefield is a first-person shooter whose

context is either the Second World War or a modern world. Also, like CoD,
Battlefield contains story-driven campaigns with rich cinematics.

7.267 Various publishers also listed Battlefield as being the most similar to and the
closest competitor to CoD, as evidenced above.

7.268 Apex Legends is another battle royale shooter game published by Electronic
Arts, and is also smaller than CoD in terms of consumer spend and gameplay
time on PlayStation. [] data showed that Apex Legends was the third
biggest shooter game in the UK after CoD and Fortnite, and contributed to

545 [] Internal Document.

154
 []% of overall gameplay time on the PlayStation in the UK (as compared to
[]% of CoD), and []% within shooters (CoD being at []%). 546

7.269 In relation to differentiation, we note that Apex Legends is more battle-royale

focused, whilst premium CoD titles are multiplayer games with different
modes (though we note CoD Warzone is battle-royale focused). Apex
Legends offers some similarities in graphics style and war settings to CoD,
though its mechanisms and gameplay are significantly different.

7.270 Overall, there are limited alternative shooter games comparable in both size
and experience to CoD. Fortnite is similar or bigger in terms of user spend on
PlayStation, but its gaming experience is significantly differentiated. Battlefield
and Apex Legends provide a more comparable experience but are [] less
successful than CoD on PlayStation (indicating that given the choice, players
prefer CoD over these other similar games). Therefore, we consider that any
withholding of CoD is likely to affect consumers that game on PlayStation in
terms of the range and quality of games available to them.

• SIE’s first-party and third party exclusive catalogue

7.271 Microsoft submitted that SIE has a strong catalogue of first and third party
exclusive games: SIE publishes bestselling first-party titles such as God of
War, The Last of Us, Marvel’s Spider-Man, Uncharted, Ghost of Tsushima,
etc. The Parties submitted that SIE also has a portfolio of high-quality
exclusive third party content including Genshin Impact, Final Fantasy 16,
Bloodborne, Sackboy: A Big Adventure, Street Fighter V, Sifu and the Silent
Hill 2 remake. 547

7.272 Results from our survey indicate that exclusives are one of the main factors
that drive console choice (noting above the point that our survey was targeted
CoD gamers and needs to be interpreted accordingly). Among those
respondents to whom content on a console is important (89% of our sample),
69% of them indicated that the availability of exclusive games would influence
their choice of purchasing a PlayStation.

7.273 However, data from [] shows that SIE exclusive titles are each significantly
smaller than CoD in terms of consumer spend on PlayStation. For example,
SIE’s best performing exclusive title in 2021 globally was []. Its total
consumer spend for SIE was [] of that from CoD in 2021, []. Another
successful title is SIE’s first-party title Marvel’s Spider-Man which generated a
consumer spend of c.$ [] million across all its titles over the period 2019-

546 [] Internal Document.

547 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.43c.

155
 2022, ie [] of the revenues generated by CoD only on PlayStation in
2021. 548

7.274 [] if we consider all SIE’s (fully) exclusive content portfolio in aggregate,
their size in consumer spend terms is [] than CoD in 2021. 549

Conclusion on alternatives to CoD

7.275 As noted above, the loss of CoD would represent a net reduction in SIE’s
range, which cannot be fully offset by games that are already available on
PlayStation.

7.276 Based on the above evidence, we conclude that there are few other games—
including franchises, shooter games and SIE’s exclusives—with comparable
consumer spends and gameplay time as CoD that are available on
PlayStation. This therefore means that the elimination of the CoD franchise
from PlayStation’s range would represent a significant reduction in range in
relative terms.

7.277 We also conclude that most of the above games offer a differentiated
experience to CoD:

7.278 Non-shooter games (including SIE’s top exclusive games) like FIFA, GTA,
NBA, Marvel’s Spider-man etc are a different genre to that of CoD; and these
games would not be a close substitute for CoD games in the absence of CoD
on PS.

7.279 Within shooter games, the closest alternative to CoD is Battlefield – however,
it was significantly behind CoD in terms of consumer spend and gameplay
time in 2021. There are other shooter games like Fortnite and Apex Legends
that are comparable to CoD in terms of performance, but there is sufficient
differentiation between them and CoD that these cannot be considered close
alternatives to CoD.

7.280 The extent of differentiation, including within shooters, of CoD’s closest

alternatives means that any foreclosure strategy will significantly impact
PlayStation’s ability to offer more of those differentiated games by contributing
to a worse range and less consumer choice.

7.281 We note that PlayStation also has the option of expanding an existing game
or creating a new game, either using SIE’s first party studios or by tying up

548 [] Internal Document.

549 [] Internal Document.

156
 with third party publishers, in response to the Merger. We discuss this in
Chapter 9.

Contractual arrangements and negotiations

7.282 In this section, we assess the impact that the Merging Parties’ contractual
arrangements and negotiations could have on the Merged Entity’s ability to
foreclose SIE. Activision has an existing contract with SIE covering the
arrangements between the two entities with regards to the availability of CoD,
to SIE PlayStation. This contractual arrangement lasts until the end of 2024
[].

7.283 In addition, Microsoft has made a separate offer to SIE to keep CoD on
PlayStation post-Merger. 550 Microsoft has also noted that it has entered into a
legally binding 10-year agreement with Nintendo to bring CoD to Nintendo
platforms post-Merger. 551 Microsoft added that an offer was made to Steam
that was ultimately declined. 552 As this theory of harm is focused on SIE for
the reasons already discussed, we have focussed our assessment on any
potential impact on the Merged Entity’s ability stemming from contractual
arrangements and negotiations with SIE.

Parties’ and SIE’s views

7.284 Microsoft submitted that contractual arrangements would limit the Merged
Entity’s ability to foreclose because SIE has more than ample time to adapt its
commercial strategy. Microsoft initially stated that this is because access to
CoD is guaranteed through to [] the end of 2024 under SIE’s existing
agreement with Activision and through at least the end of 2027 if it were to
accept Microsoft’s contractual offer. 553 Microsoft also submitted that the
existing agreement between SIE and Activision includes []. 554

550 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 3.47 and 3.50(g). See also
Microsoft, letter to the CMA and Microsoft, letter to the CMA.
551 Microsoft explained that it entered into a final agreement with Nintendo on [] February 2023 to publish CoD

titles on Nintendo post-merger. Microsoft explained that the agreement provides that Microsoft will develop and
publish future native console versions of CoD titles for Nintendo platforms for [] 10 years and that Microsoft will
publish future CoD versions for Nintendo platforms on the same date as the release of those versions on Xbox
console platforms and will maintain feature and content parity to the console versions published on Xbox console
platforms, [] (see Microsoft response to the Provisional Findings, 2 March 2023, paragraphs 1.2 and 4.6).
552 Microsoft, letter to the CMA. We note that Valve has commented publicly that an agreement with Microsoft is

not necessary to keep CoD on Steam. See, for instance, videogameschronicle.com, accessed by the CMA on 17
January 2023.
553 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.43(g).
554 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.10. See also Microsoft,

email from Microsoft to CMA.

157
7.285 Since those submissions, the offer to SIE [] to cover a ten-year period
[]. 555 As far as we are aware, this offer from Microsoft has not been
accepted by SIE. While SIE [], 556 Microsoft submitted that []; 557 we
understand that []. 558

7.286 SIE submitted that no contractual protections can ever provide proper
protections against a foreclosure strategy, nor would any contractual
arrangement be able to predict the myriad foreclosure strategies available to
Microsoft. 559 SIE also stated that Microsoft has mischaracterised its initial
offer to SIE, 560 []. 561

7.287 Both Microsoft and SIE have provided detailed chronologies to the CMA of the
negotiations between Microsoft and SIE, providing their own commentary on
why certain aspects of certain offers should or should not have been accepted
by the other party, as well as the respective motivations of Microsoft and
SIE. 562

Our assessment

7.288 The MAGs state that:

‘The CMA’s assessment of the ability of the merged entity to foreclose its
rivals is unlikely to place material weight on contractual protections, for
example, to continue supplying both the current version and future upgrades
of the input. In practice, such contracts may not completely remove a firm’s
ability to harm its rivals, given that certain rivals might not be covered by these
contracts, the contracts might not protect all ways in which the
competitiveness of rivals could be harmed, and the contracts may be of
limited duration. Moreover, over time contracts may be renegotiated or
terminated, and firms may waive their rights to enforce any breaches in light
of their overall bargaining position (reflecting the change in market structure
brought about by a merger). However, the CMA may consider any financial or
reputational costs of terminating contracts in its assessment of foreclosure
incentives.’ 563

555 See Microsoft, letter from Microsoft to the CMA. See also Microsoft, response to TOH1 working papers.
556 []; [] response to the CMA’s s109 notice.
557 See Microsoft, letter to the CMA.
558 Microsoft response to working papers.
559 SIE response to the phase 2 Issues Statement, 28 October 2022, paragraph 16; and SIE response to the

Provisional Findings, 1 March 2023, paragraph 19.

560 SIE response to the phase 2 Issues Statement, 28 October 2022, paragraph 16. See also SIE response to the

CMA’s s109 notice.

561 See SIE response to the CMA’s s109 notice.
562 See for instance, Microsoft, letter to the CMA; and SIE response to the CMA’s s109 notice. See also Microsoft

response to the Provisional Findings, 2 March 2023, paragraph 1.4.

563 CMA129, paragraph 7.15.

158
7.289 In our view, these considerations apply to the present case. Contractual
protections: (i) may not account for all the possible foreclosure mechanisms
that could be available to the Merged Entity, (ii) may be renegotiated or
terminated early, or (iii) may not be enforced depending on the respective
parties’ bargaining positions. Nor can we be sure that SIE (or any other third
party) would be able to enforce the terms of any relevant contracts should it
need to do so.

7.290 In any case, and notwithstanding the general position on contracts set out in
the MAGs, the relevant SIE protections that are currently in place with
Activision only last until end-2024 []. 564 Moreover, the statement by
Microsoft that the agreement [], 565 [].

7.291 While we acknowledge that the existing contractual arrangements between

SIE and Activision may provide SIE with some protection in the short-term
(but noting the general limitations with contracts as explained above), the
relevant protections are clearly of limited duration. As a result, even if they did
impact the Merged Entity’s ability to foreclose in the short term to some
extent, they are not of sufficient duration to have a material impact on our
competitive assessment, which considers potential concerns with a longer
time horizon.

7.292 As there is currently no agreement in place between Microsoft and SIE in

relation to Activision content, we do not consider that the negotiations referred
to above have any impact on our assessment of the Merged Entity’s ability to
engage in foreclosure. We also do not consider it appropriate or relevant to
comment on the various motivations and behaviour of Microsoft and SIE
during these negotiations.

7.293 Likewise, when considering the 10-year agreement entered into by Microsoft
with Nintendo, in addition to noting that this theory of harm is primarily
focussed on SIE for reasons already explained, we have also not been
presented with evidence to show with sufficient certainty that this will lead to
CoD ultimately becoming available on Nintendo. The agreement itself
imposes [] Microsoft to develop and publish future native console versions
of the CoD titles for Nintendo platforms after completion of the Merger.
Microsoft submitted this []. 566 As explained above, the Nintendo Switch has
certain technical limitations compared to the latest PlayStation and Xbox
consoles (including, for instance, storage capacity). These limitations would
need to be overcome for CoD to become available on the Nintendo Switch.

564 Parties Internal Document.

565 See Microsoft, email to CMA.
566 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.74.

159
 [], or any detailed plans for how the optimisation of CoD for Nintendo is
intended to be achieved. This is notwithstanding any further disagreements or
issues that may become apparent as both parties move forward with their
respective obligations under the agreement. For example, to the extent that
[]. Further, it is not clear at this stage the degree to which the CoD
experience would differ on Nintendo. This is acknowledged []. 567 We
therefore do not consider it appropriate to place material weight on this
agreement in terms of the Merged Entity’s ability and/or incentive to foreclose
given the early stage of the process and unforeseeable impacts on the
market, and the more general limitations around contractual protection as
noted above.

7.294 As such, our view is that Microsoft’s contractual arrangements are not likely to
have any significant impact on its ability to foreclose SIE.

Partial foreclosure

7.295 As set out above, we consider that the Merged Entity will also be technically
able to engage in a range of partial foreclosure strategies that do not involve
total withdrawal of access to CoD, such as releasing Activision titles on rival
consoles at a later date or with fewer features, degrading the graphical quality
of Activision content on rival consoles, or raising the wholesale price.

7.296 In this section, we assess whether the Merged Entity would have the ability to
foreclose rivals console gaming services by adopting those partial foreclosure
strategies on a standalone basis.

Parties’ views

7.297 Microsoft submitted that the Merged Entity would lack the ability to bring about
the partial foreclosure of rivals by making certain features of Activision’s
games exclusively available on Xbox, and/or degrading performance of and/or
withholding technical support for Activision’s games on rival console
platforms. Microsoft submitted that:

(a) CoD has provided exclusive or timed-exclusive downloadable content for

either SIE or Microsoft since CoD 2 in 2005, and these arrangements
have not resulted in either SIE or Microsoft being impaired in their ability
to compete at times when the other had a limited exclusive. 568

567 Microsoft Internal Document.

568 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.42 (a).

160
 (b) Activision already has a strong incentive to come up with different forms
of content and marketing exclusivity that it can monetise in its negotiations
with Microsoft and SIE. It submitted that, since 2005, these marketing
arrangements have included (i) exclusive console marketing
arrangements following the release of new titles and downloadable
content; (ii) priority access to new maps (until these were phased out
following the introduction of cross-platform play); (iii) exclusive access to
the online alpha version of the game and access to the beta version of the
game 5 days earlier than gamers on Xbox consoles or PC; (iv) game
bonuses such as extra “tier skips” on the battle pass; (v) the ability to
access additional “experience points” (eg through exclusive events); and
(vi) certain in-game character customisations and content bundles. 569
Microsoft submitted that none of these foreclosed Xbox in the console
market, and that it is not credible to suggest that there are further forms of
exclusivity which Activision could have monetised but has not considered,
and which could have a material impact on downstream competition. 570

(c) Past attempts by Activision to steer users towards PlayStation to play

CoD had very little impact on Xbox. Between Call of Duty: Black Ops 3
(2015) and Call of Duty: Modern Warfare (2019), Xbox maintained around
[]% of CoD gamers, despite Activision’s efforts to promote PlayStation
instead. The attempt to steer gamers towards SIE’s platform did not have
a foreclosure effect on Xbox. 571

Our assessment

7.298 A competitor [] submitted that Microsoft would have the incentive to engage
in one or more of the following partial foreclosure strategies: i) raising the
price of CoD on PlayStation relative to Xbox or increasing CoD’s B2P prices
on both Xbox and PlayStation while offering lower prices for Game Pass, ii)
releasing CoD on Xbox ahead of its release on PlayStation, which could have
a major impact on spend and gameplay hours on PlayStation, iii) degrading
the quality and performance of CoD on PlayStation compared to Xbox (eg,
Xbox players could enjoy greater frame rates, richer textures, higher
resolutions, or enhanced maintenance and support), iv) degrading CoD to
ignore PlayStation-specific features (eg, better controller haptics) restricting or
degrading multiplayer experience on PlayStation (eg, by refusing to enable
CoD’s multiplayer on PlayStation or, more likely, degrade the multiplayer
experience by reducing speed or quality, or by charging a PlayStation users

569 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.42 (b).
570 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.42 (c).
571 Microsoft response to the phase 2 Issues Statement, 31 0ctober 2022, paragraph 3.42 (d)

161
 additional fees to access multiplayer features), vi) making in-game content or
features exclusive to Xbox indefinitely or for a limited period, and/or vii)
bundling new releases of CoD within the next generation of Xbox. 572 This
competitor [] submitted that any degradation in the price, performance, or
quality of play on PlayStation or any delays on release would quickly harm its
reputation and cause a loss of engagement and players. 573 It submitted
evidence about the volume and nature of online conversations surrounding
consoles and games, and data which, in its view, showed that quality issues
such as crashes, freezes, and glitches, as well as graphics and load times
have a significant impact on consumer satisfaction. 574

7.299 This competitor [] also submitted that the position that Microsoft has the
ability to totally, but not partially, foreclose PlayStation is not sustainable. 575

7.300 In our survey of PlayStation CoD gamers, we asked what they would have
done at the time of their last PlayStation purchase if the most recent CoD
game they owned at the time had still been available on PlayStation, but with
less content (not able to access certain content such as specific levels, maps
or gameplay modes). In this scenario, around two thirds of gamers said that
they would have bought only the PlayStation (67%). Just over one in ten
would have bought another gaming device instead of the PlayStation (12%),
and just under one in ten would have bought the PlayStation and another
gaming device (9%). 576

7.301 We consider, however, that this question (which was drafted early in our
investigation) is not sufficiently precise to produce accurate predictions of
likely diversion rates in response to partial foreclosure strategies. Instead, it
aggregates several potential partial foreclosure strategies and asks a general
question about what gamers would do in response to one or more of them.
Furthermore, because our survey was primarily designed to capture
consumer behaviour in response to a total foreclosure strategy using CoD,
responses to the single question about partial foreclosure may have been
framed and thus biased by this. As such, we place limited weight on the data
that was produced in answer to the partial foreclosure survey question.

7.302 Based on the evidence that we have seen, we do not believe that Microsoft
would have the ability to foreclose PlayStation solely through partial
foreclosure strategies (to the extent that such strategies do not, in effect,
equate to total foreclosure). First, we place some weight on the Parties’

572 [] email to the CMA. See also [].

573 [] email to the CMA.
574 [].
575 [].
576 CMA, DJS Research report: PlayStation gamer research, 8 February 2023.

162
 arguments that Activision has already granted marketing and content
exclusivity to PlayStation and Xbox at different times and on different versions
of CoD. These do not appear to have materially affected their respective
ability to compete, eg as proxied by limited variations in their respective
market shares following these events. We acknowledge that this type of
analysis can only have a certain amount of evidential weight, as it is difficult to
control for the impact of those marketing and content exclusivity strategies on
either PlayStation’s or Xbox’s ability to compete, which is likely to be affected
by other factors. 577

7.303 More generally, we found above that CoD accounts for around []% of
gametime on PlayStation – this means that a total foreclosure strategy leading
to CoD no longer being available at all on PlayStation would represent a
[]% reduction in PlayStation’s range. In case of partial foreclosure,
however, PlayStation would not lose []% of its range; rather, []% of its
range would suffer a quality deterioration (or price increase) that would
represent only a proportion of the value that gamers derive from CoD. Even in
circumstances where gamers are sensitive to game degradation, as SIE
submits, given that []% of PlayStation’s range would not be directly
affected, and the remaining []% of its range would remain intact, albeit
deteriorated to an extent, we consider the overall impact on PlayStation’s
competitive offering would be small and insufficient to substantially reduce its
effectiveness as a competitor.

Conclusion on the Merged Entity’s ability to foreclose

7.304 Our view is that the above evidence shows that the Merged Entity would have
the ability to engage in total foreclosure of PlayStation, including its
distribution storefront and subscription services:

(a) CoD is currently an important component of PlayStation’s range of game

offerings. It is one of the three largest franchises on PlayStation and
contributes a significant share of PlayStation’s revenue and gameplay,
particularly within shooter games. CoD is important in driving console
sales, with nearly a quarter of our survey respondents stating that they
would divert away from PlayStation if CoD were no longer available on
there – these respondents amount to a significant proportion of all
PlayStation users and a significant proportion of consumer spend that
would be diverted away from PlayStation. These gamers would likely

577For example, when CoD marketing and content exclusivity was granted to PlayStation, this may have had a
negative impact on Xbox’s ability to compete. At the same time, there may have been other factors which instead
had a positive effect on Xbox and offset the negative impact caused by those strategies, leading to a net impact
on Xbox which was neutral.

163
 spend less time and money on the console than they did previously and
are likely to induce further diversion due to the presence of strong direct
network effects.

(b) There are insufficient alternatives available on PlayStation that could

offset any reduction in PlayStation’s range caused by a withholding of
CoD. The remaining games available on PlayStation, including alternative
large franchises, shooter games, and PlayStation’s exclusives, are either
(i) sufficiently differentiated from CoD that their availability would not
compensate for the reduction in PlayStation’s range, or (ii) are
significantly smaller than CoD in terms of spend and gameplay, indicating
that they are less preferred by gamers relative to CoD and would be
unable to fill the void created by a withholding of CoD.

7.305 With respect to Microsoft’s submission that a range of other games would be
available for gamers to play on PlayStation, the impact of foreclosure would
be to move from a scenario where gamers already have a choice of those
games in addition to CoD to one in which they have more limited choice. This
would represent a reduction in the game range available to them. The time
spent by PlayStation gamers playing CoD when given a free choice between
CoD and other games is a highly relevant measure of the significance of that
reduction in range. The observation that CoD accounts for []% of
PlayStation’s gameplay time in the UK, and that its removal would trigger
consumers to switch away in substantial numbers is—alongside other
evidence on the importance of CoD in consumer choice—evidence of an
ability to materially weaken PlayStation as an alternative.

7.306 We recognise that partial foreclosure strategies can be incremental and,

depending on their type and severity, can have a similar impact to a total
foreclosure strategy. We consider that partial foreclosure strategies which,
alone or in combination, amount to something similar to a total foreclosure
strategy would materially weaken PlayStation. For example, if the degradation
of or price increase in a game was so extreme that gamers stopped playing it,
or if timed exclusivity was sufficiently long-term such that it had the same
impact as total exclusivity, we consider this would be equivalent to total
foreclosure and therefore subject to the same analysis as total foreclosure
contained in this Chapter. But we also recognise that there can be milder
forms of partial foreclosure that would not amount to or even come close to
total foreclosure, such as marketing and content exclusivity, which Activision
has granted to Xbox and PlayStation at different points in time. We consider
that the ability and incentive of the Merged entity to engage in partial
foreclosure strategies that resemble total foreclosure can be analysed under
the same framework as total foreclosure. However, for the reasons explained

164
 above, we believe that the types of partial foreclosure that fall short of total
foreclosure would not materially weaken PlayStation.

7.307 On the basis of this evidence, we believe that the Merged Entity would not
have the ability to materially weaken PlayStation through partial foreclosure
strategies, where those strategies do not in effect amount to total foreclosure.

Incentive to foreclose

Introduction

7.308 In this section, we assess whether a total foreclosure strategy would be

profitable and, therefore, whether the Merged Entity would have an incentive
to engage in such a strategy. As part of this assessment, we consider the
extent to which the Merger increases the incentives to foreclose downstream
competitors relative to the situation absent the Merger.

7.309 The MAGs state that ‘the assessment of incentives typically involves a
combination of quantitative and qualitative evidence, though the balance will
vary between cases. The CMA may undertake more extensive quantitative
analysis in simple markets with high quality data, but it will focus on a
qualitative assessment in complex and dynamic markets, where firms’ current
positions and margins may not be a good guide to the future, and strategic
considerations may play a greater role. In any event, its focus will be on the
relative magnitude of the overall cost and benefit of foreclosure, not on
predicting the exact size of each element.’ 578

7.310 We have assessed the incentives to foreclose with a mix of both qualitative
and quantitative evidence. This is because financial modelling can quantify
some gains and losses of foreclosure (ie, short-term profitability), but it
ignores others (ie, long-term strategic objectives). In this case, beside our
quantitative modelling, we considered Microsoft long-term strategic objectives,
its behaviour in relation to previous acquisitions, and the relevance of
potential contractual arrangements.

7.311 Given our conclusions on the lack of ability of the Merged Entity to adopt a
partial foreclosure strategy, we focus on the gains and losses that would arise
out of a total foreclosure strategy.

7.312 The rest of this section presents:

578 CMA129, paragraph 7.18.

165
 (a) the Parties’ views,

(b) third parties’ views,

(c) our assessment of the quantitative and qualitative evidence.

Parties’ views

7.313 Microsoft submitted that it does not intend to remove CoD from PlayStation or
degrade access to the franchise. 579 Microsoft said its incentives to continue
distributing CoD to SIE are clear from the deal valuation, its public statements
and correspondence with SIE, public statements by SIE, the strategic
rationale for the Merger, and Microsoft’s internal documents in relation to the
Merger. 580

7.314 Microsoft further submitted that:

(a) its past business practices are consistent with its stated position; 581

(b) cross-platform play is a powerful disincentive to foreclosure; 582

(c) contractual protections provide a powerful disincentive to withdraw

content; 583 and

(d) the CMA’s survey of PlayStation users suffers from design flaws which
bias its results, and in any case, the CMA survey results strongly support
the absence of incentive to foreclose, 584 and show that foreclosure of S is
not possible. 585

7.315 In addition, Microsoft submitted a quantitative analysis which models the

financial gains and losses of withdrawing CoD from PlayStation. 586

7.316 In response to our Provisional Findings, Microsoft submitted that it has no

incentive to withhold CoD from PlayStation because this would involve
immediately foregoing the associated revenues, which account (in present
value terms) for approximately []% of the overall value of the deal with
Activision. Microsoft noted that this [] and would []. 587

579 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.50.
580 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.49.
581 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph (ii), page 54.
582 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph (iii), page 57.
583 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph (iv), page 60.
584 Microsoft, response to TOH 1 working papers.
585 Microsoft submission to the CMA.
586 Parties response to the EC’s RFI.
587 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.76.

166
7.317 Microsoft submitted that, when using more recent data, and when taking into
account five years of potential losses on PlayStation, our lifetime value (LTV)
model shows that Microsoft would not profit from a total foreclosure
strategy. 588

7.318 Microsoft also submitted that SIE’s claim that the gains from any alleged
strategic benefits (see below) would be large enough to compensate Microsoft
for the financial loss incurred by withholding CoD from PlayStation was based
on “unfounded speculation”. 589

Third parties’ views

7.319 SIE submitted that Microsoft would have the incentive to withhold CoD from
SIE, its closest competitor. 590

7.320 SIE submitted an economic analysis which showed that the Parties would
have an incentive to withhold the CoD franchise from PlayStation. The
analysis concluded that a total foreclosure strategy would be profitable for the
Merged Entity if more than []% of current PlayStation users were to switch
to Xbox. 591

7.321 SIE provided evidence that suggests the switching rate is higher than []%.
In particular, SIE presented four possible switching rate scenarios based on
gamers’ ‘engagement level’ with CoD (ie the share of time spent playing CoD
relative to the total time spent playing on PlayStation). In all scenarios, the
switching rate was higher than []%. 592 SIE also submitted an additional
analysis based on survey data of PlayStation users that suggested a
switching rate of []% if Call of Duty were to become exclusive to Xbox. 593

7.322 More details on this analysis are presented in Appendix E.

7.323 In response to our Provisional Findings, SIE submitted that our assessment
wrongly downplays the “significant” strategic benefits to Microsoft adding
Activision content to Game Pass. 594

588 Microsoft response to the Provisional Findings, 2 March 2023, paragraphs 2.86 and 2.92.
589 Microsoft response to SIE’s observations on the Addendum Provisional Findings.
590 SIE, submission to the CMA.
591 SIE, submission to the CMA.
592 SIE, submission to the CMA.
593 SIE, submission to the CMA. We note that, based on the information provided, we have reservations about

the survey methodology. For example, there is no indication that the sample is representative; the selected
answers are not a measure of diversion; awareness of CoD does not imply the respondent played CoD; and the
survey approach does not meet the evidence standards set out in our published good practice guide. For these
reasons, we have no given no material weight to this evidence.
594 SIE response to the Addendum Provisional Findings, 31 March 2023, paragraphs 11-12.

167
Our assessment

7.324 In this section, we discuss:

(a) evidence from quantitative analyses,

(b) less quantifiable gains and losses from foreclosure,

(c) Microsoft’s behaviour in past acquisitions of game publishers, and

(d) any protections offered by contracts.

7.325 Our assessment focuses on total foreclosure because, as discussed above,

we have concluded that Microsoft would not have the ability to materially
weaken PlayStation through partial foreclosure strategies.

Quantitative modelling

7.326 In this section, we assess the economic analyses received during our inquiry
and we present our own financial modelling.

7.327 Quantitative analyses provide only a partial assessment of the incentives to

foreclose. Several gains and losses described below are not captured by any
financial analyses. Microsoft acknowledged this and stated that quantitative
analyses are a static comparison of the potential costs and benefits of a
withholding strategy while holding everything else equal, and therefore do not
capture dynamic considerations. 595

Parties’ submissions

7.328 The Parties submitted an economic analysis of incentives to engage in total

foreclosure. They submitted that the Merged Entity would have no financial
incentive to withhold the CoD franchise from SIE’s console gaming platform
[]. 596 Based on this analysis, the Parties estimated that more than []% of
current CoD players on PlayStation would have to switch to Xbox for it to be
profitable for Microsoft to make CoD exclusive to Xbox. This is called the
‘critical diversion ratio’. The Parties submitted that a rate of consumer
switching in excess of the critical diversion ratio of []% found in their
analysis would be ‘implausibly high’. 597 More details on this analysis are
presented in Appendix E.

595 Microsoft response to working papers.

596 Parties response to RFI; and Parties, submission to the CMA.
597 Parties response to RFI.

168
7.329 The Parties relied on inputs and data from Activision and Microsoft. The
model has some advantages:

(a) it accounts for the presence of multi-homers, ie gamers who own both a
PlayStation and an Xbox and therefore might switch console without
facing the cost of buying a new one;

(b) it accounts for gamers potentially switching from consoles to PC; and

(c) it accounts for the possibility of gamers choosing between B2P and
subscription options.

7.330 However, we have material doubts about the methodology and inputs to this
economic model. Some of the model’s results also seem counterintuitive. In
particular:

(a) The Parties perform a sensitivity analysis in which they consider the
impact on the critical diversion ratio if gamers were to switch to Xbox only
(and not to PC). They estimate that the share of CoD users who would
need buy an Xbox for the foreclosure strategy to be profitable increases to
[]%. However, [] 598 [], it is not clear why the critical diversion ratio
would increase under this assumption. The Parties state that this is
because the revenues recouped from multi-homers decrease, but we find
no reason why this should be the case. We consider this casts doubt on
the reliability of the baseline results.

(b) Changing the diversion to PC and Xbox by a significant amount (eg

increasing the share of CoD gamers on PC by 10% and decreasing those
on Xbox by 10% or vice versa) changes the critical diversion ratio only by
[]%. Given that the revenues recouped on Xbox and PC are very
different on the Parties’ model, the corresponding low impact on the
critical diversion ratio seems unjustified.

7.331 In the ordinary course of business, Microsoft estimates the value of an Xbox
customer and calls this their “lifetime value” (LTV), based on actual purchase
data of Xbox Series X/S owners. The LTVs are essentially []. 599 As part of
its analysis, Microsoft used LTVs as an input to capture the value of gamers
that would hypothetically switch in response to the withholding of CoD
releases from PlayStation. These LTVs are based on [] and, the Parties
submitted, they reflected the LTVs of customers who bought []. In order to

598 Parties Internal Document.

599 Microsoft response to the CMA’s s109 notice.

169
 estimate the trend of console LTVs for later years, Microsoft has []. 600 In
relation to the above, we note that:

(a) While we recognise the reasoning behind the adjustments, we are of the
view that it does not seem sufficient to justify the large magnitude
adjustment.

(b) Given these LTVs are used, unadjusted, in the ordinary course of
business, we consider that they are likely to be more reflective of
customer value when not adjusted, and we expect that Microsoft’s
adjustments are likely to cause a significant underestimate in the Merged
Entity’s incentive to foreclose.

(c) We consider there is no reason to believe switching would not happen at

the beginning of the next console generation. Therefore, the downward
adjustment applied by Microsoft seems unjustified.

7.332 Overall, we consider these limitations have the effect of understating the
Merged Entity’s gains and overestimate the critical diversion ratio. As such,
we place no material weight on these results and focus instead on our
analysis, which takes into account the Parties’ submissions and data, but
adjusts for its apparent weaknesses.

7.333 Microsoft also presented a more stylised model during the CMA’s site visit
considering the incentive to engage in total foreclosure. Microsoft described
this analysis as a ‘mini’, ‘stylised’, and ‘simplified’ version of their original
incentives analysis which should not be preferred to their original analysis
which accounts for additional potential benefits. 601 We provide further details
on this analysis in Appendix E. However, in light of Microsoft’s preference for
the Parties’ more detailed analysis discussed above, the limitations identified
with the site visit model which are discussed in Appendix E, and our reliance
instead on our own analysis, we place no material weight on the results of this
analysis and therefore do not discuss it further here.

Third parties’ submissions

7.334 An analysis submitted by SIE estimated that []% of PlayStation users need
to switch to Xbox for total foreclosure to be profitable. 602 To make this result
comparable to the Parties’ submissions and our own survey, we converted

600 Parties response to RFI.

601 Microsoft response to working papers.
602 Based on device-level data. SIE, submission to the CMA [].

170
 this as the share of CoD gamers on PlayStation that needs to switch to
Xbox. 603 The resulting equivalent critical diversion ratio is []%.

7.335 More details on this analysis are presented in Appendix E.

7.336 We consider that this analysis has the following limitations:

(a) Absence of multi-homing: there is no recoupment through sales of CoD

on Xbox consoles and PCs already owned by PlayStation gamers. As
explained above, in theory the effect of multi-homing is ambiguous;
however, in this analysis SIE assumed all spend of diverted gamers on
PlayStation is lost and transferred to Xbox. This likely overestimates
PlayStation’s losses and Xbox’s gains.

(b) No option to switch to a new PC: there is no recoupment through sales

of CoD on a PC. By ignoring a potential source of recoupment, the
analysis underestimates the foreclosure gains.

(c) Switching rates: when calculating the expected switching rate of users
from PlayStation to Xbox, it is assumed that the probability of a user
switching is equal to that user’s current engagement level with CoD. While
we believe that the two may be correlated, we do not have any evidence
that supports a perfect link between the probability of switching and
engagement.

(d) Inconsistent profit margin: the analysis assumed a third-party revenue-

share of []% with CoD on PlayStation, which is larger than the current
third-party margin of []%. This therefore underestimates the losses the
Merged Entity would incur by withholding CoD from SIE by [], all else
equal. A sensitivity analysis examined the impact of using the effective
margin SIE currently generates for CoD, after taking into account all
payments from SIE to Activision. This results in a SIE margin of []%. 604
The proportion of PlayStation users that need to switch to Xbox for
foreclosure to be profitable under this scenario increases from []% to
[]%.

(e) Overestimates Xbox Game Pass spending: weak evidence supports

the []% uplift applied to overall user spending to account for the
additional spending that Game Pass is reported to generate. This is likely
to overestimate the additional gains that Microsoft will receive from

603 Data from [] indicates that []% of PlayStation gamers played CoD in 2021, as noted in our assessment of
Ability above. We therefore divided 8% by []% to obtain the share of CoD gamers that would need to switch for
total foreclosure to be profitable.
604 SIE submission to the CMA.

171
 spending on Game Pass, as subscription spend on PlayStation is already
incorporated. Removing the uplift in spending for XGP increases the
critical switching rate at which it becomes profitable to withhold CoD from
[]% to []%. 605

(f) Data: The analysis uses data from PlayStation as a proxy for Xbox. The
effect this has on the estimated gains and losses depends on the relative
profitability of CoD and other games on PlayStation and Xbox.

7.337 Overall, we consider that the analysis includes various factors that could
either increase or decrease the Parties’ gains (or the Parties’ losses) from the
input foreclosure strategy. Therefore, it is difficult to unambiguously say
whether the estimated critical diversion ratio is likely to be under- or over-
estimated. As with the Parties’ model, we place no material weight on the
results of this analysis and rely instead on our own analysis, which
incorporates SIE’s submissions and data.

CMA modelling

7.338 We carried out two separate quantitative analyses to assess the Parties’
incentives to engage in a total foreclosure strategy.

7.339 As with the analyses submitted by Microsoft and by SIE, our analyses have
advantages and disadvantages. However, our analyses account for some of
the drawbacks of the analyses submitted by Microsoft and SIE. For example,
we account for switching to PC and multi-homing, and we rely on diversion
ratios estimated directly from our survey. For these reasons, we consider our
model to warrant greater evidential weight than those of Microsoft and SIE.
We discuss the relative advantages and disadvantages of our quantitative
model in detail in Appendix E.

7.340 While we consider our model has some advantages relative to the other
models we have reviewed, both our models are subject to the caveats
mentioned above, namely they cannot take account of all possible gains and
losses emerging from a foreclosure strategy. An additional issue that our
model necessarily faces is that it can be difficult to cross-check the accuracy
of the profit margin estimates submitted by the Parties and third parties (other
than checking their methodology for producing it, which we have done where
possible).

605 SIE submission to the CMA.

172
7.341 With this in mind, we summarise below the main results of our quantitative
analyses.

7.342 First, we use the Parties’ submitted LTVs and our survey results to estimate
gains and losses of a potential total foreclosure strategy. Using LTVs allows
us to capture the gains from all revenue streams for Xbox. We also allow for
switching to PC, while we consider multi-homing is already reflected in the
LTV estimates. Indeed, we use the same LTV estimates as those used in the
normal course of business for Xbox; 606 this has the advantage of basing our
analysis on the same data used by Xbox for its internal decisions. We
consider that the users likely to switch from PlayStation to Xbox in response
to a foreclosure strategy are more likely to have spending patterns that
resemble early adopters than later adopters. This is based on evidence
showing that CoD gamers on PlayStation are likely to contribute considerably
to platform spend relative to the non-CoD gamers. 607 Therefore, we used an
Xbox LTV based on what the Parties defined to be ‘early adopters’. 608 In the
absence of an LTV from Activision, we computed the LTV for CoD based on
the annual profit per user for CoD that Activision makes on PlayStation. We
compute this LTV over a five-year horizon to reflect the opportunity cost of
withholding CoD from PlayStation consistently with the time horizon used for
the Xbox LTV. We combine these with our survey results. We provide more
details in Appendix E where we also address the Parties’ and third parties’
submissions on this analysis received in response to the Provisional Findings.

7.343 The analysis indicates a net loss of between $[] billion and $[] billion,
depending on how ‘users’ are measured. 609 These numbers sum five years’
worth of profits and losses on Xbox (including hardware, subscription, and
games) and Activision. The overall picture arising from this analysis is that
total foreclosure would not be profitable for the Merged Entity.

7.344 Second, we use 2021 data from Activision and SIE, as well as our survey
results, to estimate the gains and losses that would result from a total
foreclosure strategy, and in particular those gains and losses that would result
from customers switching away from PlayStation (and some of those being
recaptured by the Merged Entity). We allowed for multi-homing, both within
and across consoles generations, as well as switching to PC. We also
distinguish between profits made on other first-party and third-party games.
This analysis did also allow for modelling of a partial foreclose strategy;

606 Parties response to RFI. We use LTVs for the 2021 cohort.
607 See Appendix E for more details.
608 Microsoft, Annex to response to the Provisional Findings.
609 We used three different types of users-count: YAU who spent at least 10 hours or $100 in 2021 as measured

by Activision; the same measure from SIE; and the average number of MAUs in 2021.

173
 however, as we have found the Merged Entity would not have the ability to
engage in partial foreclosure for the reasons explained above, we do not
discuss this further.

7.345 The analysis provides a range of results, depending on the assumption made
on the switching rate of CoD gamers out of our sample and whether we
weight responses by revenue or gametime. Our survey focused on
PlayStation gamers who spent at least 10 hours or $100 on CoD. This means
that non-surveyed CoD gamers would have played CoD for a more limited
time and/or spent less money on it than surveyed gamers. For this reason,
and differently from Provisional Findings, we consider it appropriate to focus
on an outcome in which these gamers would not switch consoles in response
to total foreclosure. We also consider weighting our survey results by spend
has a more intuitive economic interpretation than weighting by gametime.
Based on this, we find that this analysis leads to a net loss of $[] million per
year, a loss equivalent to []% of Xbox’s revenues. We provide more details
in Appendix E.

7.346 Overall, our model suggests that it would not be financially profitable for the
Merged Entity to engage in a total foreclosure strategy.

7.347 In interpreting the results from our model as well as the models provided by
the Microsoft and SIE, we are conscious that quantitative modelling is
inherently subject to uncertainties and has to rely on assumptions where
information is imperfect. This limits the weight we can give to this type of
evidence. For this and the reasons explained above (ie, these analyses
cannot take account of all possible gains and losses emerging from a
foreclosure strategies), we consider it important to assess the Merged Entity’s
incentive to foreclose rivals by considering all the available evidence in the
round, including of Microsoft’s behaviour following previous acquisitions and
its stated strategy as reflected in internal documents.

7.348 In particular, we are conscious of the following elements that our financial
modelling does not account for (see below for a more thorough explanation of
their overall effect on the Parties’ incentives to foreclose):

(a) Our survey ignores the fact that Microsoft plans to add Activision content
to Game Pass. This is likely to underestimate the diversion ratio from
PlayStation to Xbox and, therefore, the Merged Entity’s incentive to
engage in total foreclosure.

(b) Our survey results account for reputational considerations to an extent.

Respondents will have taken into account their perception of Microsoft if it
were to engage in a total foreclosure strategy when responding to, for

174
 example, a question on whether they would switch to Xbox in those
circumstances. The fact that some gamers responded that they would still
switch suggests that, at least for some gamers, any such reputational
consequences would not prevent them from switching. Nonetheless, we
recognise that attitudes evolve over time, and that gamers’ future choices
may not be fully reflected in our survey results.

7.349 Strong direct network effects at a game level have two opposite effects on the
Parties’ incentives to foreclose:

(a) Diversion ratios estimated in our survey are likely to be an underestimate.

While some PlayStation gamers responded that they would not switch to
Xbox in response to a total foreclosure strategy using CoD, these same
gamers may have been willing to switch if they were to realise that other
gamers (including their friends) would do so. In this case, the Parties’
gains would be greater than estimated in our short-term financial
modelling.

(b) Another secondary effect that our survey does not capture is the effect of
cross-play. As explained below, we believe this to have a limited impact
on the gaming experience and, therefore, on the Merged Entity’s gains
and losses from total foreclosure.

Less quantifiable gains and losses from total foreclosure

7.350 In this section, we consider a range of other potential gains and losses from a
total foreclosure strategy that are more difficult to quantify on a comparable
basis. They include (i) furthering Microsoft’s strategy of expanding Game
Pass, (ii) any reputational impacts (good or bad), (iii) the strength of the Xbox
brand and user loyalty, (iv) the impact of network effects (including for games
that allow cross-play), and (v) the potential for entry, expansion, or
repositioning by rivals to disincentivise total foreclosure. We discuss some of
these broader strategic considerations below.

7.351 In response to our Provisional Findings, Microsoft submitted that long-term

strategic benefits are, by their nature, speculative and uncertain. It submitted
that there would need to be particularly clear evidence on strategic benefits to
counterweight more certain evidence suggesting no incentive to foreclose. 610

7.352 Microsoft also submitted that our LTV model already takes account of the key
issues raised as potential long-term strategic benefits. Specifically, Microsoft
said that, given the multi-year timeframe of the LTV calculations, these

610 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.109.

175
 calculations account for any longer-term benefits from a total foreclosure
strategy, including: 611

(a) New loyal customers: Microsoft observed that the LTV covers a period
over five years so, to the extent that Activision content can be used for
this strategic benefit, it is already being captured into the medium term in
the LTV calculation. 612

(b) Growing Game Pass: Microsoft submitted that, given the five-year period
captured by the LTV, any incremental growth of Game Pass is being
captured in the LTV calculation. 613

(c) Reputational benefits: Microsoft stated that, as withholding CoD would

directly contradict Microsoft’s public statements, the effect would be likely
to be very negative given the significant gamer backlash that would
inevitably occur following such a reversal. Also, Microsoft submitted that,
insofar as the suggestion is that those reputational harms would be
outweighed by having CoD on Xbox, there is no good evidential basis for
that conclusion. Microsoft said that, to the extent that CoD causes
switching, the five-year horizon in the LTV likely means that any
reputational benefits are captured in the calculation. 614

7.353 However, we note that the Xbox LTV is based on []. The CoD LTV is based
on []. As such, they do not account for less quantifiable considerations like
customer loyalty, reputational impact, or strategic considerations related to the
growth of Game Pass as it does not appear that those considerations are
captured in the underlying data from which the Xbox LTV is derived.

Expansion of Game Pass

7.354 Microsoft submitted that it intends to include Activision content on Game Pass
on a ‘day and date’ basis. 615 This is relevant to our assessment of incentives.
Indeed, any analysis that ignores this feature of the Merger would understate
the attractiveness of Xbox. In particular, this is likely to have three effects on
the Merged Entity’s incentives to foreclose:

(a) It may increase the number of gamers who switch from PlayStation to
Xbox. This is because CoD’s addition to Game Pass would add a way for
gamers to pay for CoD titles, which to some gamers may be preferable to

611 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.110.
612 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.110(a).
613 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.110(b).
614 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.110(c).
615 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 1.8(e).

176
 purchasing them on a B2P basis. 616 Our survey does not account for the
fact that CoD will be available on Game Pass. Microsoft and SIE do
attempt to capture this through the use of assumptions on profits made by
Xbox for customers who switch to Game Pass.

(b) On the other hand, CoD’s addition to Game Pass may cannibalise some
of the B2P sales of CoD. In its valuation model, Microsoft estimated this
cannibalisation to be []%, []. 617 The fact that Microsoft is nonetheless
adopting this strategy suggests that it expects it to be profitable in the long
run. This would also apply to overall revenue from customers switching
from PlayStation to Xbox.

7.355 Consistent with this, the valuation model prepared by Microsoft in anticipation
of the Transaction suggests that it expects its impact on Game Pass to be
positive:

(a) Microsoft assumed CoD and other Activision content would increase both
the number of subscribers and the revenue spent on Game Pass. 618 In
particular, the model assumes an additional [] users will subscribe to
Game Pass as a result of the Transaction. This represents an increase of
[]% in the subscriber base. 619 The model also assumes an increase
[] played on Game Pass.

(b) The net impact of the shift in mix from B2P to subscription is []. 620

(c) The growth of Game Pass appears to be an important strategic objective

for Xbox. The valuation model prepared by Microsoft assigns []% of the
valuation assigned to Activision to []. 621 This [] amounts to $[]
(including a Terminal Value of $[]). 622 Given the valuation is taken over
[] years, this amounts to a $[] valuation in net present value terms
per annum, on average, excluding the Terminal Value. Without applying
the discount rate to generate an NPV (and, therefore, on a comparable
basis to the quantitative analysis presented later in this chapter), the
annual value of this [] is $[].

7.356 This valuation does not include the effects of any foreclosure of PlayStation.
However, we consider it reasonable to expect that the benefits to Game Pass

616 We recognise that this may be beneficial to some customers. Here, our focus is on the impact on the Merged
Entity’s incentives only. We discuss benefits from Game Pass in our chapter on Countervailing factors.
617 Microsoft Internal Document; and Microsoft Internal Document.
618 Microsoft Internal Document.
619 Microsoft Internal Document.
620 CMA analysis of: Microsoft Internal Document. [].
621 Microsoft Internal Document.
622 Microsoft Internal Document.

177
 of including CoD would be larger if it were not available on another console
(including as a result of some PlayStation gamers switching to Xbox). As
such, we consider this to be relevant when assessing the profitability of a
foreclosure strategy.

7.357 Overall, we consider the potential for expansion of Game Pass represents an
important part of Microsoft’s strategy. To the extent a foreclosure strategy
would contribute to the expansion of Game Pass, we consider that this would
have at least some strategic value to Microsoft that would go beyond the sale
of consoles and games to the cohort of customers that would switch in
response to a total foreclosure strategy. When weighting all gains and losses
from foreclosure, we considered the value of an expansion in Game Pass
alongside the quantitative analysis set out above.

Reputation

7.358 Microsoft submitted that it would suffer reputational damage arising from non-
compliance with Activision’s contracts with SIE if it were to engage in a
foreclosure strategy. Microsoft said that withholding CoD from PlayStation
would result in backlash from gamers on social media. 623

7.359 For the avoidance of doubt, we are not considering a foreclosure strategy that
would require the Merged Entity to breach existing contracts with SIE.
However, we assessed Microsoft’s arguments in relation to the reputational
costs associated with a foreclosure strategy in relation to prospective
contracts and/or new releases.

7.360 We recognise that withholding CoD from PlayStation could have some
reputational consequences for Microsoft. For example, an industry report
stated that Activision and SIE faced some backlash in the past over temporary
exclusivity of a CoD game mode. The third party report states that, when CoD
Spec Ops: Survival was released exclusively on PlayStation for the first year,
there was some backlash on social media. The document links this backlash
to the gamers’ expectation of playing CoD: Modern Warfare on cross-play. 624

7.361 However, the evidence indicates that reputational effects may not prevent a
foreclosure strategy:

(a) First, we note that any negative impact on Microsoft’s reputation did not
prevent it from making some ZeniMax games exclusive to Xbox. In
response to the backlash due to Microsoft’s decision to make ZeniMax’s

623 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.58(d).
624 [] Internal Document.

178
 game Starfield exclusive to Xbox and PC, ZeniMax’s head of marketing
apologised. 625 This reputational cost did not result in Microsoft making the
game available on other platforms. While CoD [] than any ZeniMax
games, such that any reputational effect may reasonably be expected to
be larger in absolute terms, so too would the other gains and losses from
making it exclusive. We have not seen reasoning or evidence to suggest
that reputational effects would ‘scale’ disproportionately with size in a way
that would negate any incentive to foreclose CoD.

(b) Second, in addition to reputational costs, there are also potential

reputational gains from a foreclosure strategy. Given CoD’s popularity,
Xbox would benefit from an increased expectation that Microsoft would in
future be willing to invest in successful studios and release more games
exclusively on Xbox.

(c) Third, negative reputational effects can often be managed. For example,
exclusivity can be presented as a positive differentiating factor for Xbox,
and any limitations of CoD on PlayStation may be attributed to technical
reasons. In a scenario where no agreement is reached to supply CoD on
PlayStation, it would be difficult for consumers to discern whether this is
as the result of a foreclosure strategy or the failure to reach a deal on
reasonable terms for both parties. Given this ambiguity, a proportion of
affected consumers may not blame Microsoft.

(d) Fourth, we have no evidence of where a reputational impact would arise

or how it would affect customer behaviour. For example, if any backlash
against Microsoft were primarily emanating from loyal PlayStation gamers
who are in any event less inclined to switch to Xbox, it would not
significantly affect Microsoft’s incentives to engage in a foreclosure
strategy (since it would be unlikely to gain these consumers in any event).

7.362 Overall, we consider that negative reputational effects could have some
impact on Microsoft’s incentive to make CoD exclusive to Xbox, but that these
can be managed to a large extent by broader commercial and strategic
objectives. Therefore, we place limited weight on reputational effects in the
overall assessment of the Parties’ incentives.

Loyalty and switching costs

7.363 Microsoft submitted that, once the choice of console has been made, gamers
tend to remain loyal to their choice. Microsoft stated that rival consoles have

625‘Bethesda Apologizes for Starfield Xbox Exclusivity: Can Be 'Frustrating' For PlayStation Owners – IGN’,
accessed by the CMA on 22 November 2022.

179
 built a strong reputation and enjoy a high degree of brand loyalty. 626 Microsoft
also submitted that there are switching costs for gamers on PlayStation due to
the backward compatibility of its games and the fact that not all SIE’s games
are available on PC. 627 Microsoft submitted that other switching costs such as
entitlements (eg achievements and trophies earned through gameplay time),
as well as intangibles such as familiarity with the console hardware and
controller, and connections to the broader gamer network keep gamers
engaged with their console. 628

7.364 We note that, according to the survey referred to by Microsoft, not only rival
consoles, but also Xbox has built a significant degree of brand loyalty. 629

7.365 We acknowledge the Parties’ submissions that a degree of loyalty to gamers’

platforms is likely to exist. Such loyalty is likely to have two effects:

(a) First, loyalty to a platform will tend to reduce the total amount of switching
in response to a total foreclosure strategy using CoD. Our survey
nonetheless indicates that a significant proportion of PlayStation
customers would be willing to switch to Xbox. This suggests that the value
of being able to continue playing CoD (or the full version of CoD) can
outweigh loyalty to a given platform and associated switching costs.

(b) Second, if customers are loyal to their platforms, this increases the value
of consumers that eventually do switch from PlayStation to Xbox in
response to a foreclosure strategy. This is because inducing a switch to
Xbox will give rise to an opportunity to engender loyalty among those new
Xbox customers.

7.366 Evidence provided by a third party [] showed that as gamers played more of
Microsoft’s first-party games, they were less likely to upgrade from
PlayStation 4 to PlayStation 5 than other gamers. This held true even when
accounting for total gametime spent on PlayStation 4. 630 This suggests that
switching post-Merger may be higher than suggested by our survey. Given
the potential to generate strong brand loyalty, this also suggests that the value
of each switch may be higher (because each switch raises the chance of
future console sales).

626 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.78(b).
627 Microsoft response to TOH 1 working papers.
628 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.65.
629 ‘The Best SEO Companies’ survey Generational Brand Loyalty’, accessed by the CMA on 5 January 2023.
630 Based on account-level data. [] response to the CMA’s RFI.

180
 Cross-play and network effects

7.367 Microsoft stated that cross-play represents a disincentive to total and partial
foreclosure. According to Microsoft, a foreclosure strategy in a cross-play
enabled game like CoD would hurt gamers’ playing experience on all
platforms, cause severe reputational damage and gamer backlash (we
considered this above), and lead to significant financial damage. 631 In
particular, Microsoft submitted that:

(a) cross-play benefits gamers on all platforms because it leads to

improvements in user experience due to more efficient matchmaking.
Gamers benefit from shorter waiting times to find a match, lower disparity
in the average level of experience across gamers in the online match, and
improvements in the overall stability of the internet connection to the
server hosting the online match; 632

(b) a larger user base due to cross-play amplifies the impact of direct network
effects which operate at the market level, not the individual console or
platform level; 633 and

(c) CoD game titles have allowed gamers to cross-play since 2019, and
cross-play is now widespread. For example, over the last twelve months,
[]% of [] of CoD: Modern Warfare and CoD: Warzone had cross-play
enabled while playing online, which allowed them to play with gamers on
Xbox and PC. 634

7.368 As discussed above under features of the market for console gaming
services, the gaming industry is characterised by strong direct network
effects. They are especially strong for large multiplayer franchises like CoD.
Strong direct network effects at a game level have two opposite effects on the
Parties’ incentives to foreclose:

(a) Some gamers will switch from PlayStation to Xbox because other gamers
(including their friends) have switched, and they want to remain on the
same platform. This would increase the Merged Entity’s gains from
foreclosure.

(b) The presence of cross-play means that withholding CoD from PlayStation
could in principle also worsen the experience of Xbox users. A total
foreclosure strategy would reduce CoD’s overall user base. If some Xbox

631 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.56.
632 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.56.
633 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.56.
634 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.56.

181
 gamers stopped playing CoD as a result of the foreclosure strategy, the
Merged Entity’s losses would be higher than estimated in our short-term
financial modelling. Moreover, if the lack of cross-play means that gamers
will reduce their overall level of engagement with Xbox because they
cannot connect with gamers (eg friends) on PlayStation, then Xbox’s
losses would increase.

7.369 Evidence suggests that any impact from the second point above is likely to be
limited:

(a) Microsoft submitted that cross-play leads to shorter matchmaking times,

lower skill disparity, and better online match stability. 635 However, data
from the Parties indicates []. 636 Comparing the average time to search
for a user with cross-play relative to one without cross-play across
platforms in Europe, the difference between the quickest and slowest
matchmaking time is []. For context, global data from the Parties shows
that the waiting time for a match is []. Only after []. 637 UK data shows
that search times for users are [] for [] cross-play and non-cross-play
users. 638 Even if the impact on wait times were more significant, it is
unclear that this would have an impact on Microsoft’s profits on Xbox and
induce gamers to switch away from Xbox or significantly reduce their
overall engagement with the platform.

(b) Our survey estimates that []% of PlayStation gamers will buy an Xbox,
either in addition or instead of a PlayStation. This increases CoD’s user
base on Xbox, somewhat reducing the benefits from a larger user base
across platforms.

(c) Cross-play is a tool introduced by Activision in 2019. CoD’s popularity was

established well before then, indicating that its success is not dependent
on cross-play. In fact, there are successful games on both Xbox and
PlayStation that support cross-play (between console and PC) and yet are
not available on the other console, for example Halo, Genshin Impact,
and Forza. This suggests commercial and strategic considerations may
override any potential benefits from cross-play.

7.370 Given the evidence above, we consider the impact of cross-play to be limited.
As such, we consider direct network effects are still likely to have a positive

635 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.56.
636 Activision response to the CMA’s RFI.
637 [].
638 Activision response to the CMA’s RFI.

182
 impact on the Parties’ incentives to foreclose by inducing some gamers to
follow others in their choice of console.

7.371 Indirect network effects also exist, with publishers wanting to publish their
games on the platform with the most users. However, most publishers choose
to make their games available on as many platforms as possible. Publishers
told us that their incentives in this respect would not change as a result of a
potential withholding of CoD from PlayStation. 639 Therefore, we do not
consider indirect network effects to be central to our assessment.

Entry, expansion, and repositioning of other games

7.372 When considering the long-term incentives of the Parties, we have also
considered the long-term incentives of other game publishers to take
advantage of the gap left from CoD on PlayStation following a foreclosure
strategy. Without CoD on PlayStation (or with a reduced CoD offering on
PlayStation), other publishers may be incentivised to develop new games,
expand, or reposition existing games to gain users. If Microsoft expected this
to occur, it would be less incentivised to engage in foreclosure strategies.

7.373 In the Chapter below on Countervailing Factors, we assess this possibility of

entry and expansion in response to the Merger including barriers to entry and
expansion for game publishers. We find that barriers to entry and expansion
are high, especially to achieve a similar success to that of CoD. Barriers
include development costs, staff and franchise loyalty. We have not seen any
evidence of entry and/or expansion plans of rivals who will enter or expand
irrespective of whether the Merger proceeds.

7.374 On this basis, we consider entry, expansion, and repositioning of other games
would not be sufficiently likely to materially reduce the incentive of the Merged
Entity to engage in a total foreclosure strategy.

Past acquisitions

7.375 We considered whether Microsoft’s strategy following previous gaming

acquisitions could be informative of its likely strategy following the Merger.
The Parties stated that decisions regarding past mergers, which involved mid-
sized games, are not evidence of Microsoft’s incentives in this Merger
because they are fundamentally different, 640 in particular because CoD is a

639 [] response to the CMA’s RFI; and [] response to the CMA’s RFI.
640 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.55(c).

183
 multiplayer franchise, 641 and because mid-sized games are not evidence of
Microsoft’s incentives related to CoD. 642

7.376 Microsoft submitted that its acquisition of Minecraft is the closest analogue to
its proposed purchase of CoD, and that Microsoft kept Minecraft available on
PlayStation and expanded its reach to other platforms, including Nintendo,
after its 2014 acquisition of Minecraft’s publisher Mojang. 643 Microsoft
explained that Minecraft, like CoD, is a globally popular multi-player franchise
with a strong player community and social element that was available on
multiple platforms when Microsoft acquired it. Microsoft explained that, similar
to Minecraft, CoD monetises its content through: (i) subscription payments for
multi-player functionality (seasonal battle passes); (ii) merchandise (from its
dedicated online CoD Shop); and (iii) game enhancing features (such as
player skins and map packs). 644

7.377 A console gaming competitor [] submitted that Minecraft is not a close
analogue to CoD, including because (i) it is a single-release title that was
published in 2011, (ii) it is based on a legacy monetisation model of a one-
time fee, after which users receive lifetime updates and content, (iii) it is not
graphically intensive and offers a visual experience that relies on blocky and
pixelated visuals, and (iv) it does not drive anything close to the level of
gameplay, engagement, or purchasing decisions as CoD, having fewer
players, fewer MAUs, lower engagement levels, less than [] of CoD
gameplay hours, and limited impact on console purchasing decisions. 645
According to this competitor, the closest analogue to the current transaction is
Microsoft’s acquisition of ZeniMax. 646

7.378 We consider that Minecraft has some similarities to CoD (ie, it was an
established multiplayer game with multiple revenue streams) but that it also
has important differences. One important difference is Minecraft’s legacy
monetisation model. Before being acquired by Microsoft, Mojang Studios sold
Minecraft for a one-time fee, after which users received lifetime updates and
content. Following the acquisition, Microsoft continued to charge a one-time
fee (of around $30) and developed other ways of monetising the game,
including subscription payments for multi-player functionality, merchandise
from the Minecraft Shop (eg, t-shirts, lamps, mugs, etc), and game-enhancing
features from the Minecraft marketplace (eg, skin packs, texture packs,

641 Microsoft response to working papers.

642 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.55(c).
643 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 3.55(a). See also

Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.118-2.119

644 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.119.
645 [] submission to the CMA.
646 []; See also [].

184
 adventure maps, etc). Despite CoD having similar in-game monetisation
streams to Minecraft, this legacy monetisation model of a one-time fee for
lifetime access and updates, which predated Microsoft’s acquisition, differs
significantly from CoD, where users generally buy the new premium iteration
of the game every year for a higher fee.

7.379 With respect to other past acquisitions, the Parties submitted that, as of
October 2022, from the development studios Microsoft has acquired, there
are [] upcoming titles, [] of which are not planned to be available on
PlayStation or Nintendo. 647 Examples as of October 2022 include:

(a) [].

(b) [].

(c) [].

(d) [], 648 [].

7.380 Having reviewed Microsoft’s strategy following its previous acquisitions of

game studios and publishers, we consider that most of these were, in effect,
acquisitions of talent. The majority of studios that Microsoft has acquired (with
some notable exceptions including ZeniMax) did not have regular releases of
popular gaming franchises available on different platforms. As such, Microsoft
did not have to decide whether to make multiplatform games with a large
customer base exclusive to Xbox following these acquisitions; it acquired
those studios with the specific purpose of making exclusive games for its
platform.

7.381 The main exception is Microsoft’s 2021 acquisition of ZeniMax:

(a) Prior to the acquisition, 24 of the 26 games ZeniMax released on

PlayStation or Xbox consoles since 2010 were released on both
consoles. 649

(b) ZeniMax has published three games since its acquisition by Microsoft.
The most recent was Hi-Fi Rush, a single player action game which was
released exclusively on Xbox and PC in January 2023. Prior to this it
released Deathloop in September 2021 and Ghostwire: Tokyo in March

647 Microsoft Internal Document.

648 Wasteland 3 was released on Xbox, PlayStation and Windows in August 2020 and on Linux and Mac in
December 2020
649 'Wikipedia | List of Bethesda Softworks video games', accessed by the CMA on 23 January 2023. The two

games not published on both consoles were Quake III Arena Arcade released only on Xbox 360 and Fallout
Shelter released only on Xbox One, android and iOS.

185
 2022. 650 Both games were released under one-year exclusive contracts
with PlayStation which were agreed prior to the acquisition. Therefore, it
seems that Microsoft has honoured existing contractual agreements and
acted in accordance with the statements it made at the time of the
ZeniMax acquisition. 651 Deathloop has since been released on Xbox (and
Game Pass) in September 2022 with the addition of new content available
across all platforms.

(c) Microsoft and ZeniMax have confirmed that some future releases of
games from Zenimax studios will be Xbox exclusives (which includes a
Windows PC version). 652 Microsoft has also confirmed that all future
releases will be on Xbox Game Pass on release. 653 Microsoft has publicly
confirmed that new games Starfield and Redfall expected in 2023 will be
Xbox/PC exclusives. 654

(d) Microsoft Gaming’s CEO has publicly alluded to the next Elder Scrolls
(Elder Scrolls VI) release being an Xbox exclusive (which includes a
Windows PC version). 655 Both Starfield and Elder Scrolls: VI were
announced in 2018, before Microsoft’s acquisition, and were not
anticipated to be Xbox exclusives. 656

(e) ZeniMax also publishes Elder Scrolls Online, a live service massively
multiplayer online role-playing game within the Elder Scrolls franchise. It
was first released in 2014 but receives annual content releases. The
game continues to be available on Xbox, PC and PlayStation.

7.382 This strategy of acquiring talent to make new games that are exclusive to
Xbox is consistent with our finding that console providers place significant
value in having exclusive content to differentiate their platform and attract

650 Deathloop is a single player first-person shooter (FPS) game with a 2-person campaign mode. Ghostwire:
Tokyo is a single-player action-adventure game.
651 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 3.55(a)-(b).
652 We note that, in the context of the European Commission merger investigation in relation to the

Microsoft/ZeniMax acquisition whose report was published in March 2021, Microsoft submitted it had strong
incentives to continue making ZeniMax games available for rival consoles and their related storefronts (see Case
M.10001 – Microsoft / ZeniMax, paragraph 107). Microsoft’s decisions described above regarding new titles
Starfield and Redfall., and the suggestion of future exclusive releases in the Elder Scrolls franchise on Xbox,
which reveal its real-world incentives, suggest that static incentives analyses developed in the context of a
merger inquiry may fail to capture significant unstated commercial incentives.
653 ‘Microsoft News | Get The Facts: How Microsoft is Committed to Growing Gaming Communities’ accessed by

the CMA on 7 February 2023.

654 Starfield is a single player role-play game (RPG). Redfall is a FPS with a 4-person co-operative multiplayer

mode. ‘Gamerant | Xbox three exclusive Bethesda games explained’, accessed by the CMA on 19 January 2023.
See also https://www.xbox.com/en-US/games/starfield and https://www.xbox.com/en-GB/games/redfall.
655 Elder scrolls is a role-playing game (RPG); all releases in the franchise (except for Elder scrolls online) have

been single player and Elder Scrolls VI is expected be single player. Microsoft gaming’s CEO publicly alluded to
Elder Scrolls VI being an Xbox exclusive in an interview in November 2021. 'GQ magazine | xbox Phil Spencer
Todd Howard interview', accessed by the CMA on 30 January 2023.
656 ‘E3 2018: Bethesda’s Press Conference News Recap – Fallout 76, Elder Scrolls 6, Starfield, and More’,

accessed by the CMA on 5 January 2023.

186
 more users. Most first-party Xbox and PlayStation games are exclusive to
their respective platform, and almost every studio that Microsoft has bought
now makes games exclusive to Xbox. Moreover, where Microsoft has seen
value in making multiplatform third-party studio games exclusive to Xbox, it
has done so (eg, the upcoming release of their new game Redfall following
the ZeniMax acquisition).

7.383 We consider, however, that the financial and strategic calculation of creating
new exclusive games for Xbox may be different from that of making CoD
exclusive to Xbox: making an existing multiplatform gaming franchise
exclusive leads to losses (ie, lost revenues from customers on other
platforms), which have to be weighed against expected gains (ie, increased
revenues from new Xbox customers). Moreover, in this case, at least part of
CoD’s value comes from the size of its community of gamers, and that would
be eroded by removing it from PlayStation. By contrast, making new gaming
IP available on other platforms has some costs associated with it (eg,
optimizing for a different OS and possible diversion of users to a rival console)
and leads to uncertain gains. We consider that these differences could
reasonably lead Microsoft to make a different strategic decision in relation to
new releases of CoD than it has done in relation to entirely new games
generated after previous acquisitions, especially in light of the significant
quantifiable losses suggested by the revised LTV model.

7.384 Overall, our view is that the evidence arising from Microsoft’s past acquisitions
alone is inconclusive. It suggests that exclusivity matters, that Microsoft
strives to acquire and produce exclusive content, but that when it acquired
Minecraft, it had the incentive to keep it on PlayStation and Nintendo.

Contractual arrangements and negotiations

7.385 In this section, we assess the impact that the Merging Parties’ contractual
arrangements (and negotiations that might end up in contractual
arrangements) could have on the Merged Entity’s incentive to foreclose SIE.
As explained above, Activision has an existing contract with SIE which will
expire in the short term covering the arrangements between the two entities
with regards to the availability of CoD on PlayStation. Microsoft has also
made a separate offer to SIE to keep CoD on PlayStation for a period of ten
years post-Merger. 657 Microsoft also explained that it has entered into a
legally binding 10-year agreement with Nintendo to bring CoD to Nintendo
platforms post-Merger and an offer was made to Steam. 658 As this theory of

657 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 3.47 and 3.50(g). See also
Microsoft, letter to the CMA; and Microsoft, letter from to the CMA.
658 Microsoft, letter to the CMA and Microsoft response to the Provisional Findings, 2 March 2023, paragraph 1.2.

187
 harm is focused on SIE for the reasons already discussed, we have focussed
our assessment on any potential impact on the Merged Entity’s incentives
stemming from contractual arrangements and negotiations with SIE.

Parties’ and SIE’s views

7.386 Microsoft submitted that SIE’s existing contractual arrangements with

Activision, as well as Microsoft’s separate offer to SIE to keep CoD on
PlayStation post-Merger, limit the Merged Entity’s incentive to foreclose. 659

7.387 Microsoft submitted that the financial and reputational damage arising from
non-compliance with Activision’s contracts with SIE would be severe.
Microsoft explained that, given [] Activision’s existing contracts with SIE of
c. [] per annum, there would be costly penalties for breaching these
contracts. Microsoft also submitted that the reputational damage would be
even worse. 660

7.388 Microsoft referred the CMA to its acquisition of ZeniMax. Microsoft submitted
that, post-acquisition, it honoured ZeniMax’s existing contractual
commitments with SIE. 661 Microsoft also referred the CMA to public
comments made by SIE that it expects Microsoft to abide by existing
contractual commitments post-Merger. 662

7.389 SIE submitted that Microsoft’s past conduct shows that its ‘public utterances’
should be treated with ‘extreme scepticism’. 663 However, SIE did not suggest
that Microsoft has breached existing contractual arrangements following
previous acquisitions; instead, it referred to new releases made by particular
game studios post-acquisition. 664 This is consistent, for instance, with the fact
pattern outlined by Microsoft regarding the ZeniMax acquisition: Microsoft
states it has honoured existing contractual arrangements in place between
SIE and ZeniMax at the time of the acquisition. Microsoft explained that two
titles (Deathloop and Ghostwire: Tokyo) were released first on PlayStation
post-acquisition in accordance with the terms of the relevant contractual
arrangements. 665

659 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 3.58(a)-(d). See also
Microsoft, letter to the CMA and Microsoft, letter to the CMA.
660 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.58(d).
661 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.47.
662 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 3.47(c), 3.50(h).
663 SIE response to the phase 2 Issues Statement, 28 October 2022, paragraph 17.
664 SIE response to the phase 2 Issues Statement, 28 October 2022, paragraph 17.
665 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.47. See also Microsoft

response to the working papers.

188
7.390 Regarding the agreement with Nintendo, Microsoft submitted that its 10-year
agreement with Nintendo demonstrates its lack of incentive to foreclose by
showing a general intention to distribute CoD on more consoles, not fewer,
post-Merger, as well as the commercial reality that Microsoft is looking to
monetise CoD as much as possible and, as a result, its clear incentive is to
expand access to the franchise. 666

Our assessment

7.391 As there is currently no agreement in place between Microsoft and SIE in

relation to Activision content, we do not consider that the negotiations referred
to above have any impact on our assessment of the Merged Entity’s incentive
to engage in foreclosure.

7.392 As regards the existing agreement between Activision and SIE, the MAGs
make clear that the CMA may consider any financial or reputational costs of
terminating contracts in its assessment of foreclosure incentives. 667 The
question of reputational impacts and the financial implications of foreclosing
CoD from SIE has already been incorporated into our assessment above.
More generally, we also consider that the points discussed above in relation
to ability to foreclose are relevant to the CMA’s assessment on the Merged
Entity’s incentives to engage in foreclosure in this market. In particular, we
note that while the existing contractual arrangements between SIE and
Activision provide SIE with some protection in the short-term, the relevant
protections are of limited duration. We do not consider they are of sufficient
duration to have a material impact on our competitive assessment, which
considers potential concerns with a longer time horizon.

7.393 Regarding the Nintendo agreement, in addition to noting that this theory of
harm is primarily focussed on SIE for reasons already explained, we also
consider the points discussed in the ability assessment regarding the
uncertainty created by certain terms of this agreement also apply to our
incentive analysis. With regards to Microsoft’s submission that this agreement
demonstrates a general intention to distribute CoD on more consoles, we note
that this agreement has been entered into during the course of our Merger
investigation (and those of other authorities). We therefore do not consider
this is reliable evidence of what Microsoft’s incentives would otherwise be in
the ordinary course.

666 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 2.74 and 2.115.
667 CMA129, paragraph 7.15.

189
7.394 As such, our view is that Microsoft’s contractual arrangements are not likely to
have any significant impact on its incentive to foreclose SIE, at least beyond
the short term once the existing contract with SIE comes to an end.

Conclusion

7.395 We assessed whether a total foreclosure strategy would be profitable and,

therefore, whether the Merged Entity would have an incentive to engage in
such strategy. As part of this assessment, we considered a mix of both
qualitative and quantitative evidence.

7.396 Our quantitative modelling indicates that the Merged Entity would face
substantial losses in case of total foreclosure. Given the significant losses that
this model suggests the Merged Entity would incur, under all plausible
scenarios, we place significant weight on this evidence when assessing it in
the round, together with Microsoft’s behaviour following previous acquisitions
and its longer-term strategic objectives.

7.397 Our assessment also shows that Microsoft has acquired a range of gaming
studios and, with very few exceptions, has redirected the efforts of those
studios to produce exclusive Xbox games. We consider that this evidence is
not sufficiently conclusive to suggest that Microsoft would have an incentive to
foreclose PlayStation. This is because most of the studios that Microsoft has
acquired did not have regular releases of popular gaming franchises available
on different platforms; therefore, Microsoft did not have to decide whether to
make multiplatform games with a large customer base exclusive to Xbox
following these acquisitions. Although Minecraft has some similarities to CoD,
we do not consider it to be sufficiently similar to provide strong evidence that
Microsoft’s strategy with CoD would be the same as its strategy with Minecraft
insofar as making it available on PlayStation post-Merger.

7.398 We have also seen that there are longer-term strategic benefits that are
difficult to quantify, but which nonetheless contribute to Microsoft’s incentive
to engage in foreclosure strategies to some extent. These include Microsoft’s
plans to grow Game Pass, which Microsoft’s own valuation model estimates
to be significant. Longer-term strategic benefits also include the potential to
acquire new loyal customers, and the reputational benefits from owning a
console with exclusive CoD content. We have not found evidence to suggest
that these strategic benefits are of a scale that they would mitigate the
significant losses that our LTV model suggests Microsoft would incur as a
result of a total foreclosure strategy.

7.399 Based on our assessment above, and placing particular weight on the
significant losses that our LTV model suggests Microsoft would incur if it were

190
 to engage in a total foreclosure strategy, we conclude that the Merged Entity
would not have the incentive to engage in a total foreclosure strategy using
CoD.

7.400 We have, therefore, concluded that the Merged Entity will not have the
incentive to engage in total foreclosure of PlayStation in the market for
console gaming services in the UK.

Effect of foreclosure on competition

7.401 In light of our conclusions that (i) the Merged Entity will not have the ability to
engage in partial foreclosure strategies vis-à-vis PlayStation, and (ii) that the
Merged Entity would not have the incentive to engage in total foreclosure, it is
not necessary for us to conclude on the effects of foreclosure on competition
in the supply of console gaming services in the UK.

Conclusion on TOH1

7.402 We conclude that the Merger may not be expected to result in a substantial
lessening of competition in the market for the supply of console gaming
services in the UK.

191
8. Theory of harm 2: Vertical effects in cloud gaming
services

Framework for assessment

8.1 The concern under this input foreclosure theory of harm is that the Merger
may lead to the Merged Entity using Activision’s games to foreclose cloud
gaming service rivals, harming their ability to compete.

8.2 In relation to Activision’s content, we consider whether the Merged Entity

could harm its rivals’ competitiveness and thus lessen current and future
competition in cloud gaming services through total foreclosure, ie making
either current and/or future Activision content unavailable on rival cloud
gaming services (ie exclusive to Game Pass). 668

8.3 In our assessment of whether the Merged Entity may harm rival cloud gaming
services’ ability to compete by denying or worsening its access to Activision’s
games, we follow the framework set out in the Merger Assessment Guidelines
for assessing input foreclosure theories of harm. 669 We therefore consider
whether three cumulative conditions are satisfied: 670

(a) Would the Merged Entity have the ability to use its control of content to
harm the competitiveness of rival cloud gaming services?

(b) Would it have the incentive to actually do so, ie would it be profitable?

(c) Effects of foreclosure: would the foreclosure of rival cloud gaming

services substantially lessen overall competition between cloud gaming
services?

668 During our investigation we also considered whether the Merged Entity could harm its rivals’ competitiveness
and thus lessen current and future competition in cloud gaming services through partial foreclosure strategies, in
particular by making Activision content available for release on rival cloud gaming services at a later day
compared to Xbox (ie timed exclusivity). However, in circumstances where Activision content is not currently
available on rival cloud gaming services, and – for reasons explained in this Chapter, we consider the Merged
Entity would have the ability and incentive not to make it available to rival cloud gaming services post-Merger –
we have not considered it necessary to assess this further. To the extent timed exclusivity was longer-term such
that it was essentially akin to total foreclosure, this would be covered by the analysis in this Chapter more
generally. Other partial foreclosure strategies may include i) degrading the technical quality of Activision gaming
content available to cloud gaming rivals (eg content having worse graphics, worse latency, lower frame per
second, longer load times; ii) making features or upgrades of Activision games unavailable to other cloud gaming
rivals (ie content exclusivity); and/or iii) raising the wholesale price of Activision content to cloud gaming rivals. It
has not been necessary to assess the extent to which Microsoft may have the ability and incentive to pursue any
such strategies in light of our findings in relation to total foreclosure.
669 CMA129, paragraphs 7.9-7.22.
670 CMA129, paragraph 7.9.

192
8.4 An important question in this context is the likely development of this market,
and the Parties’ likely position within it, absent the Merger. We have therefore
assessed the long-term viability of cloud gaming services, []. 671

8.5 Microsoft has a wide range of products and services, some of which may be
important assets or inputs into cloud gaming. These include the Windows and
Xbox operating systems, its Azure cloud infrastructure, and its Xbox gaming
library. Any ability on Microsoft’s part to use these assets to foreclose rivals
does not form part of this theory of harm, as Microsoft already benefits from
these assets and inputs without the Merger. However, as detailed below,
these assets and inputs may nonetheless affect Microsoft’s incentive to
engage in a foreclosure strategy using Activision’s content, and they may also
magnify the effect of any such foreclosure strategy in the market for cloud
gaming services. We have therefore carried out an assessment of Microsoft’s
pre-existing strengths in cloud gaming services and assessed their impact on
its incentives to foreclose rivals and the effects of any such foreclosure:

(a) In relation to incentive, Microsoft would have a greater incentive to

foreclose rivals the stronger it is downstream, as it will capture more of the
sales from any foreclosed rival (and we note that it is already strong given
its pre-existing strengths in cloud gaming services). As noted in the
MAGs, incentives will be greater if the Merged Entity has a more
successful downstream offering. 672

(b) In relation to effect, as noted in the MAGs, competition concerns may be

particularly likely to arise if a merger firm has a degree of pre-existing
market power in the downstream market, and already faces limited
competitive constraints pre-merger. 673

8.6 We are also mindful that Activision content has only been available on cloud
gaming services to a limited extent thus far. We have therefore considered the
extent to which Activision would have made its content available on cloud
gaming services absent the Merger.

8.7 This chapter is structured as follows:

(a) First, we assess the likely future development and long-term viability of
cloud gaming services.

(b) Second, we assess the competitive landscape in cloud gaming services.

671 Microsoft response to working papers.

672 CMA129, paragraph 7.19 (b).
673 CMA129, paragraph 7.21.

193
 (c) Third, we assess Microsoft’s pre-existing strengths in cloud gaming
services, focusing on the Windows and Xbox OSs, Azure, and its Xbox
gaming library.

(d) Fourth, we assess Microsoft’s overall strengths in cloud gaming services

relative to rivals.

(e) Fifth, we assess the extent to which Activision content would have been
available on cloud gaming services absent the Merger.

(f) Sixth, we assess whether the Merged Entity would have the ability to
foreclose rival cloud gaming services as a result of the Merger.

(g) Seventh, we assess whether the Merged Entity would have the incentive
to foreclose rival cloud gaming services.

(h) Finally, we assess the effect that any foreclosure strategy would have on
competition in the market for cloud gaming services.

8.8 As set out in the MAGs, in the context of sectors that are characterised by
fast-moving technological and commercial developments or assessments of
potential or dynamic effects that are particularly dependent on the evolution of
competitive conditions, the types of evidence that are available to the CMA
may be more restricted. In such cases, the CMA may place particular weight
on evidence such as internal documents, the expected number of competitors
after the merger, similarities between the characteristics of the products or
services that are under development, and the views and expansion plans of
market participants. As with uncertainty, the absence of certain specific types
of evidence such as historical data will not in itself preclude the CMA from
concluding that the SLC test is met on the basis of all the other available
evidence assessed in the round. 674

Future development of cloud gaming

8.9 This section considers evidence on how the market for cloud gaming services
will develop in the future, and Microsoft’s plans for its cloud gaming service.

8.10 As set out in Chapter 5, we consider that there is a single market for cloud
gaming services. We recognise, however, that cloud gaming service providers
are testing a range of different business models to monetise their service.
They all currently offer a subscription-based model for access to their servers
(and some offer a free tier with advertising for this purpose). Different services

674 CMA129, paragraph 2.28.

194
 monetise their gaming content in different ways, including (i) the traditional
B2P model, whereby users must purchase a game through the cloud service
in order to play it (eg, Google Stadia), (ii) the BYOG model, where users can
play games bought in third party storefronts (eg, NVIDIA GeForceNow), (iii)
free-to-play offerings paid through advertising revenue and in-game
purchases, and (iv) multi-game subscription services, whereby users pay a
subscription fee for access to gaming servers and a catalogue of games (eg,
Amazon Luna and Xbox Game Pass Ultimate). The evidence shows that
these business models are not fixed, and that cloud gaming service providers
are open to exploring different ways of monetising their services.

8.11 Under this theory of harm, we explore competition concerns in the market for
cloud gaming services as a whole. Given our findings set out below that
Activision’s content is most likely to become available on cloud gaming
services under a B2P or BYOG approach, we pay particular attention to the
potential impact of the Merger on these services. We note, however, that
industry participants in this growing industry are continuously experimenting
with different payment models, and even a single participant may have more
than one way of monetising content. For example, Amazon has implemented
a BYOG option in Luna for Ubisoft games, and NVIDIA [] on GFN. 675

8.12 As such, we believe that any competition concerns arising from the Merger
would affect the market as a whole, rather than just existing participants
following a B2P or BYOG approach.

Parties’ views

8.13 Microsoft submitted the following on cloud gaming:

(a) It stated that cloud gaming is currently small. According to Microsoft, it

accounted for a de minimis proportion of consumer spend on gaming in
2021 – just []%. 676 Microsoft expects that it will account for only []%
of total consumer spend on gaming by 2025. 677 It submitted that on PC
specifically cloud gaming is also de minimis with cloud gaming users
accounting for less than []% of PC gamers in the UK. 678

(b) It submitted that consumer adoption was uncertain and expected to be

limited, 679 and that downloading games was likely to remain the most
popular option for playing games on PC and console. Microsoft noted that

675 ‘Amazon Luna Ubisoft Store', accessed by the CMA on 17 January 2022; and [] call note.
676 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.3.
677 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.8.
678 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.32.
679 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.3.

195
 this was in part because cloud gaming requires the user to be connected
to the internet and gameplay can be affected by latency. 680

(c) It submitted that cloud gaming on mobile devices had been unsuccessful,
citing the performance of Fortnite on xCloud as an example of difficulty in
attracting and retaining gamers. 681 It also submitted that with the
increasing computational power of mobile devices, many gaming
companies are increasingly developing native mobile games, meaning
there is unlikely to be material demand for cloud gaming on mobile
devices (which Microsoft also submitted depends on the user having a
stable internet connection). 682

(d) It submitted that it did []. It submitted that this had been demonstrated
by its experience with streaming Fortnite, and provided data purporting to
show that []. 683 Microsoft provided further data that suggested that
[]. 684

(e) It submitted that it considered using []. 685

(f) It explained that, as a result of these issues, it had []. 686 []. 687 It
submitted that []. It submitted that it []. 688

Our assessment

Internal document evidence

8.14 This section considers the evidence from Microsoft’s internal documents that
discuss its plans and expectations for cloud gaming. The evidence is
presented in chronological order to track the development of Microsoft’s view
of cloud gaming services over time.

8.15 Internal documents from [] suggest that Microsoft is optimistic about cloud
gaming and committed to bolstering its already strong position in the market
[]. These documents, set out in more detail below, include views from
senior Microsoft employees such as []. 689

680 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 5.4-5.6.
681 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 5.10-5.13.
682 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.31.
683 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 5.14-5.15.
684 Microsoft site visit.
685 Microsoft response to the CMA’s s109 notice.
686 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.16.
687 Microsoft, response to working paper.
688 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.37.
689 Microsoft Internal Document; Microsoft Internal Document; Microsoft Internal Document.

196
8.16 A strategy document from October 2019, prepared by the Xbox gaming team
for Microsoft’s senior leadership team, states that ‘cloud streaming’ represents
an []. The document also notes that the transition to a []. 690

8.17 In an email exchange from April 2020, [] discuss the possibility of putting
[]. 691

8.18 A May 2020 presentation sets out a number of projects that would support a
long-term strategy to support expansion in cloud streaming. In particular, the
document defines []. The document describes []. 692

8.19 An email from [] in December 2020 states that []. 693 In the same email
chain [] states: []

8.20 A draft document from May 2021 discusses []. It describes []. It suggests
that []. 694

8.21 An email from [] in December 2021 []. 695

8.22 In an email from February 2022, [] argues that []. 696 []. In this email
[] also notes [].

8.23 Another email from February 2022 provides an update on []. 697 Microsoft
submitted that this project []. It has publicly stated that it still expects to
produce a streaming device at some point, but that this could be ‘years
away’. 698

8.24 An email from [] in March 2022 notes []. 699

8.25 Emails from [] in May 2022 discuss []. In the same email chain []
describes []. 700

8.26 In relation to Microsoft’s argument that [], emails from [] to [] in July
2022 discussing []. 701

690 Microsoft Internal Document.

691 Microsoft Internal Document.
692 Microsoft Internal Document.
693 Microsoft Internal Document.
694 Microsoft Internal Document.
695 Microsoft Internal Document.
696 Microsoft Internal Document.
697 Microsoft Internal Document.
698 Microsoft, response to working papers.
699 Microsoft Internal Document.
700 Microsoft Internal Document.
701 Microsoft Internal Document.

197
8.27 In submissions to the CMA during these proceedings, Microsoft has submitted
that this analysis had been prepared relatively early []. 702 The analysis [].
Microsoft also described []. In this respect we make the following
observations:

(a) The Parties have not pointed to any internal documents generated in the
ordinary course of business that update these analyses and show [].

(b) Other games are likely to be [].

(c) Whilst different assumptions would have an impact on [], the analysis
nevertheless broadly demonstrates that [].

(d) A more recent internal document from September 2022, providing an

email update on [], describes how []. This indicates that Fortnite has
[] on the basis of more recent data. 703

8.28 Microsoft also provided data showing that []. 704 As such, Microsoft
considers that []. 705 However, we note that this data involves []. It was
also based on []. In addition, Microsoft highlighted issues with
discoverability on mobile due to app store restrictions which limited user
uptake. 706 This means that the experience [] is likely to underestimate
consumer uptake and the viability of cloud gaming services outside the mobile
ecosystem.

8.29 Slides from May 2022 provide an update on xCloud quarterly results, showing
MAUs of [] in Q3, with year-on-year growth in MAUs of []%, []. []% of
MAUs were NTX, which was []%. It is also noted in the slides that []. 707

8.30 Emails from [] in July/August 2022, primarily discussing opportunities in VR,
note the opportunities and Microsoft’s strength in cloud. In the emails he
states that he sees []. He also notes that in relation to Microsoft’s game
development studios, [] 708

8.31 Slides from July 2022 presenting analysis on XGP [] describe [], and
that, according to a survey, []. The slides also show that survey
respondents were []. 709

702 Microsoft response to working papers.

703 Microsoft Internal Document.
704 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 5.14-5.15.
705 Microsoft site visit.
706 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.13e.
707 Microsoft Internal Document.
708 Microsoft Internal Document.
709 Microsoft Internal Document.

198
8.32 Slides from August 2022 provide an update on []. 710

8.33 A July 2022 email exchange between [] describes []. It also notes that
[]. 711

8.34 An internal chat in July 2022 describes how [].’ 712 This suggests that [].

8.35 An email from 4 August 2022 that discusses []. The same document also
describes that [] 713

8.36 An email from 11 August 2022 to the Xbox Everywhere [] further sets out
[]. For example, []. It describes []. It also describes that []. 714

8.37 Microsoft provided a chart showing that, []. In February 2022 it forecast
approximately [] xCloud MAUs by []. In September 2022, this forecast
[], which []. 715 We note, []. It may also be expected to [].

8.38 Microsoft highlighted further internal documents from April to July 2022, which
[]: 716

(a) An email exchange from April 2022 between [] and [], both of whom
work in Xbox marketing, 717 []. It notes that the []. However, the same
email also states that []. 718

(b) An internal message from May 2022 from [] highlights []. It notes
that, []. It states that that []. 719

(c) An internal chat from July 2022 between [] 720 suggests that []. In
relation to the [] discussed above, it []. 721

(d) Another internal chat from July 2022 between the same people highlights
[]. It describes how []. It also states that []. 722

8.39 Despite these views and [], more recent internal documents continue to
discuss []. For example:

710 Microsoft Internal Document.

711 Microsoft Internal Document.
712 Microsoft Internal Document.
713 Microsoft Internal Document.
714 Microsoft Internal Document.
715 Microsoft response to the CMA’s s109 notice.
716 Microsoft response to working paper.
717 Microsoft Internal Document.
718 Microsoft Internal Document.
719 Microsoft Internal Document.
720 Microsoft Internal Document.
721 Microsoft Internal Document.
722 Microsoft Internal Document.

199
 (a) Slides from August 2022 describing FY23 plans for [] note that [].
However, []. 723

(b) Emails from September 2022 discuss []. 724

(c) The September 2022 email to [] sharing [] referenced above also
notes that Microsoft is continuing its []. 725

(d) A document from September 2022 discusses in detail []. 726 Microsoft
submitted that this document represents []. However, it discusses [].

(e) A memo from September 2022 describing priorities for Fiscal Year 2023
sets out how []. The new fiscal year is described []. It describes
[]. 727

(f) Two documents dated September 2022 and November 2022 discussing
plans for next generation consoles show that []. One slide describes
[]. Another slide discusses []. 728

8.40 In our view, the overarching impression from Microsoft’s internal documents is
that cloud gaming is expected to be a growing area in gaming []. Whilst
these documents highlight that [] we do not consider that they suggest that
cloud gaming services will stop growing. At most, we consider that these
documents show []. As described above, more recent documents continue
to discuss [].

Third party evidence

8.41 Evidence from rival cloud gaming providers, including from their internal
documents, shows that they expect the market to grow and be profitable.
Internal documents from one rival [] from October 2021 []. The document
suggests that this rival []. 729

8.42 Microsoft submitted that []. 730 Whilst it is difficult to directly []. Despite this
[] 731 and []. 732

723 Microsoft Internal Document.

724 Microsoft Internal Document.
725 Microsoft Internal Document.
726 Microsoft Internal Document.
727 Microsoft Internal Document.
728 Microsoft Internal Document; and Microsoft Internal Document.
729 [] Internal Document.
730 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.33 and 3.34.
731 [] response to the CMA’s RFI.
732 [].

200
8.43 Another rival [] submitted that it expected that many users would switch to
cloud gaming. 733 It said that it expected cloud gaming could and would be
delivered profitably, and that getting the right content to attract paid users was
the main barrier to succeeding in this market, rather than any technical or
economic aspects of cloud gaming. 734 An internal document from this provider
shows that []. 735 Whilst Microsoft submitted that these [], 736 in our view
[].

8.44 The same rival [] submitted that technological barriers to streaming,
including latency, were quickly dropping and were likely to continue to drop. It
described how it had reduced latency on its service []. 737 On a call the
same rival stated that it had solved the latency problem, and that its service
now outperformed playing on a local console device including from a latency
perspective, which it had achieved by using more powerful graphical
processing units (GPUs). 738

8.45 Evidence from this rival [] on play time for different games showed that
multiplayer and ‘fast-twitch’ games 739 are amongst the most popular,
indicating that latency does not present a problem for streaming these types
of games. []the [] most played games on the platform included [], all of
which include multi-player and can be considered fast-twitch. 740

8.46 Another rival [] expressed that it expected cloud gaming to increase
substantially and ultimately replace consoles, stating that cost and
technological issues such as latency could be overcome to provide an
‘amazing gaming experience’. 741

8.47 Another provider [] submitted that it had reached profitability in 2022 having
started operating in 2019, although this excludes hardware expenses. It
stated that it has high capital expenditure due to hardware investments, and
that a hardware solution with efficient balance between cost and performance
is key to profitability in cloud gaming. 742 This provider also stated that cloud
gaming will be the main way users access gaming content in 7-10 years. 743

733 [] response to the CMA’s RFI.

734 [] response to the CMA’s RFI.
735 []Internal Document; and [] Internal Document.
736 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.34.
737 [] response to the CMA’s RFI.
738 [] call note.
739 Games that require fast response times such as first-person shooters.
740 [] Internal Document.
741 [] call note.
742 [] response to the CMA’s RFI.
743 [] response to the CMA’s RFI.

201
8.48 A potential entrant [] stated that there were challenges in cloud gaming, but
that it was only a matter of time before they were solved, and cloud gaming
really takes off. 744 The same company stated that ‘fast-twitch’ multiplayer
games such as Call of Duty or Fortnite are suitable for cloud gaming services
due to advancements in home networking performance, and that there may
be some benefit to playing these types of games through a cloud gaming
service because the high performance and reliability of cloud networking can
increase the scalability (eg, number of players) in a multi-player game. It also
stated that it had tested such games on its service and believes it offers
competitive performance. 745

8.49 A competitor [] described that GPUs are increasing in cost as the cost of
transistors and required number of transistors are increasing, and that the
benefit of cloud gaming is that it allows consumers to access the latest
gaming hardware at a lower cost by providing a shared computing
environment. It further submitted that in the future some games might only be
able to run in the cloud due to the ability to have large memory and
increasingly large GPUs. 746

8.50 A third party publisher [] indicated that it did not expect cloud gaming to
replace console in the near future, but that it would become an alternative for
some consumers. It noted that it has not yet reached mass adoption, with one
of the main reasons being that it can still be associated with ‘lag’ or latency. It
described how to replace console, cloud gaming services needed to prove
that the latency question has been addressed, and that internet coverage and
data plans need to improve. It stated that it is interested in developing games
for cloud gaming services where the 'quality of service is there’ with respect to
eg latency and bandwidth, []. 747

8.51 Another third party publisher [] stated that it thought it likely that cloud
gaming can support a transition away from PC and console gaming, although
noting that is still an emerging technology and its development is associated
with uncertainties, and it is therefore difficult to estimate approximate
timescales. It described the main challenges for such a transition as mainly
technical, including the requirement for low latency. 748

8.52 Other major third party publishers also expressed opinions about the future
development of cloud gaming.

744 [] call note.

745 [] response to the CMA’s RFI.
746 [] call note.
747 [] response to the CMA’s RFI.
748 [] response to the CMA’s RFI.

202
 (a) [] noted that cloud gaming is a developing technology and that if it
continues to develop it will likely further increase the competitive nature of
game development and benefit consumers unable to purchase the
hardware for console or PC gaming. It also noted the need for stable,
high-speed internet access. 749

(b) [] stated that cloud gaming is still nascent, and cloud gaming service
providers are currently in the early adopter stage. It noted that in deciding
whether to publish on a cloud gaming service it would, among other
things, evaluate whether it has the capacity to provide users with a good
gaming experience. It also noted that it has published games on GFN and
xCloud. 750

(c) [] described cloud gaming as still nascent. It stated that it would
consider the in-game player experience including latency when evaluating
opportunities to publish a game on a cloud gaming service, and has
published a select number of games on GFN, xCloud and Stadia. 751 It
also stated that it thinks cloud gaming offers one possible route to device
agnostic gaming in the future. 752

(d) [] stated that is likely that cloud gaming services will grow especially in
markets with free fast internet access and low console penetration. It
noted that in the UK ‘machine gaming’ (ie on console or PC) is most
popular as there is no latency. It described cloud gaming as being early in
its life cycle, and that as a rough guess it could be 10-15 years before
cloud gaming replaces consoles. 753

(e) [] stated that it did not think it was likely that cloud gaming would
overtake console gaming in the next five years due to latency
concerns. 754 It however stated that it did anticipate that cloud gaming will
be a viable alternative to native devices in major markets within five
years. 755

8.53 Some cloud infrastructure providers also expressed views on the future
development of cloud gaming.

749 [] response to the CMA’s RFI.

750 [] response to the CMA’s RFI.
751 [] response to the CMA’s s109 notice.
752 [] call note.
753 [] response to the CMA’s RFI.
754 [] response to the CMA’s RFI.
755 [] response to the CMA questionnaire.

203
 (a) [] suggested that the main developments are likely to be a continued
move to console-less games which are increasingly becoming cloud-
based. It noted that this will likely lead to increased cloud demand. 756

(b) [] stated that the closing of Google Stadia will ‘leave a sizable hole not
only in market share but also in regard to performance’ and noted that the
market is ‘searching for someone else’ as Microsoft and SIE do not wholly
focus on cloud gaming. It stated that whilst many believe cloud gaming is
the future, until cloud infrastructure and business models catch up, similar
failures of cloud gaming services will continue to happen. 757

8.54 Several market analyst reports demonstrate an expectation that the cloud
gaming services market will grow, with estimates ranging from $6.1 to $11.4
billion by 2025, and $11.9 to $13.5 billion by 2026, but with differing views on
the extent to which it will replace consoles.

(a) One recent industry market report from [] estimates that in 2021 there
were 21.7 million paying users of cloud gaming services spending $1.5
billion. It forecasts an increase of almost triple in terms of user base (to
58.6 million paying users) and more than quadruple in terms of spending
(to $6.3 billion) in 2024, according to the most-likely scenario. 758

(b) Another industry market report from [] in 2021 indicates that cloud
gaming’s share of total consumer spend on games is expected to double,
from 2.2% in 2021 to 5.3% in 2025, from $3.7 to $10.0 billion revenue. It
is expected to reach $11.9 billion by 2026. 759

(c) Another estimate from [] from 2021 forecasts cloud gaming to reach
101 million revenue-driving users by 2025, with $6.1 billion of revenue. 760

(d) A report from [] in 2021 predicts that over the next five years cloud
gaming revenue is expected to grow to $13.5 billion and account for over
5% of revenue in gaming (by 2026). Revenue is expected to be $11.4
billion by 2025. It expects that cloud gaming will not necessarily grow the
market for ‘fast-twitch’ action games, but instead grow based on new
forms of high production content. 761

756 [] response to the CMA’s RFI.

757 [] response to the CMA’s RFI.
758 [] Internal Document.
759 [] Internal Document.
760 [] Internal Document.
761 [] Internal Document.

204
 (e) A 2021 report from Oppenheimer describes how []. It also described
[]. 762

(f) A report from 2022 from Enders predicts that []. 763

8.55 None of the market analyst reports referenced above provide estimates for
the size of the UK market specifically. Based on the shares of supply analysis
described below, in 2022 the UK represented 10% of global MAUs of paid
cloud gaming services. To estimate the future size of the UK market we have
applied this proportion to the above forecasts for the size of the global market
for cloud gaming services. Using this approach suggests that the UK market
for cloud gaming services will be worth $0.6 to $1.1 billion by 2025, and $1.2
to $1.3 billion by 2026.

8.56 Data on cloud gaming MAUs gathered for the shares of supply analysis also
shows that the number of UK cloud gaming MAUs tripled between the start of
2021 and the end of the 2022, demonstrating substantial growth in the UKs
use of cloud gaming services.

Conclusion on expectations for cloud gaming

8.57 Based on this evidence, we consider that cloud gaming will continue to grow
and is likely to become profitable in the next five years. Although it is difficult
to predict exactly how big cloud gaming will eventually become, the evidence
supports the conclusion that it is a growing and promising market in which
several market participants are investing considerable amounts.

8.58 We consider that []. Most of Microsoft’s internal documents show that its
[] and there are indications in these documents that cloud gaming could be
transformative for the gaming industry []. As such, we believe it is likely that
[].

8.59 There is also clear consensus from third party respondents that cloud gaming
users and revenue will increase substantially in the next few years. Although
current revenues compared to streaming costs present challenges for
profitability, responses indicate that this will change as demand grows and
costs are driven down. We also note that short-run loss-making is not
uncommon in nascent technology markets, and that this does not preclude
their viability as the technology and market develop.

762 Microsoft Internal Document.

763 Microsoft Internal Document.

205
8.60 With respect to the Parties’ argument that the increasing computational power
of mobile devices will limit the growth of cloud gaming, we note that, whilst the
computational power of mobile devices has increased and may continue to
increase, cloud infrastructure and hardware will continue to be more powerful
with more memory, and therefore offer higher performance gaming. In the
same way that there will continue to be demand for consoles and gaming PCs
which are able to offer higher performance gaming than smaller, handheld
devices, there is likely to continue to be demand for cloud gaming which can
also offer higher performance from hardware housed in data centres. Playing
via cloud also offers additional benefits such as using less battery life as less
processing is happening locally.

Competitive Landscape in cloud gaming services

Features of the cloud gaming market

Network effects in cloud gaming

8.61 This section considers the extent of network effects in cloud gaming services.
Significant network effects can be a barrier to entry or expansion in a market.
In addition, the larger the network effects are in cloud gaming, the more likely
it is that the market could ‘tip’, ie, reach a position where only one or a few
providers are able to compete. 764 In a market with significant network effects,
the effect of any foreclosure strategy can therefore be increased. As part of
our assessment of network effects we also look at evidence on multi-homing,
as this may mitigate the strength of network effects to some extent.

Parties’ views

8.62 Microsoft submitted that cloud gaming reduces network effects. It stated that,
as cloud gaming is effectively device agnostic and does not require the
consumer to invest in hardware, gamers can easily switch and multi-home
across services, reducing network effects. It cited Ofcom evidence that in the
video streaming market many consumers multi-home, thereby reducing the
importance of network effects (as smaller platforms can still attract content
creators to distribute on their services) and decreasing barriers to entry. 765 It
also stated that direct network effects operate at the game level rather than
the platform level. 766

764 CMA129, paragraph 4.25.

765 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 5.51-5.52.
766 Microsoft response to working papers.

206
 Our assessment

• Evidence from ToH1

8.63 As described above, there is evidence that direct network effects (ie the
benefit to gamers of there being more other gamers on a platform) are
prevalent in the gaming industry, at both a game-level and platform-level, and
work similarly across different platforms. There is evidence that network
effects are stronger for large multiplayer social franchises like CoD and WoW,
with gamers wanting to play with their network of friends. This suggests that
direct network effects are likely to be similarly prevalent in cloud gaming,
though mitigated to some extent by multi-homing and switching being
relatively easier than in console gaming (described below). As cross-play and
cross-progression are implemented at the game level, they are available on
cloud platforms at similar levels to console, reducing the strength of direct
network effects at the platform level.

8.64 As described above, there is also evidence that indirect network effects (ie the
benefit to publishers of there being more gamers on a platform, and vice-
versa) exist in gaming. In principle, indirect network effects should be more
limited in this market. Unlike gaming consoles, which operate with exclusive
and proprietary console OSs, cloud gaming service providers predominantly
use OSs designed for PCs. If they decide to use Windows, which already has
the largest catalogue of compatible games, they would be able technically to
run all Windows games on their service. They would, however, have to pay
Windows licensing fees, and they would still have to reach an agreement with
the relevant game publishers for the right to run the relevant games on the
cloud gaming service. If they decide to use Linux, the availability of games
would be significantly more limited. Although the availability of Windows
(albeit for a licensing fee) reduces the impact of indirect network effects to
some extent, we have seen evidence (see below) that game publishers are
also more likely to license their content to cloud gaming services with a
significant user base, meaning that indirect network effects are still relevant
for individual cloud gaming platforms in this market.

• Internal documents

8.65 An internal email from [] at Activision in February 2021, discussing its
approach to [], notes that there are currently few [] noting that []. 767

767 Activision Internal Document.

207
 We consider this email shows that the number of users on a platform is a
factor in attracting content.

• Third party evidence

8.66 On direct network effects, one competitor [] described their importance but
pointed to ways of mitigating them. It described how consumers look at which
platform other users choose as this is important to ensuring full enjoyment of
multiplayer games. It noted that cross-play was a solution to this, and that
cross-play was therefore extremely important for new cloud gaming platforms
with small user bases, but that this option was not always available. 768

8.67 Another competitor [] also submitted that direct network effects are
important. It described that games have become social media platforms and
that gamers play games with their friends, creating game-specific social
networks, and increasing the cost of switching to another game even within
the same genre. On cross-play it noted that the ability to play across cloud
gaming platforms was not universal and dependent on how each platform is
structured. 769

8.68 On indirect network effects, one competitor [] also noted their existence and
importance for developers. It described how a large user base is necessary to
attract developers to release games on a platform, especially for games that
primarily generate revenue from in-game transactions which it described as a
model that is growing in popularity. It further stated that game developers
want assurance that their games will reach a certain group and number of
users, and that if a cloud gaming platform already has an established user
base, it would therefore be easier to reach an agreement with a game
developer to license their games. It also stated that having a large number of
users could increase a cloud gaming provider’s bargaining power with
developers to put games on its service and negotiate exclusivity. 770

8.69 In discussing commercial negotiations between a competitor [] and

Activision, this competitor [] noted that [] given its [] size. 771 This
competitor [] noted that when thinking [] 772

8.70 An internal document from the same competitor dated April 2021 further
demonstrates the significance of indirect network effects, with a chart showing

768 [] response to the CMA’s RFI.

769 [] response to the CMA’s RFI.
770 [] response to the CMA’s RFI.
771 [] call note.
772 [] call note.

208
 a cycle of getting more games, leading to having more engaged gamers, and
therefore better relationships with publishers. 773

8.71 Another competitor [] also submitted that indirect network effects were
important. It noted that to publishers, the number of users on a service is key,
and that without sufficient users the cost of maintaining a game on a service
may not be worthwhile, particularly if the service requires the publisher to port
and update the game on a new OS. This competitor noted ‘without a sufficient
number of users, the administrative burden of maintaining a game on a
particular cloud gaming service may not be worth the publisher’s effort.’ It
stated that []. 774 This competitor explained that it understood []. 775

8.72 An internal document from another competitor [] dated August 2020 []. 776
[].

8.73 Responses from several publishers indicated that they aimed to make games
available on as many platforms as possible subject to considerations such as
user experience and cost:

(a) One publisher [] described how it generally tries to make its games
available to users on as many platforms as practicable, and that it
evaluates a cloud gaming service provider’s technical capacity to provide
an optimal user experience and ‘various business considerations’ before
deciding to publish a game on the cloud gaming service. 777

(b) Another publisher [] stated that its strategy is to ‘reach players
wherever they are’, and that in reaching that goal, it would consider
developing/publishing its games on cloud gaming services. When
deciding on which cloud gaming service to develop/publish a game,
financial considerations are taken into account by weighing the
opportunity cost with the added production costs. 778

(c) Another publisher [] submitted that in deciding whether to make its
games available on a cloud gaming service it will consider the player
experience, development costs, and the expected financial return. 779

773 [] Internal Document.

774 [] response to the CMA’s RFI.
775 [] response to the CMA’s RFI.
776 [] Internal Document.
777 [] response to the CMA’s RFI.
778 [] response to the CMA’s RFI.
779 [] response to the CMA’s s109 notice.

209
 (d) Another publisher [] stated that it will take into account commercial and
technological factors such as the additional costs incurred and potential
consumer base. 780

(e) A further publisher [] stated that it would look at the potential audience,
financial incentives and sales forecasts, also noting that each platform
would require some engineering work to publish on it. 781

8.74 The evidence on multi-homing specific to cloud gaming was mixed and
generally third parties seemed to lack detailed analysis or thorough views on
the topic, mainly because the market is still developing:

(a) One competitor [] submitted that as cloud gaming services have
untethered gamers from consoles, it is eliminating obstacles that prevent
consumers from choosing the devices and platforms for gaming. 782 It
noted that cloud gaming eliminates an important barrier to multi-homing in
the purchase of expensive hardware. However, it suggested that multi-
homing in cloud multi-game subscription services is less likely than in
video streaming due to the high cost of subscribing. It also explained that
gamers may not multi-home between cloud gaming services due to the
loyalty they gain from both investing time and progress in games, and any
investment in platform specific hardware such as controllers. 783

(b) In a call, a competitor [] suggested that the existence of established

players with their own content makes new entry in cloud difficult but not
impossible. It further suggested that, for example, Nintendo, SIE and
Microsoft’s content ecosystems are deep, meaning that other entrants
may struggle to obtain the necessary compelling content to attract users
to multi-home away from incumbent services. 784

(c) Another competitor [] commented that some degree of multi-homing

does and will exist, but []. 785

(d) A further competitor [] suggested that multi-homing may reduce indirect
network effects but that the monthly subscription plan of cloud gaming is
not low cost, so multi-homing can be expected to be less prevalent. 786

780 [] response to the CMA’s RFI.

781 [] response to the CMA’s RFI.
782 [] response to the CMA’s RFI.
783 [] response to the CMA’s RFI.
784 [] call note.
785 [] call note.
786 [] call note.

210
 • Conclusion on network effects and multi-homing in cloud gaming

8.75 We consider the evidence shows that cloud gaming is currently characterised
by both direct and indirect network effects.

8.76 Direct network effects exist because the multiplayer and social aspect of
gaming means that gamers prefer to play on platforms with a large user base,
where they can play with their friends and benefit from better multiplayer
matchmaking. They are also more likely to stick to a platform in which their
gaming progress has been saved (although the BYOG model allows for cross-
progression which may mitigate this). As cross-play between a cloud gaming
platform and other platforms exists at similar levels to console platforms,
direct network effects are mitigated to a similar extent. 787 The strength of
direct network effects seems to be mitigated further by multi-homing when
compared to console gaming, as the financial cost of switching to a new cloud
gaming service is lower than in the market for gaming consoles. However,
since the non-financial costs remain relatively high (ie, moving away from
friends and losing game progress), we consider that direct network effects are
also likely to be prevalent in this market.

8.77 Indirect network effects also seem to be strong. The number of users of a
cloud gaming service is an important consideration for publishers when
deciding whether to make their games available to that service. Whilst
developers generally aim to make games available on as many platforms as
possible, there seem to be cost considerations when deciding whether to
make a game available on a cloud gaming platform. These costs are
especially high if the cloud gaming service provider does not use Windows as
its OS, as most games are optimised for Windows. However, the fact that
cloud gaming service providers are able to use Windows as their OS (albeit
for a licensing fee) means that indirect network effects are likely to be weaker
in cloud gaming services than in gaming consoles.

Overview of competitors

8.78 As explained in our section on market definition in Chapter 5, we consider that

there is a UK market for cloud gaming services, primarily because not every
provider is present in the UK and there are price differences between different
countries. From a supply-side perspective, however, cloud gaming service
providers are global players, and they have been expanding their presence to
different regions in recent years. Since this is a rapidly changing market in

In principle players could cross play amongst PC based cloud gaming providers (such as GFN and Luna),
787

PCs and consoles (if available and PC-console cross play is supported), but this will depend on each specific
game.

211
 which the competitive landscape will likely continue to develop over the next
few years, we consider that an assessment of the major global cloud gaming
providers is relevant to understanding competition in this market today and in
the near future.

8.79 We have identified five main global competitors offering cloud gaming
services today. They are:

(a) NVIDIA GFN. NVIDIA’s GeForce NOW provides streaming services for
PC games using high-end GPUs on its servers, building on its strengths
as a GPU supplier. Its service offers one free and two premium tiers, with
premium tiers providing improved performance, priority servers access,
and longer session lengths. It uses a BYOG model where consumers use
the service to access games already owned on PC storefronts such as
Steam, Epic Games Stores, EA Origin, and Ubisoft Connect. 788 It is
available across several devices including PCs, mobiles and smart
TVs. 789

(b) Amazon Luna. Amazon is a leading provider of cloud computing services

and has invested in improving its virtualisation technology to provide a
better streaming performance to gamers. Luna streams PC games and
has primarily an MGS model with different channels that users can
subscribe to. 790 It has recently also incorporated BYOG for Ubisoft. 791 It
first launched in the US in March 2022, and subsequently launched in the
UK, Germany, and Canada in March 2023. 792 It is available across
several devices including PCs, mobiles, and smart TVs. 793

(c) Microsoft xCloud. Microsoft offers cloud gaming as part of the Game
Pass Ultimate MGS. The service uses current generation console
hardware to stream console games from the Game Pass Ultimate game
library. It is available across several devices including Xbox consoles,
PCs, mobiles, and smart TVs. 794

(d) SIE PlayStation Plus. SIE offers cloud gaming as part of the Premium
tier of its PlayStation Plus MGS. PlayStation Plus provides access to a
large catalogue of games. Its cloud gaming offering, however, is currently
limited to older titles, as the cloud infrastructure uses older console

788 [] response to the CMA questionnaire.

789 ‘GFN system requirements’, accessed by the CMA on 23 January 2023.
790 Parties FMN.
791 ‘Amazon Luna Ubisoft Store', accessed by the CMA on 17 January 2023.
792 ‘Amazon Luna Blog’, accessed by the CMA on 29 March 2023.
793 ‘Luna – getting started’, accessed by the CMA on 23 January 2023.
794 ‘xCloud – ways to play’, accessed by the CMA on 23 January 2023.

212
 hardware. 795 It is only available on PlayStation consoles and Windows
PCs. 796

(e) Boosteroid. Boosteroid offers streaming of PC games using a BYOG

model. It has only one paid tier available through a monthly or annual
subscription. It is available in Europe, Latin America, and North America,
operating 18 data centres where its hardware is deployed, and across
several devices including PCs, mobiles and smart TVs. 797

8.80 Google Stadia was a cloud gaming service that offered streaming of PC
games using a Linux OS. It included a free and premium tier, and both B2P
and MGS features. The service closed down in January 2023. Whilst it is no
longer a competitor in the market, we consider it in the evidence below as its
exit from the market provides insights into the importance of certain assets
and inputs.

8.81 There are several other providers that are potential entrants or are already
active in cloud gaming to some extent. We do not consider their strengths or
position in the market in detail here for the reasons given below. However, we
would still be concerned if the Merger increased barriers to entry or
expansion, thereby constraining their ability to enter and compete effectively:

(a) Blacknut. Blacknut is a provider offering cloud gaming through a multi-

game subscription model. 798 It submitted that it is also planning to add
different models such as BYOG and F2P. 799 However, it offers a very
limited number of AAA games. 800 We have not received sufficient
evidence of its size to consider it as a key global competitor alongside
those mentioned above. However, we note its growth ambitions as
evidenced by a recent investment from Square Enix and availability on LG
TVs. 801

(b) Tencent. Tencent does not currently have a consumer facing cloud
gaming service. 802 One competitor [] told us that, although Tencent has
been exploring avenues for using cloud gaming, these initiatives have
been small and focused on the PRC. 803

795 [] response to the CMA’s RFI.

796 ‘PS Plus – what you need’, accessed by the CMA 23 January 2023.
797 Boosteroid response to the CMA’s RFI.
798 We note that Blacknut also has a significant B2B offering. See Blacknut.biz.
799 Blacknut response to the CMA’s RFI. See also NAB Tradeshow announcement.
800 ‘Blacknut Games’, accessed by the CMA 29 March 2023.
801 See Square Enix looks to the future with strategic investment in cloud gaming pioneer Blacknut and Blacknut

for LG TV.
802 Parties FMN.
803 [] response to the CMA questionnaire.

213
 (c) Shadow. Shadow is a cloud computing service—a service where the user
can access a powerful Windows PC remotely for uses including but not
limited to gaming—rather than a specific cloud gaming service. 804 We
understand it to be a very small competitor and has not featured in any of
the shares of supply submitted to us—see Appendix C: Shares of supply.

(d) Meta. Meta has a small presence in gaming of which cloud gaming [].
[], and in our view is therefore materially different to cloud gaming
services offering the latest high-performance games. 805

(e) Nintendo. Nintendo also offers a very limited cloud gaming service that is
only available on the Nintendo Switch device. Nintendo explained that
[]. 806

(f) Antstream Arcade. Antstream Arcade is a cloud gaming service that

offers access to a library of retro games, such as Pac-Man. 807 Given it
focuses on much older, niche content, we do not consider it as a
competitor in the market for cloud gaming services offering the latest high-
performance games. Antstream Arcade [], supporting this
interpretation.

(g) Others. There are also some lesser-known providers such as Utomik,
Playkey and Netboom. 808 These providers are likely to be much smaller
and are therefore not considered in detail. 809

8.82 In the course of the investigation, we contacted Shadow, Utomik, Playkey,

Netboom, and Blacknut, with a view to obtaining MAU data to calculate
shares of supply, but were unable to obtain this information. We do not expect
that data for these providers would make a significant difference to our
calculated shares of supply, as they have not been mentioned by any other
market participants as significant suppliers, so we expect them to be small.
They are also not included in any shares of supply estimates from the Parties
or third parties.

804 ‘Shadow Cloud Computing’, accessed by the CMA on 17 January 2023.

805 [] response to the CMA’s RFI.
806 [] response to the CMA’s RFI.
807 ‘PC Mag—Antstream Arcade review’ accessed by the CMA on 17 January 2023.
808 The Parties identified three additional competitors: Gamestream, Wiztivi, Ubitus, and Rainway. We consider

that these companies are not close competitors as Gamestream, Wiztivi, and Ubitus are cloud gaming platform
providers to other businesses rather than consumer facing cloud gaming services, and Rainway is an application
allowing users to stream games from their own gaming PC.
809 [] response to the CMA’s RFI.

214
Shares of supply

8.83 We have looked at market shares as an indicator of the relative strength of

currently available cloud gaming services. Our assessment is based on data
and forecasts from several sources, including the Parties’ submissions and
third party evidence.

8.84 We recognise that market share data may not be particularly informative to
assess relative strength in this market. Cloud gaming services remain in their
infancy and, as such, static measures of market share are unlikely to
accurately reflect the relative strengths of different cloud gaming services
providers. We also note that cloud gaming service providers may face some
degree of competition from out-of-market constraints, such as from PCs and
consoles. These constraints are not captured within our market share data,
meaning that market shares may understate the constraints on cloud gaming
providers. We therefore treat our market share estimates for cloud gaming
services as indicative only and interpret them alongside qualitative evidence
on the competitive landscape.

8.85 Microsoft submitted global shares for cloud gaming services in 2021.
Microsoft estimates that xCloud had a [0-10]% share, 810 NVIDIA GFN had a
[50-60]% share, and Facebook Gaming a [10-20]% share.

8.86 Using the Parties’ data and third party evidence, we estimated global shares
of supply for cloud gaming for both 2021 and 2022 (set out at Table 8.1 and
Table 8.2, respectively). Our estimates include broadly like-for-like services
that provide a means to play high-performance games via cloud gaming and,
therefore, exclude Facebook Gaming, Nintendo Switch Online, and Antstream
Arcade. 811 Whilst we have found a UK market for cloud gaming services, we
have presented both UK and worldwide shares of supply here, as we consider
both are informative about the relative size and recent growth of cloud gaming
service providers. We also present the shares of supply for paid cloud gaming
services only.

810 Microsoft response to the CMA’s s109 notice. Given the limited available information on competing providers'
cloud streaming offerings, these shares were produced by Microsoft on a best-efforts basis.
811 A more detailed explanation of our methodology and the caveats surrounding the shares is contained in

Appendix C: Shares of supply.

215
Table 8.1: Shares of cloud gaming services in terms of average MAUs, 2021-2022, worldwide,
including paid and unpaid services.
Service 2021 Average MAUs % 2022 Average MAUs %

xCloud [] [20-30] [] [50-60]

NVIDIA GFN [] [20-30] [] [20-30]
PlayStation Cloud Gaming [] [30-40] [] [10-20]
Boosteroid [] [0-5] [] [0-5]
Google Stadia []* [5-10] [] [0-5]
Amazon Luna [] n/a []† [0-5]
Total [] 100 [] 100

Source: CMA’s calculations on the Parties’ and third parties’ data.

* Figure for April to December 2021.
† Figure for March (launch date) to September 2022, and does not include users who only played games via Free Games with
Prime and/or games purchased from an external partner (eg, Ubisoft)
Note: Figures rounded to nearest thousand. Due to rounding, figures may not sum to 100%.

Table 8.2: Shares of cloud gaming services in terms of average MAUs, 2021-2022, worldwide,
paid services only.
Service 2021 Average MAUs % 2022 Average MAUs %

xCloud [] [30-40] [] [60-70]

PlayStation Cloud Gaming [] [40-50] [] [20-30]
Boosteroid [] [5-10] [] [5-10]
NVIDIA GFN‡ [] [5-10] [] [0-5]
Amazon Luna [] n/a []† [0-5]
Google Stadia* n/a n/a n/a n/a
Total [] 100 [] 100

Source: CMA’s calculations on the Parties’ and third parties’ data.

* Google Stadia was unable to split its MAUs by tier.
† Figure for March (launch date) to September 2022 , and does not include users who only played games via Free Games with
Prime and/or games purchased from an external partner (eg, Ubisoft)
‡ NVIDIA was [].
Note: Figures rounded to nearest thousand. Due to rounding, figures may not sum to 100%.

Table 8.3: Shares of cloud gaming services in terms of average MAUs, 2021-2022, UK,
including paid and unpaid services.
Service 2021 Average MAUs % 2022 Average MAUs %

xCloud [] [30-40] [] [60-70]

PlayStation Cloud Gaming [] [40-50] [] [10-20]
NVIDIA GFN [] [10-20] [] [10-20]
[]Google Stadia* n/a n/a [] [0-5]
Boosteroid [] [0-5] [] [0-5]
Amazon Luna† n/a n/a n/a n/a
Total [] 100 [] 100

Source: CMA’s calculations on the Parties’ and Third Parties’ data.

* Google Stadia did not provide its MAUs split by region for 2021 or January 2022. The 2022 average MAUs figure is for
February—December.
† Amazon Luna was not available in the UK until March 22, 2023. 812
Note: Figures rounded to nearest thousand. Due to rounding, figures may not sum to 100%.

812 ‘Amazon Luna Blog’, accessed by the CMA on 29 March 2023.

216
Table 8.4: Shares of cloud gaming services in terms of average MAUs, 2021-2022, UK, paid
services only.
Service 2021 Average MAUs % 2022 Average MAUs %

xCloud [] [40-50] [] [70-80]

PlayStation Cloud Gaming [] [50-60] [] [20-30]
Boosteroid [] [0-5] [] [0-5]
NVIDIA GFN [] [5-10] [] [0-5]
Amazon Luna† n/a n/a n/a n/a
Google Stadia* n/a n/a n/a n/a
Total [] 100 [] 100

Source: CMA’s calculations on the Parties’ and Third Parties’ data.

* Google Stadia was unable to split its MAUs by tier.
† Amazon Luna was not available in the UK until March 22, 2023. 813
Note: Figures rounded to nearest thousand. Due to rounding, figures may not sum to 100%.

8.87 These estimates show that worldwide Microsoft’s share increased from [20-
30]% in 2021 to [50-60]% in 2022, and that, in 2022, it had [] as many
average MAUs as the next biggest service, NVIDIA GFN.

8.88 For users of paid services only, worldwide Microsoft’s share was higher, with
[30-40]% in 2021, and [60-70]% in 2022. The share of NVIDIA GFN was []
when only considering paid users.

8.89 The estimates for the UK show that Microsoft’s share increased from [30-
40]% in 2021 to [60-70]% in 2022. In 2022 it had over [] as many average
MAUs as the next biggest service, PlayStation Cloud Gaming.

8.90 For users of paid services only in the UK, Microsoft’s share was higher, with
[40-50]% in 2021, and [70-80]% in 2022.

8.91 The Parties submitted that shares of supply analysis is misleading and
overstates Microsoft’s strength as [], and it is therefore not comparable to
standalone cloud gaming services such as GFN. 814 Whilst it is not possible to
determine how many users xCloud would have as a standalone service, the
evidence described above indicates that cloud gaming attracts users to XGPU
and that a significant proportion ([]%) of users would be willing to pay extra
for it, even though they already have access to offline play, which suggests for
those customers, streaming and offline play are not substitutes, and these
customers may be interested in cloud gaming as a standalone service
(although we recognise that it is difficult to fully separate Microsoft’s ability to
‘upsell’ xCloud to Game Pass customers from Microsoft’s ability to compete
for stand-alone xCloud customers). In any case our concern is about
Microsoft’s future position based on our assessment of its strength, and its

813 ‘Amazon Luna Blog’, accessed by the CMA on 29 March 2023.

814 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.86.

217
 substantial number of users now indicates that it may continue to attract a
large number of users as cloud gaming grows.

8.92 One third party []’s internal document contained a third party analyst report
from March 2022 which showed 2021 shares of supply for cloud gaming
services in terms of revenue. 815 It showed that PS Now had a share of more
than []% in 2021, with xCloud, NVIDIA GFN and Luna significantly smaller.

8.93 The same report contains estimates for the future of the cloud gaming market
as a whole. It estimated that revenue would increase by almost 300% from
2021 to 2026, increasing from less than c. 0.4 billion to c. 1.05 billion USD. 816
Within this, xCloud’s revenue was expected to increase from under 50 million
to c. 450 million USD in the same period, estimating it to be the largest cloud
gaming service by revenue in 2026.

8.94 Whilst we place less weight on independent reports than on the data that we
gathered from third parties, we note that the report does corroborate our own
findings regarding the growth of xCloud.

8.95 Our shares of supply may not be a good proxy for the future state of the cloud
gaming market, given the caveats in the methodology and the nascent nature
of cloud gaming. However, we consider that these shares provide a useful
indication of the current strength of different cloud gaming providers.

Microsoft’s strengths in cloud gaming

8.96 This section considers whether Microsoft holds pre-Merger advantages over
current and potential cloud gaming rivals as a result of its wider multi-product
ecosystem. The section is structured as follows:

(a) First, we consider the extent to which individual assets could give
Microsoft an advantage in cloud gaming services. We focus our
assessment on the following products and services:

(i) Microsoft’s Windows and Xbox OSs;

(ii) Microsoft’s Azure cloud infrastructure;

(iii) Microsoft’s first- and third party content;

815[] Internal Document

816We consider forecasts for the size of the cloud gaming market in more detail in the future development of
cloud gaming section. We note that this forecast is considerably lower than those from other market reports, with
the forecast for 2026 being lower than many estimates of the market size for 2021.

218
 (b) Second, we consider the assets held by rivals, and whether these are
likely to offset any advantage Microsoft may have.

(c) Third, we consider the overall view of the respective strengths of different
cloud gaming competitors.

Operating systems

8.97 In this section, we consider evidence on the extent to which Microsoft may
have a strong position in OSs that can be used for cloud gaming, in particular
through its ownership of Windows and the Xbox OSs. We also consider
whether any such strong position would be likely to confer on Microsoft a
competitive advantage in cloud gaming services.

Importance of operating systems

8.98 OSs are system software products that control the basic functions of
computing devices, such as servers, PCs, tablets, and smartphones, and
enable the user to use the device and run application software on it, including
games. 817 A cloud gaming service requires an OS to run on the cloud
infrastructure it uses to be able to provide its service.

8.99 The OS also has an important impact on the availability of gaming content.
Games are developed to be compatible with specific OSs. Cloud gaming
services can offer more content if they run on an OS for which many games
have been developed.

8.100 In the following section, we consider Microsoft’s pre-existing strength in cloud

gaming services as a result of owning Windows OS and Xbox OS. We also
consider evidence on the ability of rivals to access Windows OS for their own
cloud gaming service (including the cost) and any alternatives available to
them, such as porting games to rival OSs (eg, Linux) and the use of
compatibility layers (eg, Proton). 818

817Microsoft response to the CMA’s s109 notice.

818Porting requires the game developer to create a version of the game that is compatible with a different OS.
Once developed the game can then run natively on that OS. With a compatibility layer, the Windows version of
the game is still used, meaning the developer does not need to create a new version of the game, but an
additional layer of software is used to translate the game allowing it to run on a different OS.

219
 Parties’ views

8.101 The Parties submitted shares of supply indicating that the Windows PC OS is
the predominant OS for PCs. Windows’ share of PC OSs is []% worldwide
and []% in the UK (January to June 2021). 819

8.102 We note that Windows’ market share for PC OSs used for gaming is even
higher, upwards of 95%, based on a Steam survey. 820

8.103 The Parties submitted that because xCloud runs on Xbox console hardware
(running the Xbox OS and streaming Xbox console games), Microsoft does
not benefit from any Windows-related advantages when competing in the
supply of cloud gaming services. 821

8.104 The Parties submitted that []. 822

8.105 With respect to the cost of Windows for rival cloud gaming services, the
Parties submitted that this is not relevant unless it results in rivals facing
significantly higher costs than Microsoft. 823

8.106 Regarding alternatives to Windows, Microsoft submitted that []. 824

8.107 Activision submitted that, based on the experience of porting [] to [],
porting typically takes a team [], with (taking the specific example of []
roughly [] USD upfront cost and [] USD annual ongoing cost. It also
noted the [] cost of []. 825

8.108 Microsoft stated that Linux is an alternative to Windows that is extensively

used in cloud computing and is viable for cloud gaming. 826 Microsoft
submitted that it expects []. It submitted that Valve had successfully used
Proton on its Steam Deck handheld gaming device and []. It submitted that
[] shows that Windows games can be run on Linux servers using Proton
and other compatibility layers. 827 It also submitted that there are already game
streaming providers (eg Maximum Settings) which use Proton. 828 Microsoft

819 Microsoft response to the CMA’s RFI.

820 ‘Steam Hardware & Software Survey: December 2022’, dated December 2022, accessed 23 January 2023.
821 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.21.
822 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.24.
823 Microsoft Annex 4 to response to the Provisional Findings.
824 Microsoft response to the CMA’s s109 notice.
825 Activision response to the CMA’s s109 notice.
826 Microsoft, Annex 4 to response to the Provisional Findings.
827 Microsoft response to the CMA’s Supplementary Evidence paper.
828 Microsoft response to the CMA’s s109 notice; and Microsoft response to working paper.

220
 stated that the large majority of the best-selling Windows PC games can be
run on Proton and that it is highly rated by gamers. 829

8.109 Microsoft noted the complexities that can arise from anti-cheat software
meaning that some game titles cannot run on certain open-source OS and
compatibility layers because of security concerns. 830

8.110 With regard to Google’s recent decision to shut down Stadia, Microsoft
submitted that this was not linked to the use of a non-Windows OS. Microsoft
submitted instead that the failure was linked to it having []. 831

8.111 Microsoft also submitted that it had [] referencing a statement from
CodeWeavers, the developers of Wine—the compatibility layer for running
Windows applications which Proton is based on. 832 The quote stated that a
certain application programming interface 833 (API) could not be used to make
it difficult for Wine to keep up, as changing the API would break Microsoft’s
existing codebase as well as those of many other software developers.

8.112 It also submitted that using Linux has certain advantages for cloud gaming
services, notably its wide use in cloud computing which means there are
greater opportunities for re-use of Linux servers for non-gaming uses,
allowing for recouping of costs. 834

Our assessment

• Internal document evidence

8.113 An internal document from Microsoft recognises the []. 835

8.114 Internal documents from Microsoft also describe the []:

(a) One email chain from []. The same email says that ‘[]. While noting
that [] 836 []. 837

(b) Another email from []. 838

829 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.46.
830 Microsoft response to the Remedies Working Paper
831 Microsoft response to the CMA’s s109 notice.
832 Microsoft response to working paper.
833 An API is a way for two or more computer programs to communicate with each other.
834 Microsoft response to the working papers.
835 Microsoft Internal Document.
836 []
837 Microsoft Internal Document.
838 Microsoft Internal Document.

221
 (c) An email from [] recognises []. The email says that []. 839

(d) An email from []. 840

(e) An Activision internal document also describes that a key barrier to putting
content on [] is that []. 841

8.115 Several internal documents discuss Proton and show that []:

(a) In an email chain [] 842

(b) Another email [], 843 []. 844 In our view the statement that [],
suggests [].

(c) An email from []. 845

(d) Some internal discussions suggest that []. 846

8.116 The Xbox OS also has a large library of compatible games and, in particular,
the full range of games that Microsoft already uses to support the Xbox
console business. Data provided by Microsoft indicates that [] games are
playable on Xbox Series X (which runs Xbox OS). 847 This has allowed
Microsoft to set up cloud infrastructure using Xbox consoles running Xbox OS
in a quick and low-cost way, whilst maintaining access to a large library of
games. This is discussed in more detail in the next section. Microsoft has
submitted that [] and therefore does not gain an advantage from this library
of games. 848 However, in our view the advantage still arises because, where a
game is already compatible with Xbox OS, there is no additional development
work or porting required to make the game available on xCloud, reducing the
cost of making games available for streaming.

• Third party evidence

8.117 The strong position of Windows in OSs, especially among PCs used for
gaming, is reflected in a similarly large proportion of PC games that are
developed for Windows. Of the games available on Steam, 99.97% are
compatible with Windows, compared to 21% for MacOS, and 14% for

839 Microsoft Internal Document.

840 Microsoft Internal Document.
841 Activision Internal Document.
842 Microsoft Internal Document.
843 [].
844 Microsoft Internal Document.
845 Microsoft Internal Document.
846 Microsoft Internal Document.
847 Microsoft response to the CMA’s RFI.
848 Microsoft, Annex 4 to response to the Provisional Findings.

222
 Linux. 849 This means that a cloud service provider that uses Windows will
benefit from access to a significantly wider range of games that work on its
system without any porting or emulation than will a cloud service provider that
chooses other non-console OSs such as Linux or MacOS. Several third
parties submitted that Windows is very important for cloud gaming:

(a) One competitor [] stated that Windows was an essential input for cloud
gaming because, as a result of its large installed user base, publishers
tended to write games for Windows/DirectX. 850

(b) One potential entrant [] described Windows as crucial for gaming and
noted that rivals that had attempted to use alternative OSs had failed. 851 It
also described its reliance on Microsoft’s agreement to use Windows for a
specific element of its service, not covered by the standard licensing
agreement. 852

(c) A third party publisher [] stated that it mainly developed PC games for
Windows as it was the most commonly used OS amongst gamers. 853
Activision also stated that all its available PC games run on Windows. 854

8.118 The OS can have a significant impact on a cloud gaming service provider’s
costs. Whilst some OSs are free to use, others—including Windows—have
licensing fees which can be substantial and represent a significant proportion
of a cloud gaming services overall costs. Evidence, including from internal
documents, provided by several rival services demonstrated that the cost of
licensing Windows is substantial:

(a) One competitor [] said that Microsoft could charge high fees for
Windows and that this provided Microsoft with a competitive advantage. It
stated that the Windows licence fee it paid []. 855 Internal documents
from this competitor show that Windows fees represent []% of its
streaming cost per hour. 856

(b) Another competitor [] used Windows for early prototypes of its service
but indicated that use of Windows would have almost doubled its
streaming cost per hour, and that it would therefore be difficult for anyone
other than Microsoft to provide a financially sustainable cloud gaming

849 Analysis of 'Steam store page', accessed by the CMA on 24 October 2022.
850 [] response to the CMA’s RFI.
851 [] call note.
852 [] call note.
853 [] response to the CMA’s RFI.
854 Activision response to the CMA’s s109 notice.
855 [] response to the CMA’s RFI.
856 [] Internal Document.

223
 service using Windows that could be profitable in view of Windows
licensing costs. 857

(c) Internal documents from another competitor [] showed that []. 858

8.119 One option for competitors using an OS other than Windows is to have
developers port games and make them compatible with other OSs, such as
Linux.

8.120 Responses from various third parties highlight the difficulties and cost
associated with this approach. For example, one competitor [], submitted
that the cost of porting games to Linux is a significant barrier to making
content available on a cloud gaming platform and, therefore, is a key
challenge in attracting users to a Linux-based platform. 859 It said that there
was a substantial cost involved in porting a game built on the
Windows/DirectX platform to the Linux/Vulkan platform. 860 [], the competitor
described the development of a cloud gaming service on a non-Windows OS
as time and resource intensive, dependent on acquiring a significant customer
base to recoup the initial investment, and not viable without sufficient breadth
and depth of content. 861 [].

8.121 A publisher [] described the OS as decisive in its decision to make a game
available on a cloud gaming service, with a service using Windows requiring
minimal effort. 862 In comparison it noted the difficulty of creating a port for
cloud gaming services requiring publishers to use their own software
development kits (SDKs), describing it as taking some time, effort and
investment. 863 It described the cost of porting as dependent on several
factors, but generally ranging between [], also suggesting that porting costs
could decrease over time if a developer has a library including games or
game engines that have previously been ported to the same OS. 864

8.122 We have considered the experience of Google’s Stadia, which was a Linux-
based cloud gaming platform that relied on porting games from Windows to
Linux. Stadia announced its closure in September 2022 and ceased operating
as a cloud gaming service provider to customers in January 2023. [] noted
that its failure was linked to [] to attract sufficient users and that the choice

857 [] response to the CMA’s RFI.

858 [] Internal Document.
859 [] response to the CMA’s RFI.
860 [] response to the CMA’s RFI.
861 [] response to the CMA’s RFI.
862 [] response to the CMA’s RFI.
863 [] response to the CMA’s RFI.
864 [] response to the CMA’s RFI.

224
 of a [] was a factor in this as it ‘increased friction’ for development
studios. 865

8.123 Another option for providers using a non-Windows OS is to use a compatibility

layer which effectively translates a Windows game, allowing it to be played on
a non-Windows OS. Proton is a notable example of a compatibility layer for
Linux, which was introduced by Valve in 2018. 866 It is already in use by some
cloud gaming providers and is under consideration by others.

8.124 Some third parties were confident in the effectiveness of Proton as an

alternative to Windows. One competitor [] is already partially using Proton
(alongside Windows) for game streaming []. 867 Internal documents from this
competitor []. 868 It submitted that it was too early to be sure of its
effectiveness in practice but that the relevant individual was not aware of any
major issues so far. 869

8.125 Others noted it as an option but expressed certain reservations about its
effectiveness. One competitor [] described that []. 870 It described the
disadvantages as [], the reduced performance associated with using a
compatibility layer, and the potential for unexpected issues and bugs, eg, due
to anti-cheat software. 871 It also noted that []. 872

8.126 Another competitor [] stated that Proton was less effective for games built
on the latest version of DirectX. 873 It also stated that Proton would perform
less well for new games and therefore would not be suitable for offering
games on release date. 874

8.127 Another competitor [] stated that it uses different technical setups to run
Windows games, including Linux, Proton, and other compatibility layers. It
submitted that, although it is generally possible to use such solutions, it is
often impossible to reach good levels of performance, especially with AAA
games. 875

865 [] call note.

866 Microsoft response to the CMA’s s109 notice.
867 [] response to the CMA’s s109 notice.
868 [] Internal Document.
869 [] call note.
870 [] call note.
871 [] response to the CMA’s RFI. Publishers use anti-cheat software to try to identify and prevent efforts to

obtain unfair advantages during gaming. As an unexpected intervention in the gaming software, compatibility
layers can become a target for anti-cheat software.
872 [] call note.
873 [] response to the CMA’s RFI.
874 [] call note.
875 [] response to the CMA’s RFI.

225
8.128 One potential entrant [] similarly noted disadvantages relating to
performance, additional work required to resolve issues and bugs, and the
difficulty of using Proton to provide games on release date. 876

8.129 One publisher [] noted that compatibility layers could reduce performance
and the key consideration for [] in deciding whether to put its games on a
PC OS by way of a compatibility layer is whether its games performed well on
the PC OS and hardware, notwithstanding the presence of a compatibility
layer. 877

8.130 One publisher [] stated that Proton enables Windows games to run on
Linux with an almost seamless experience and that it could envision offering
its games on a platform using it. 878

• Data on Game Performance

8.131 Data on game performance using Proton suggests that it is an imperfect

substitute for gaming on Windows, and a significant proportion of games
cannot be played without issues. Valve categorises games based on how well
they run using Proton on Valve’s Steam Deck handheld gaming device. 879
Data on the proportion of Steam’s top 100 games in each category showed
that 16% were playable with no issues, with a further 33% playable with some
tweaking required. 31% were unplayable and 20% had not been tested. 880

8.132 A separate community-based categorisation is also available, and this also

suggested that Proton was still subject to material performance issues. 881
Data for Steam’s top 100 games using this categorisation suggested that 19%
ran perfectly, 53% ran perfectly after some tweaking, 14% ran with minor
issues, 2% ran but with more substantial issues, and 12% were unplayable.
Whilst Microsoft submitted that this data demonstrates the effectiveness of
Proton, 882 in our view it is consistent with the interpretation that it is an
imperfect substitute to using the Windows OS.

876 [] call note.

877 [] response to the CMA’s RFI.
878 [] response to the CMA’s RFI.
879 steamdeck.com | ‘Deck Verified', accessed by the CMA on 9 December 2022.
880 Analysis of data from 'ProtonDB dashboard', accessed by the CMA on 19 October 2022. Proportions will be

subject to change as the top 100 games changes. Some performance issues will relate to Steam Deck as a
handheld device rather than Proton.
881 'ProtonDB', accessed by the CMA on 9 December 2022.
882 Microsoft response to working papers.

226
 • Conclusion on Microsoft’s position in operating systems

8.133 OSs affect both the range of content available and cost to cloud gaming
service providers. We believe that Windows OS and Xbox OS both give
Microsoft a wide range of games that readily work without any need for
porting or use of a compatibility layer. Further, as both OSs are owned by
Microsoft, they can be self-supplied at little or no incremental cost.

8.134 With respect to Microsoft’s submission that it does not benefit from Windows-
related advantages on the basis that xCloud currently does not use Windows,
we make the following observations:

(a) First, Microsoft has two strong OSs to choose from, and each has a
significant library of compatible games that can run without the need to
port them or emulate another OS.

(b) Second, Microsoft []. We assess the importance of Azure below, [].

8.135 We consider the evidence above to be consistent in showing that Microsoft’s

ability to self-supply Windows represents a significant cost saving relative to
rivals (as Microsoft does not have to pay an expensive license fee for
Windows OS).

8.136 Use of a console OS provides similar content advantages, but this option is
limited to a small number of competitors. A proprietary console OS is only
available to providers who can use an existing console hardware OS for game
streaming, which is limited to Microsoft, SIE, and Nintendo, the latter of which
does not currently have its own cloud infrastructure to provide its cloud
gaming service. 883

8.137 For providers other than Microsoft, there is a significant cost to using
Windows as licensing fees must be paid to Microsoft. This is demonstrated by
several third party responses and recognised by Microsoft. As described
above, Microsoft acknowledges that this licensing fee is relevant if it results in
rivals facing higher costs than Microsoft. Several competitors have submitted
that the licensing fee is a significant cost, and consequently the licensing fee
in many cases gives Microsoft a cost advantage. Whilst this section
specifically considers advantages arising from OSs, we consider other
advantages (including those of rivals) and overall strength in subsequent
sections.

883 [] response to the CMA’s RFI.

227
8.138 Evidence suggests that porting a game to a different OS is expensive, and
developers are generally unwilling to develop games for OSs without a large
installed user base. Google Stadia, for example, attempted to run its cloud
gaming service on Linux and its business failed. Many see the lack of content
on Stadia’s Linux-based platform as a major contributor to its failure. We
believe that this shows that porting games to an alternative OS (ie requiring
developers to create a new version of the game compatible with that OS) is
not a viable alternative to Windows currently and is unlikely to be in the near
future.

8.139 Using a Proton compatibility layer on a Linux-based cloud gaming service

may be a better alternative to porting, and it is already in use by some gaming
platforms. [] and [] [].

8.140 There are, however, limitations to Proton. The majority of games require
additional work to run well on Proton, with many not working at all, and it
needs to be maintained to keep up with updates to Windows or DirectX. The
presence of a compatibility layer can also reduce performance, and
complications can arise from anti-cheat software. Releasing games day and
date using Proton is also difficult given the additional work required. It is
possible that further investment and increased use of Proton in gaming could
reduce these limitations but that remains uncertain.

8.141 We consider that Microsoft’s rivals in cloud gaming services, therefore, face
the choice between high costs associated with using Windows or using an
alternative OS (such as Linux) with limited native content. The ability to use its
Windows and Xbox OS without a licensing fee, therefore, provides Microsoft
with a significant advantage in cloud gaming services.

Cloud infrastructure

8.142 This section considers whether Microsoft has a strong position on cloud
infrastructure given its ownership of the Azure cloud infrastructure network,
and its option of using console-based infrastructure. First we describe
Microsoft’s current use of cloud infrastructure and the possibility of using
Azure in the future. We then consider the strength of Microsoft’s position by
comparing its cloud infrastructure options to the alternative available to rivals.

8.143 Cloud infrastructure is an essential input for cloud gaming service providers.
Cloud gaming platforms operate by storing and executing games remotely on
hardware in a data centre and streaming the game to a gamer’s device.
Within the data centre, games which are being streamed are stored and run
on hardware. The type of hardware depends on the provider and can range

228
 from bespoke gaming hardware to servers which can be used for multiple
workloads. 884

Using Azure for cloud gaming

8.144 This section considers Microsoft’s current use of cloud infrastructure for its
cloud gaming service, and its plans for the future. We consider Microsoft’s
submissions as well as internal document evidence on the evolution of its
plans over time.

Parties’ views

8.145 Microsoft submitted that it currently does not use Azure for its cloud gaming
service, which instead is provided on dedicated Xbox consoles located in
Microsoft data centres, streaming console games running on the Xbox OS. It
submitted that it cannot derive any advantage from an asset that it does not
use. 885 It noted that this []. Microsoft submitted that this []. 886 It also only
allows for streaming of console games, not PC games. 887

8.146 Microsoft submitted that there are expected benefits to using Azure PC
infrastructure for cloud gaming, []. It also noted the advantages of scale to
cloud infrastructure providers such as Azure. 888

8.147 Microsoft submitted that it []. 889 It stated the reasons for this were []. 890
Microsoft also stated []. 891 Microsoft also submitted that internal documents
show that []. 892 As described previously, Microsoft submitted that it []. 893
It also submitted that []. 894

Our assessment

• Internal document evidence

8.148 Microsoft’s internal documents show that, []:

884 Microsoft response to the CMA’s s109 notice.

885 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 5.21-5.22.
886 Microsoft, Annex 4 to response to the Provisional Findings.
887 Microsoft response to the CMA’s s109 notice.
888 Microsoft response to the CMA’s s109 notice.
889 Microsoft response to the CMA’s s109 notice.
890 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.24.
891 Microsoft, Annex 4 to response to the Provisional Findings.
892 Microsoft, Annex 4 to response to the Provisional Findings.
893 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 5.25-5.26.
894 Microsoft response to the working papers.

229
 (a) Slides from May 2020 []. 895 Microsoft has submitted that []. 896
Microsoft submitted that []. 897

(b) Emails from [] in []. 898

(c) An internal evaluation of rival cloud gaming services [] describes

benefits from using PC hardware []. 899

(d) Discussion between []. 900

(e) An email discussion []. 901

(f) An internal chat []. 902

(g) A [] cloud gaming performance evaluation report []. 903

(h) Emails from []. 904

(i) Similarly, []. 905

(j) Emails from []. 906

(k) A Skype chat between [] and [] from May 2022 discusses the plan for
PC streaming. Although [] describes [], []states that for long-term
compatibility issues such as [], that []. 907

(l) Another email []investment are also highlighted however. 908

(m) Slides []. 909

(n) Slides from August 2022 outlining the FY23 budget plan include []. 910
This suggests that [].

895 Microsoft Internal Document.

896 Microsoft response to working papers.
897 Microsoft response to working papers.
898 Microsoft Internal Document.
899 Microsoft Internal Document.
900 Microsoft Internal Document.
901 Microsoft Internal Document.
902 Microsoft Internal Document.
903 Microsoft Internal Document.
904 Microsoft Internal Document.
905 Microsoft Internal Document.
906 Microsoft Internal Document.
907 Microsoft Internal Document.
908 Microsoft Internal Document.
909 Microsoft Internal Document.
910 Microsoft Internal Document.

230
 • Conclusion on using Azure for cloud gaming

8.149 With respect to Microsoft’s cloud infrastructure plans, internal documents [],
with this described as being []. Some internal documents note [], and
Microsoft has highlighted that [] and documents [].

8.150 We [], we believe that Microsoft’s use of Xbox consoles provides it with an
advantage over most rivals already as described in the following section.

Microsoft’s position in cloud infrastructure

8.151 The previous section considered whether Microsoft is likely to use Azure for
cloud gaming services. In this section, we consider Microsoft’s position in the
upstream market for cloud infrastructure, including the availability of
alternatives.

Parties’ views

8.152 Microsoft submitted that there are plenty of cloud infrastructure alternatives.
According to Microsoft, cloud infrastructure is standardised, commoditised,
and competitive. It stated that there are many providers, all of which source
GPUs from the same suppliers and have comparable offers that can, and in
some instances already are, being used for cloud gaming. 911 It submitted that
Amazon and Google had [], and that NVIDIA had []. 912 It also submitted
that it had [], and that Azure can therefore not be considered a competitive
advantage. 913

Our assessment

• Internal document evidence

8.153 One Microsoft internal document describes various advantages Microsoft has
from Azure. []. 914 In submissions to the CMA, Microsoft submitted that
[]. 915

8.154 Other internal documents indicate that Azure []. A document [] discussing
infrastructure services opportunities in cloud gaming notes that []. 916 A note
of discussions with a customer from [] describes again that Azure []. The

911 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.45.
912 Microsoft response to working papers.
913 Microsoft response to the CMA’s Supplementary Evidence paper.
914 Microsoft Internal Document.
915 Microsoft response to working papers.
916 Microsoft Internal Document.

231
 same note describes []. 917 A follow up email from September 2022 notes
that the customer is interested in using []. 918 In our view this is an indication
that []. Although this provider [], 919 we do not think this contradicts the
evidence about [], and the internal document referenced above shows that
the provider [].

8.155 An internal document from Activision also states that, despite [] in []
cycles, [] is well positioned for [] with its ability to leverage []. 920

8.156 Microsoft internal documents also provide some comparison of expected cost
differences between major providers, []. One document discussing
infrastructure services opportunities in cloud gaming notes [], but states
[] 921 []. 922 An email [] but notes that []. 923

• Third party evidence

8.157 The cost of cloud infrastructure is a substantial component of a cloud gaming

service’s overall costs. One independent report submitted by Microsoft notes
that compute cost per hour (ie the cost per hour of cloud usage) is []. 924
Responses from third party cloud gaming service providers estimated that
cloud infrastructure represents from [] to []% of overall costs. 925

8.158 The technical capability of cloud infrastructure is also important in determining

the performance of games on a cloud gaming service, and therefore the
customer experience. One third party described several key requirements for
cloud gaming infrastructure: low latency, high bandwidth, and reliability. 926
Another third party described how cloud infrastructure used for gaming
requires GPUs to provide the smooth graphics and dynamic experience that
customers expect. 927

8.159 Cloud gaming services can self-supply cloud infrastructure, either utilising an
existing cloud network or building one, or use a third party cloud infrastructure
provider. The decision will depend on the firm’s available assets and as such,

917 Microsoft Internal Document.

918 Microsoft Internal Document.
919 [].
920 Activision Internal Document.
921 Cloud-based game streaming services are computationally intensive and also require significant network

bandwidth (ie network egress) on account of being video-heavy. All providers of streaming and other online
services need to pay egress charges, which is the cost to send information over the internet from a data centre to
an end user. The greater the volume of data transported over the internet, the lower the per unit charge. As such,
providers can benefit from economies of scale by providing a range of streaming and other online services.
922 Microsoft Internal Document [].
923 Microsoft Internal Document [].
924 Microsoft Internal Document.
925 Third party responses to the CMA questionnaire: [], [] and [].
926 [] response to the CMA questionnaire.
927 [] response to the CMA questionnaire.

232
 any advantage Microsoft has will vary according to the rival we compare
Microsoft to.

8.160 Third party responses indicate that most current competitors, with the
exception of Nintendo, self-supply cloud infrastructure. For example:

(a) Amazon [], 928 which has the []. 929

(b) SIE generally [] to provide cloud gaming services. 930

(c) NVIDIA developed its own custom cloud gaming servers utilising its GPU
assets []. 931

(d) Google hosted Google Stadia []. 932 Its GCP cloud infrastructure
network has []. 933

(e) Nintendo uses a third party cloud infrastructure provider [] to run its
limited cloud gaming service. 934 The Parties submitted that this provider
[] uses cloud infrastructure from multiple cloud infrastructure providers
[]. 935

8.161 Third party responses noted the cost advantages of having a large cloud
infrastructure network.

8.162 One rival [] noted that, by being a leading provider of cloud infrastructure
services, Microsoft can leverage a much lower cost structure for its cloud
gaming service. 936 This rival described that large cloud providers enjoy a
significant cost advantage in cloud gaming services because they contract at
much higher volumes than others including itself, and have cloud
infrastructure located throughout the world. 937 It described that []. 938

8.163 Another rival [] described how, as it is unable to replicate Microsoft’s cloud
infrastructure network, it would never be able to run its cloud gaming service
as efficiently. 939 It also provided internal document evidence demonstrating

928 [] call note.

929 Parties, FMN.
930 [] response to the CMA’s RFI.
931 [] response to the CMA’s RFI.
932 [] response to the CMA’s RFI.
933 Parties FMN.
934 [] response to the CMA’s RFI.
935 [] response to the CMA’s s109 notice.
936 [] response to the CMA’s RFI.
937 [] response to the CMA’s RFI.
938 [] call note.
939 [] response to the CMA’s RFI.

233
 that it considers Microsoft to have unique advantages in cloud gaming
services, of which Azure is one. 940

8.164 A market report from January 2021 described []. 941

8.165 Third party responses indicated that several cloud infrastructure providers
offered servers suitable for cloud gaming. One cloud infrastructure provider
[] said that its platform offered the required capabilities for cloud gaming,
[]. 942 Another potential provider stated that several cloud infrastructure
providers have similar capabilities. 943 As it uses older console hardware, []
cloud gaming service is currently behind rivals and can only offer older
games. 944

8.166 Some third parties stated that providing a cloud gaming service using third
party cloud infrastructure, ie procured from a public cloud infrastructure
provider such as Azure, AWS or GCP, was not viable.

8.167 One cloud gaming service rival [] described how it had built its own [] for
its service []. It further stated that it []. 945 [] notably is a [], a key
input for cloud infrastructure, which will provide [].

8.168 Similar comments were made by another provider [] in a market report. It
noted that it would not be possible to provide a cloud gaming service using a
public cloud infrastructure offering as servers are not optimised for gaming,
hardware is often outdated, and coverage is limited. 946 The same provider told
us that it does not use third party cloud infrastructure due to the low
performance, poor customization, and high costs offered by most of the public
cloud providers. This provider uses its own server hardware designed in
collaboration with several technology companies, and deploys its servers in
third party data centres. 947

8.169 A market report [] stated that []. 948

8.170 However, other third party responses stated that, whilst high costs may make
use of third party infrastructure difficult currently, it is likely to become feasible
as streaming costs decrease and providers develop cloud gaming optimised

940 [] Internal Document.

941 Microsoft Internal Document.
942 [] call note.
943 [] call note.
944 [] response to the CMA’s RFI.
945 [] response to the CMA’s RFI.
946 []Internal Document.
947 [] response to the CMA’s RFI.
948 Microsoft Internal Document.

234
 servers in the longer run. One provider [] described it as expensive today
but as being feasible in the ‘not-too-distant future’. 949

8.171 As described above, cloud infrastructure represents a substantial proportion

of a cloud gaming service’s costs, therefore streaming cost is likely to be a
key determinant of a cloud gaming service’s competitiveness.

8.172 It is difficult to make an exact comparison across providers as methods for

calculating cost per streaming hour may differ, the cost is heavily influenced
by the usage which can vary substantially, and costs are developing over time
as most providers aim to reduce them through various means. Despite these
caveats, comparison of the evidence submitted by providers shows:

(a) Microsoft submitted that current cost per streaming hour is [] using
console-based servers. 950 Internal documents note that []. 951 A further
document describes that []. 952 Another document on [] outlines
targets []. 953

(b) One competitor’s [] internal documents show []. 954 []. 955 []. 956

(c) Another competitor’s [] internal document from January 2021 indicates
a cost per streaming hour of []. 957 It is important to note that [] uses
[] hardware for its streaming service. []. 958 It is therefore not directly
comparable to other cloud gaming services currently. This will explain at
least some of the large cost difference compared to others []. According
to internal documents [] intends to move to []. 959

(d) Another competitor’s [] internal document from January 2021 shows
[]. 960 A more recent August 2022 internal document indicates []. 961

(e) Finally, one competitor [] submitted that its cost per streaming hour was
[]. 962

949 [] call note.

950 Microsoft response to the phase 2 Issues Statement, 31 October 2022, table 10.
951 Microsoft Internal Document.
952 Microsoft Internal Document.
953 Microsoft Internal Document.
954 [] Internal Document.
955 [] Internal Document.
956 [] Internal Document.
957 [] Internal Document.
958 [] accessed by the CMA on 15 December 2022.
959 [] Internal Document.
960 [] Internal Document.
961 [] Internal Document.
962 [] response to the CMA’s RFI.

235
8.173 Based on the available evidence, and noting the difficulties of making direct
comparisons, [] has the lowest current streaming cost, but this is due to
[]. Its costs can be expected to []. At that point, it is likely to have higher
costs than []. [], [], and [] have relatively similar costs currently. []
current costs are considerably higher, but it has []. Both [] and [] have
similar long-term targets for cost per streaming hour.

8.174 It is clear from Microsoft’s internal documents that it []. It is worth noting that
Microsoft’s predicted costs for [] are also below [], meaning that the
combination of [] significantly increases Microsoft’s advantage in cloud
gaming.

• Conclusion on Microsoft’s position in cloud infrastructure

8.175 Cloud infrastructure is an essential long-term input for cloud gaming services
and has an important impact on costs and performance.

8.176 Microsoft currently has the option of using Xbox consoles [] and which
enables it to offer cloud services at a competitive cost. Whilst Microsoft has
submitted that this infrastructure [], in our view Microsoft can [] its Azure
infrastructure (ie if Microsoft needed to [], not only would that drive the
impetus to [], but this [].

8.177 Evidence suggests [], given advantages such as []. Whilst the current
[].

8.178 Some rivals have their own cloud infrastructure solution. []. Only Microsoft
has a short-term, lower cost hardware option in the form of console
infrastructure, and a large cloud infrastructure network that can enable it to
efficiently scale its service as cloud gaming grows.

8.179 Alternatively, cloud gaming service providers can also access third party cloud
infrastructure. Evidence suggests that this may not be a financially viable
option at present but may become viable in future.

8.180 In any event, we consider that Microsoft has a strong position in cloud gaming
as a result of its ownership of Azure, one of the biggest cloud infrastructure
providers globally. This will likely give Microsoft a significant cost advantage
over rivals without an in-house cloud infrastructure (although not over rivals
with their own cloud infrastructure []). When this cost advantage is
combined with Microsoft’s advantage from owning Windows OS, a license
which is a significant proportion of rivals’ costs, it becomes difficult to see how
any cloud gaming service provider—whether or not it has its own cloud
infrastructure—can match Microsoft’s cost advantages. We therefore believe

236
 that Azure will likely help to compound Microsoft’s advantages over rival cloud
gaming service providers.

First and third party content

8.181 Microsoft is a game publisher and currently owns 24 game development

studios, several of which it acquired in recent years. These studios make
games such as Minecraft, Forza, Elder Scrolls, and Halo for Xbox and other
consoles, PC, and mobile devices. Some of this content is available
exclusively on Xbox and some is licensed to rival console providers. 963

8.182 Microsoft also has established relationships with third party developers and
publishers, including a large existing library of third party content available on
Game Pass.

8.183 This section will consider Microsoft’s pre-existing position in first and third
party content.

Microsoft’s position on first and third party content

Parties’ views

8.184 Microsoft submitted that its first and third party content does not give it an
advantage over SIE and Nintendo, given these players’ strong content
portfolios. It also submitted that its cloud gaming competitors, including
NVIDIA, Amazon, and Netflix, will have access to a large, diverse, and high-
quality pool of content for their cloud streaming services. 964 Microsoft also
stated that it agrees that content is an important factor driving the value of
game streaming services, although Microsoft clarified that it is not the only
factor. 965

8.185 Microsoft also submitted that its agreement with NVIDIA to allow streaming of
all Microsoft titles on GFN eliminates any potential content advantage, and
that Microsoft is at a clear disadvantage compared to Nintendo and SIE on
content. 966

963 CMA, phase 1 Decision, 12 October 2022, paragraph 12

964 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 1.10 and 3.67.
965 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.48(b).
966 Microsoft Annex 4 to response to the Provisional Findings.

237
 Our assessment

• Internal document evidence

8.186 Microsoft’s internal documents suggest that rival cloud gaming service
providers without an existing gaming console may lack the games and
relationships with third party game publishers to compete effectively in cloud
gaming:

(a) One Microsoft internal document from October 2019 shows that []. The
document explains that []. We note that this document pre-dates the
agreement of the Merger. 967

(b) An email to the [] in March 2020 explains that rivals like []. According
to the email, these rivals are []. The email explains that this is []. 968

(c) Another document from April 2020 describes gaming content as []. 969

8.187 We note that these documents precede Microsoft’s acquisition of Zenimax, so

Microsoft’s position on content is likely to have been strengthened since their
creation.

8.188 Further internal documents described below on Microsoft’s overall strengths in

cloud gaming also consistently highlight content as an area of strength. In
these documents [].

• Third party evidence

8.189 Several rivals highlighted Microsoft’s content library as a significant

strength: 970 One described how Microsoft had a large catalogue of existing
titles, with access to its own game development studio to increase this
offering. 971 Another noted Microsoft’s significant content advantages,
including a leading first-party exclusive content library. 972 Another stated that
Microsoft owns a significant library of must-have content. 973 And another
described that XGP offered a robust game library, including games from
companies Microsoft has already purchased. 974

967 Microsoft Internal Document.

968 Microsoft Internal Document.
969 Microsoft Internal Document.
970 Third party responses to the CMA questionnaire: [], [], [] and []
971 [] response to the CMA questionnaire.
972 [] response to the CMA questionnaire.
973 [] response to the CMA questionnaire.
974 [] response to the CMA questionnaire.

238
8.190 One rival [] described how Microsoft has access to exclusive gaming
content, through its pre-merger ownership of 23 first party game studios. 975

8.191 An internal document from one competitor [] in October 2021 describes
how Microsoft has leveraged the Game Pass game catalogue for xCloud, and
has secured day and date releases from AAAs like Square-Enix in addition to
putting all 1P content into Game Pass on day one. 976

8.192 A third party report described Microsoft as best positioned for subscription
services generally, and for cloud gaming specifically as a large first-party
content library will help the platform scale up faster and with better marginal
economic benefits. The document explains that the cost of technology (ie
processing and bandwidth costs) increases broadly with user growth, whilst
the marginal cost of content can drop significantly if more subscribers join the
platform. 977

• Shares of supply

8.193 As an indicator of the strength of providers' respective first-party content

libraries, we have considered data submitted by the Parties on different
publishers’ shares of supply of PC and console AAA games. 978 Of the current
main global cloud gaming providers, only Microsoft and SIE feature amongst
the top publishers. In 2021 Microsoft had a share of [5-10]% in the UK, and [0-
5]% worldwide. SIE meanwhile had a share of [0-5]% in the UK, and [0-5]%
worldwide. In our view this demonstrates that both providers have a strong
library of first party content compared to other rivals.

8.194 Of the other main competitors, although Amazon has a first-party studio, it has
published relatively few successful games, and therefore does not feature
amongst the top listed publishers (indicating that it has a less than []%
share worldwide). 979 Other rivals such as NVIDIA and Boosteroid have no
first-party content at all.

8.195 Although Nintendo only has a limited cloud gaming offer currently, it has a
higher share of supply of PC and console AAA games than both Microsoft and
SIE, with a share of [10-15]% in the UK in 2021, and [5-10]% worldwide. 980

975 [] response to the CMA’s RFI.

976 [] Internal Document.
977 Microsoft Internal Document.
978 Microsoft response to the CMA’s RFI.
979 Microsoft response to the CMA’s RFI.
980 Microsoft response to the CMA’s RFI.

239
 This may indicate that Nintendo has the strongest position on first party
content, should it expand its cloud gaming offer.

8.196 We note that there are difficulties in defining AAA games and that this will
impact the described shares, however using shares of supply for all PC and
console games does not alter the shares substantially or change the
assessment of the strength of Microsoft and SIE relative to other cloud
gaming providers. The data is also compiled from various third party market
intelligence sources meaning it may not be a completely accurate reflection of
actual market shares, however we believe it is useful as a guide to the relative
strength of different publishers.

• Conclusion on Microsoft’s position on first- and third party content

8.197 Based on the evidence, we believe that the existing content portfolio and
developer/publisher relationships that Microsoft has built over time may give
Microsoft a significant advantage over cloud gaming services rivals without
similar assets. Of current rivals, only SIE and Nintendo have comparable
strength in this area.

Microsoft’s overall strengths in cloud gaming services relative to rivals

8.198 We believe that Microsoft’s strengths in cloud gaming services should not be
assessed in isolation. The evidence suggests that the combination of
Microsoft’s multi-product ecosystem gives it a stronger position in cloud
gaming than would be suggested by assessing each of its products and
services individually. As described above, this may affect Microsoft’s incentive
to engage in a foreclosure strategy using Activision’s content, and may also
magnify the effect of any such foreclosure strategy in the market for cloud
gaming services.

8.199 Microsoft is not, however, the only industry participant with a combination of
assets that could be leveraged for cloud gaming. There are rivals with their
own multi-product ecosystem, several of which have entered, are present, or
intend to enter the market for cloud gaming services. These include Amazon,
SIE, NVIDIA, Nintendo, Google, [].

8.200 In this section, we first consider the potential strengths that each of these
rivals has. We then consider how these compare to Microsoft’s advantages
when assessed in the round.

240
Advantages of Microsoft’s rivals

Amazon Luna

8.201 Amazon’s key assets for cloud gaming services include:

(a) AWS cloud infrastructure. As described above, []. Similar to Microsoft

with Azure, this can be leveraged to provide cloud gaming services at a
low cost.

(b) Twitch. Twitch is an interactive livestreaming service for content, primarily

used for gaming ie watching streams of others playing games. 981 As
described above this gives Amazon access to a large community of
consumers interested in gaming.

(c) Promotional channels. As well as Twitch, Prime Video and Amazon’s

retail site offer useful channels for promoting its cloud gaming service.

(d) First and third party content. Amazon has a game development studio
and has published the games Lost Ark and New World (although neither
is currently available on Luna). 982 This game development studio is not
comparable to rivals such as Microsoft, SIE, and Nintendo in the volume
and range of first party content. On third party content, Amazon has been
successful in placing some third party content on Luna (eg, from
Ubisoft). 983

SIE PlayStation Plus

8.202 SIE’s key assets for cloud gaming services include:

(a) PlayStation hardware and OS. Similar to Microsoft, SIE has the option
to use console hardware and the associated OS to stream games,
providing a lower cost solution with a large library of compatible games.
As discussed above, however, there are cost advantages in using PC
servers in the long term, so any advantage that SIE may have by using its
console hardware for cloud gaming services is not likely to extend into the
future.

(b) First and third party content. SIE publishes several high profile and
popular game franchises such as God of War, The Last of Us and
Uncharted, and owns several studios capable of developing high quality

981 ‘Twitch', accessed by the CMA on 15 December 2022.

982 ‘Amazon Games’, accessed by the CMA on 7 December 2022.
983 ‘Amazon Luna – Ubisoft’, accessed by the CMA on 24 January 2023.

241
 games. 984 Like Microsoft and Nintendo, it also has strong relationships
with third party developers and publishers, including exclusive titles.

NVIDIA GFN

8.203 NVIDIA’s key assets for cloud gaming services include:

(a) GPUs. NVIDIA is one of two leading manufacturers of GPUs, which are
an essential input into the cloud infrastructure required for cloud gaming.
Access to these at lower cost than rivals provide NVIDIA with an
advantage in developing its cloud infrastructure and this has helped it
develop its own infrastructure without having an existing cloud network. 985

(b) Third party content. GFN has a large library of third party content
available on its service with over 1,400 games available. 986 However we
note that these games are not part of NVIDIA’s service and must be
purchased by the consumer separately, and that it does not have any
exclusive titles.

Nintendo

8.204 Nintendo’s key assets for cloud gaming services include:

(a) First- and third party content. Nintendo publishes several high profile
and popular game franchises such as Mario and Zelda. 987 Like Microsoft
and SIE it also has strong relationships with third party developers and
publishers, including exclusive titles.

8.205 Whilst Nintendo has its own console hardware that could be used as cloud
infrastructure, given its lower grade specification, it would not be suitable for
streaming higher performance games, and has not been used by Nintendo for
its cloud gaming offer.

Google Stadia

8.206 Although Google shut down its cloud gaming service, Stadia, it is worth
considering Google’s pre-existing strengths, given that these were insufficient
for it to succeed in cloud gaming. Google’s key assets for cloud gaming
services include:

984 'TechRaptor article', accessed by the CMA on 15 December 2022.

985 ‘NVIDIA website’, accessed by the CMA on 15 December 2022.
986 'Games available on NVIDIA GeForce NOW', accessed by the CMA on 7 December 2022.
987 ‘PlayDiaries article’, accessed by the CMA on 15 December 2022.

242
 (a) GCP. As described above, Google is the [] provider of cloud
infrastructure services []. Similar to Microsoft with Azure, this can be
leveraged to provide cloud gaming services at a low cost.

(b) Promotional channels. As the owner of YouTube and the Google Play
Store, Google has useful channels for promoting Stadia to consumers.

[]

8.207 [] is a potential entrant to cloud gaming. [] key assets for cloud gaming
services include:

(a) []. 988

Microsoft’s overall strengths relative to rivals

Parties’ views

8.208 Microsoft submitted that it does not have any pre-existing advantage or
strength in cloud gaming due to its ecosystem, and that whilst it has several
assets or products, these products are not integrated or linked and cannot be
used to cause the market for cloud gaming services to tip. 989

8.209 Microsoft also submitted that it does not hold a uniquely strong position in
cloud gaming services. It submitted that NVIDIA has become the leading
cloud gaming provider without having any first party content. It stated that
both SIE and Amazon have first party content similar to Microsoft. 990 It has
also suggested that Microsoft is behind both Amazon and Google on cloud
infrastructure. 991

Our assessment

• Internal document evidence

8.210 Internal documents from Microsoft since 2019 describe the key inputs and
assets required for cloud gaming service providers and indicate that Microsoft
has a strong position in the market due to its strength across the full set. For

988 [] call note.

989 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.17.
990 Microsoft response to working papers.
991 Microsoft response to working papers.

243
 example, one Microsoft internal document displays a chart showing that
[]. 992

Figure 8.1 Microsoft internal cloud gaming competitive assessment

[]

Source: Microsoft internal document.

8.211 We consider that developments since the document was produced have not
significantly altered the situation. Notable developments include:

(a) Microsoft has strengthened its position on first party content with the
acquisition of Zenimax Media;

(b) Amazon has increased third party content available on Luna with, for
example, Ubisoft games now available. 993 Amazon has also published
first party games Lost Ark and New World, although neither is currently
available on Luna; 994

(c) Electronic Arts has not launched a cloud gaming service;

(d) Google announced the winding down of Google Stadia in September

2022, with the service closing in January 2023; 995

(e) NVIDIA has increased third party content accessible on GFN (over 1,400
games available), 996 and has developed cloud infrastructure utilising
NVIDIA’s GPUs; 997

(f) SIE has []. 998

8.212 Another Microsoft draft internal document [], finding that []. 999 Our view is
that both of these disadvantages are being addressed through []. Microsoft
submitted that this document had not been finalised and was not used as the
basis for any decisions by Microsoft’s Gaming Leadership Team.

8.213 An internal document from early 2021 notes that Microsoft is []. 1000 [].
However we note that Microsoft’s [].

992 Microsoft Internal Document.

993 ‘Ubisoft+ Now Available on Amazon Luna’, accessed by the CMA on 7 December 2022.
994 ‘Amazon Games’, accessed by the CMA on 7 December 2022.
995 ‘A message about Stadia and our long term streaming strategy’, accessed by the CMA on 7 December 2022.
996 'Games available on NVIDIA GeForce NOW', accessed by the CMA on 7 December 2022.
997 [] response to the CMA’s RFI.
998 [] response to the CMA’s RFI.
999 Microsoft Internal Document.
1000 Microsoft's Internal Document.

244
8.214 In another document from 2020 comparing []. 1001

8.215 Another internal document from early 2021 focusing on game development in
the Xbox ecosystem highlights []. 1002

• Third party evidence

8.216 Internal documents from one competitor []. A document from August 2020
describes []. The same document notes that []. 1003 Another document
from the same competitor describes Microsoft as [] of the group of listed
competitors []. 1004

8.217 Another competitor [] described Microsoft as having a unique advantage in

the expected transition to cloud gaming due to its capabilities across a range
of products and services needed for gaming but noted the advantages of
others. Microsoft’s advantages included the Windows OS, its leadership in
consoles, the game developers and publishers it has acquired, and its cloud
infrastructure business. 1005 In comparison it described SIE as having strength
only in content. Amazon was described as having a huge advantage through
its cloud infrastructure network but lacking in content. NVIDIA was noted to
compete by making rapid technological innovations, and constantly adding
cutting-edge technology to its world-class GPU hardware and software but
relies on others for access to content.

8.218 Industry market reports also described Microsoft’s strong position. One report
stated that Microsoft’s strengths across most of cloud gaming's value chain
mean it is much better positioned to succeed than other game companies. 1006
Another third party report described Microsoft as best positioned for
subscription services generally, and for cloud gaming specifically as in-house
cloud infrastructure is expected to provide a critical advantage, and a large
first-party content library will help the platform scale up faster and with better
marginal economic benefits. 1007

1001 Microsoft Internal Document.

1002 Microsoft Internal Document.
1003 [] Internal Document.
1004 [] Internal Document.
1005 [] response to the CMA’s RFI.
1006 [] Internal Document.
1007 Microsoft Internal Document.

245
Conclusion on Microsoft’s potential strengths in cloud gaming

8.219 The evidence supports the conclusion suggests that Microsoft has a multi-
product ecosystem that places it in a very strong position relative to most or
all rivals in the growing market for cloud gaming services.

8.220 We believe that Windows alone is one of Microsoft’s strongest advantages in

this market. Most PC games are developed for Windows and not for other
OSs, meaning that a cloud gaming service running on Windows can
potentially access a vast library of content. Since Microsoft owns Windows, it
has a significant cost advantage over rival cloud gaming providers, who face
substantial licensing fees reported to be around []% of hourly streaming
costs by some providers, and even up to []% by others.

8.221 Alternative OSs are available but have significant disadvantages. There is a
high cost of porting and maintaining games from Windows to Linux. And the
Proton compatibility layer that allows Windows games to be played on servers
running Linux is not seen as sufficiently reliable to guarantee gameplay parity
with cloud gaming services running Windows.

8.222 On cloud infrastructure, Microsoft’s xCloud infrastructure is a cost-effective

short-term solution that gives it some advantage over most competitors, with
the possible exception of SIE []. In the longer term, [], we believe
Microsoft will be able to leverage the full extent of its Azure infrastructure for
its cloud gaming service, providing it with a significant cost advantage over
most rivals. Only Amazon and Google (the latter having shut down its cloud
gaming service) have similar internal cloud infrastructure capabilities. Neither
of them, however, have the combined cost advantage of Azure plus Windows,
(or the content advantages of Microsoft) which places Microsoft in a uniquely
advantageous position. The failure of Stadia demonstrates that even a strong
cloud infrastructure network cannot necessarily compensate for the difficulties
associated with using an alternative OS.

8.223 Finally, we consider Microsoft already has a strong position on first-party

content compared to most competitors, except for SIE and Nintendo. We
believe that content is particularly important to the success of a cloud gaming
service, particularly considering Google’s failure with Stadia, which our
evidence suggests was caused at least in part by a lack of gaming content,
which was connected to its use of a Linux OS.

8.224 We recognise that a few rival cloud gaming service providers also have multi-
product ecosystems that could offer an advantage in cloud gaming services.
Some have an in-house cloud gaming infrastructure (eg, Amazon and
Google), others have access to first- and third party content (eg, SIE), and

246
 others have different cost advantages, such as GPUs (eg, NVIDIA).
Consistent with Microsoft’s own internal documents, however, we believe that
none of these rivals can match Microsoft’s advantage arising from its
ownership of Windows, Azure, and the Xbox gaming catalogue combined. As
such, we believe that Microsoft is already in a uniquely strong position in the
market for cloud gaming services.

8.225 In the next sections, we first consider the evidence on the extent to which
Activision would have made its content available on cloud gaming platforms
absent the Merger, then assess whether the Merger would give Microsoft the
ability and incentive to use Activision’s content to foreclose rivals, and
consider what effect, if any, this would have on competition.

Availability of Activision’s content on cloud gaming services

8.226 As of April 2023, Activision content has only been available to cloud gaming
platforms to a limited extent, in particular during the beta testing phase of
NVIDIA GFN and has since been removed.

8.227 In this section we consider evidence on the extent to which Activision would
have made its content available on cloud gaming platforms absent the
Merger. In particular, we consider:

(a) The likely growth of cloud gaming services. The greater the growth, the
greater we would expect Activision’s incentive to be to make its content
available on these services.

(b) Whether latency is a challenge likely to be overcome in cloud gaming

services within the next five years.

(c) Additional evidence on Activision’s incentives to make its content

available on cloud gaming services, including from the Parties’ internal
documents and third party evidence.

Parties’ views

8.228 Microsoft submitted that Activision content would not be available via cloud-
based game streaming absent the Merger. It submitted that the evidence
shows that Activision does not consider that cloud-based game streaming
constitutes a good value proposition as it has major reservations about
[]. 1008

1008 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 4.29.

247
8.229 Microsoft also submitted that [] due to []. In relation to any success that
Fortnite—another shooter that has been made available for streaming on
cloud gaming platforms—has enjoyed, Microsoft submitted that []. 1009

8.230 Activision submitted that the technical limitations of cloud gaming mean []. It
further submitted that it considers cloud gaming []. Activision also submitted
that cloud gaming is a transient technology and that computing power of
consumer electronics hardware – in particular mobile phones – is developing
so rapidly that it will soon [], further constraining the future growth and
reach of cloud gaming. 1010

8.231 Activision submitted that [] has [] setting out a strategy or plan to do so in
the future. 1011

8.232 Activision submitted that, [] – especially at the highest levels, whose
approval would be required – have never been convinced that []. Moreover,
Activision Blizzard has no involvement in cloud gaming today. 1012

8.233 Activision also submitted an excerpt from the Main Parties Hearing, where
[] stated: ‘[].’ 1013

8.234 Microsoft submitted that []. It also submitted that Activision’s []. 1014
Similarly, Activision submitted that []. 1015 On the same point, Activision
submitted that CoD: Mobile has ~[]% MAUs of the total MAUs of the
franchise, and that mobile gaming revenues from the King division and titles
such as CoD: Mobile, as well as ancillary revenue, represented approximately
47% of Activision Blizzard’s revenues in 2022. Activision also submitted that it
will []. 1016

8.235 We have already presented the Parties’ arguments on the availability of

Activision’s content on MGS services in Chapter 7. We consider that these
arguments, such as Microsoft’s submission that MGS sales would cannibalise
Activision’s B2P sales, are relevant for those cloud gaming platforms that
have an MGS-based business model.

8.236 Before setting out our assessment, we make the following observations
relevant to the Parties’ submissions set out above:

1009 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 5.42-5.44.
1010 Activision response to the Provisional Findings, 2 March 2023, paragraph 3b.
1011 Activision response to the Provisional Findings, 2 March 2023, paragraph 6.
1012 Activision response to the Provisional Findings, 2 March 2023, paragraph 26.
1013 Activision response to the Provisional Findings, 2 March 2023, paragraph 7.
1014 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.38.
1015 Activision response to the Provisional Findings, 2 March 2023, paragraph 3b.
1016 Activision response to the Provisional Findings, 2 March 2023, paragraph 22.

248
 (a) While we have considered in our assessment the Parties’ submissions to
the CMA regarding Activision’s perceptions of cloud gaming opportunities,
including Activision’s statements to the CMA in the context of this merger
investigation, we have done so alongside other evidence, including in
particular from internal documents generated in the ordinary course of
business (which are less susceptible to being coloured by the merger
process).

(b) Even if it was the case that, instead of pursuing cloud gaming, Activision’s
[], that would not be inconsistent with Activision being willing to add its
games to cloud gaming services. We also note that several F2P games
are currently available on cloud gaming platforms (eg Fortnite and Apex
Legends on NVIDIA GFN). 1017 Plus, as we noted in the section on future
development of cloud gaming above, whilst the computational power of
mobile devices has increased and may continue to increase, there is likely
to continue to be demand for cloud gaming which can also offer higher
performance from hardware housed in data centres. Playing via cloud
also offers additional benefits such as using less battery life as less
processing is happening locally.

(c) The fact that Activision’s games are not on cloud gaming services today
and have been to a limited extent in the past does not imply they will not
be in the future. In particular, any discussions of cloud gaming
opportunities within Activision and between Activision and cloud gaming
providers would show that Activision is still interested in cloud gaming,
even though its games are not on one of those services yet.

Our assessment

Future developments in cloud gaming

8.237 We identified a number of likely developments in cloud gaming that we

consider would contribute to an increase in the incentive for publishers to
make their content available on cloud gaming platforms over time.

8.238 First, as set out in our assessment above, we found that cloud gaming is likely
to continue to grow and is also likely to become profitable in the next five
years. We consider that growth in the size of the cloud gaming market would
make placing content on cloud gaming platforms increasing attractive.

1017 Play Your Games Anywhere | GeForce NOW | NVIDIA, accessed by the CMA on 30 March 2023.

249
8.239 Second, we received evidence that MGS-based cloud gaming platforms are
[]. We consider that this would further increase Activision’s incentives to
add content to those platforms, given [] would not lead to any
cannibalisation of game sales by MGS.

(a) One Microsoft internal document dated April 2022 shows that []. 1018

(b) An internal document provided by a competitor [] shows that []. 1019
[]. 1020

8.240 Third, with respect to latency and technical quality of games played on cloud
gaming platforms, third party evidence suggests that these are not an issue or
that any outstanding issues will be solved soon. To the extent any such issues
imply that adding Activision’s content to cloud gaming platforms would create
a risk of harm to the perceived quality of that content, progress in reducing
latency and improving technical experience on cloud gaming platforms would
reduce any such risk. In particular:

(a) A competitor [] submitted that cloud gaming outperforms most PCs and
consoles and is technologically viable for all game genres. 1021 The same
competitor [] submitted that technological barriers to streaming,
including latency, were quickly dropping and were likely to continue to
drop. It described how it had reduced latency on its service []. 1022 As
described above, evidence from this competitor [] on play time for
different games showed that multiplayer and ‘fast-twitch’ games are
amongst the most popular, indicating that latency does not present a
problem for streaming these types of games. 1023

(b) Another competitor [] submitted that most players on most internet
connections can now have a great quality gaming experience via cloud
gaming. It further submitted that there are many viable ways to make
cloud gaming economically and technically viable in the current state. 1024

(c) Another competitor [] submitted that there are challenges to be solved
in cloud gaming with respect to latency. However, this competitor
submitted that it is a matter of time before these challenges get solved.
According to this competitor, [] would solve the latency issue. 1025 The

1018 Microsoft Internal Document.

1019 [] Internal Document.
1020 []
1021 [] submission to the CMA.
1022 [] response to the CMA’s RFI.
1023 [] Internal Document.
1024 [] call note.
1025 []call note.

250
 same competitor stated that ‘fast-twitch’ multiplayer games such as Call of
Duty or Fortnite are suitable for cloud gaming services due to
advancements in home networking performance, and that there may be
some benefit to playing these types of games through a cloud gaming
service because the high performance and reliability of cloud networking
can increase the scalability (eg, number of players) in a multi-player
game. It also stated that it had tested such games on its service and
believes it offers competitive performance. 1026

(d) As set out in the section below in which we consider other publishers’
behaviour in relation to adding content to cloud gaming platforms, a
number of publishers have already listed games in which reaction times
are relevant to gameplay and low latency is therefore likely to be
beneficial to gamers’ experience. This includes shooter games such as
Battlefield 2042, Fortnite, and Tom Clancy’s Rainbow Six Extraction, as
well as other games, such as FIFA.

8.241 We have taken these anticipated developments in cloud gaming into account
in our assessment below. This is particularly relevant to our assessment of
historical evidence, including evidence from Activision’s contemporaneous
internal documents. In particular, the observation that Activision has not yet
licensed its content to cloud gaming platforms will not tell the full picture,
particularly in the context of developments that will tend to increase its
incentive to do so.

Publishers’ behaviour on adding content to cloud gaming

8.242 Activision’s incentives are likely to be impacted by what other publishers do,
any additional costs in porting games to cloud gaming platforms, and the
extent of any financial incentives offered by cloud gaming providers. We have
considered the approach that rival publishers have taken to placing games on
cloud gaming services. We consider that this is relevant evidence when
assessing Activision’s likely incentive to do so.

8.243 Rival publishers such as Ubisoft, Electronic Arts, and Epic Games already
offer a range of games, including AAA titles and new releases, on cloud
gaming services. They offer these games either on a standalone basis or
through their own MGS bundles. In particular:

(a) Electronic Arts currently offers EA Play as part of Game Pass Ultimate,
which allows subscribers to stream several games via xCloud. It also

1026 [] response to the CMA’s RFI.

251
 offered a selection of games, such as the latest FIFA day and date, on
Google Stadia (before it was shut down), and offers several games,
including the latest Battlefield release (Battlefield 2042, an AAA shooter
game) on NVIDIA GFN. 1027

(b) Ubisoft offers several games, including latest releases and day-and-date
AAA titles on Amazon Luna through its Ubisoft+ subscription and through
a BYOG feature. 1028 Ubisoft also has several games available on NVIDIA
GFN – these include latest releases of AAA games such as Assassin’s
Creed Valhalla, Tom Clancy’s Rainbow Six Extraction, and Far Cry 6. 1029

(c) Epic Games offers several games, including Fortnite, on NVIDIA GFN.

8.244 Activision has also offered games – including CoD – on cloud gaming
services in the past. It offered games on NVIDIA GFN when this was in the
beta phase, before they were removed from the service for reasons that were
commercial in nature and not evidently related to technical performance. 1030
These were both back catalogue titles and day-and-date releases of the latest
titles, including various CoD games 1031.

8.245 This evidence shows that some rival publishers have been willing to make
their latest AAA games, including shooters, available on third party cloud
gaming services both on a MGS (Amazon Luna) and BYOG (Amazon Luna
and NVIDIA GFN) basis, and in some cases even on a day-and-date basis
(on Google Stadia).

Activision’s internal documents

Activision’s general stance on cloud gaming

8.246 Although some Activision internal documents show that it has []. In
particular:

1027 Launch & Play EA SPORTS™ FIFA 23 in Seconds - Stadia (google.com), accessed by the CMA on 17
January 2023; we note that Google announced it is closing down Google Stadia in January 2023, even though
these games were still available to play in the meantime: A message about Stadia and our long term streaming
strategy (blog.google); Play Your Games Anywhere | GeForce NOW | NVIDIA, accessed by the CMA on 17
January 2023.
1028 Play Ubisoft Games You Own on Amazon Luna, accessed by the CMA on 17 January 2023.
1029 Play Your Games Anywhere | GeForce NOW | NVIDIA, accessed by the CMA on 17 January 2023.
1030 Activision Blizzard pulls all games from GeForce Now a week after launch | GamesIndustry.biz, accessed by

the CMA on 2 February 2023, and Nvidia says a 'misunderstanding' led to Activision's departure from GeForce
Now | PC Gamer, accessed by the CMA on 2 February 2023.
1031 For example, we understand that Call of Duty: WWII was previously available day-and-date on NVIDIA GFN.

Source: [] response to the CMA’s RFI.

252
 (a) One internal document dated May 2020 notes that [].1032 We note this
document was a working draft of [] and that the final version did not
include [].1033

(b) A document dated October 2020 from Activision’s [] team provides an
overview of cloud gaming in general and []. The document states that
[].1034

(c) The same document also explores different ways for Activision to [].
The document explains that []. The document also notes that [].1035

(d) The same document also sets out []. It explains that []. It adds that
[]. However, it notes that, in the long term, Activision should []. 1036

(e) A draft presentation by [] in March 2021 discusses []. The document
notes that Activision has []. The document also notes that all of []
when it comes to technological capabilities. However, overall the
document notes that [].1037

(f) A document dated August 2021 on strategic planning notes that cloud
gaming is emerging and will []. It adds that Activision should []. The
same document expresses [].1038

(g) In an email dated January 2022, []. The document considers [].1039

8.247 Activision submitted that these documents show that []. 1040

8.248 Activision also submitted that the documents above are either [], and that
[]. 1041 For example, Activision submitted that the []. 1042 We note that the
employee who [] in this email chain is [], a senior Activision employee in
a team responsible for []. We consider that the job title and position of this
employee within Activision give this employee the ability to influence
Activision’s strategy in relation to cloud gaming.

1032 Activision Internal Document.

1033 Activision Internal Document; and Activision Internal Document.
1034 Activision Internal Document.
1035 Activision Internal Document.
1036 Activision Internal Document.
1037 Activision Internal Document.
1038 Activision Internal Document.
1039 Activision Internal Document.
1040 Activision response to working paper.
1041 Activision response to the Provisional Findings, 2 March 2023, paragraphs 33-35.
1042 Activision response to the Provisional Findings, 2 March 2023, paragraph 35.

253
8.249 We note the Parties’ submission that [] cloud gaming. 1043 However, as
noted above, draft versions of these documents show that Activision
employees []. We also consider that, as cloud gaming grows over the next
five years, the likelihood of it being reflected in [] increases significantly.

8.250 In relation to [] stance on cloud gaming, we note that the evidence we have
seen from Activision internal documents in this respect is mixed. In particular:

(a) One internal Activision email by [] from January 2020, which discusses
[], notes that [] is ‘[]’. The document also notes that []. 1044 As the
email is focused on [], it is not clear whether [] stance as emerging
from these statements refers to cloud gaming as a whole or is instead
[]. For example, another document from September 2020 discussing
[] refers to ‘[]’. 1045 This document suggests that any concerns by []
around [].

(b) Another internal Activision email chain from March 2020 discusses a []
on cloud gaming that did not come to fruition and involves multiple senior
employees. []. In response, [] notes: ‘[]’. In turn, []responds to
the email chain by noting: []. 1046 This document suggests that [] was
[], and that senior Activision employees [] which they believed were
aligned with [] stance.

8.251 Regardless of the general stance on cloud gaming as emerging from the
documents above, we have seen evidence of Activision, including Activision’s
senior leadership and [], actively discussing potential opportunities with
cloud gaming providers in the last three years. These are assessed more in
detail below.

8.252 More generally, we consider that the main factors currently acting as a drag
on cloud gaming, and therefore on Activision’s interest in it, are the scale of
the market (and therefore the scale of the opportunity) and latency in cloud
gaming services. On the former, we have assessed what the future of cloud
might look like in detail above. We found that cloud gaming is likely to
continue to grow and is also likely to become profitable in the next five years.

8.253 With respect to latency, some Activision documents raise questions and
doubts in relation to certain cloud gaming providers ([]) – these documents
are discussed more in detail below. However, with the [], we consider that
none of these documents or the documents above raise []. In fact, one

1043 Activision response to the Provisional Findings, paragraph 26.

1044 Activision Internal Document.
1045 Activision Internal Document.
1046 Activision internal document.

254
 Activision internal email by [] from February 2021 states that []. The
document recognises that [] as []. However, in the same document []
states that []. The document also states that []. 1047 On this latter point, we
note elsewhere in this section other Activision documents which suggest that
cloud gaming will []. 1048

8.254 Overall, we consider that Activision’s stance on cloud gaming as emerging

from these documents is positive. This is despite some documents pointing
towards scepticism from [] to partnerships with cloud gaming providers,
[]. In particular, as we assess below, we have seen evidence of Activision,
including Activision’s senior leadership and [], actively discussing potential
opportunities with cloud gaming providers in the last three years.

8.255 We now turn to internal documents relating to interactions with specific cloud
gaming providers, including [].

[]

8.256 Activision’s internal documents show that []. The documents show senior
Activision employees were involved in these discussions. In particular:

(a) One email from Activision to [] dated December 2021 []. 1049

(b) A subsequent internal Activision email exchange from January 2022

shows that []. The document shows that he planned []. 1050

(c) Another internal Activision email exchange dated December 2021

discusses []. The document shows that Activision was considering
[]. 1051

(d) Another internal Activision email exchange from January 2022 shows that
Activision was considering []. 1052

(e) Another internal Activision email exchange from February 2022 discusses
[] and notes that Activision ‘[].’ The document also notes that
Activision’s ‘strategy of []’ and that ‘There’s no need to mention
anything related to [] as this is our regular course of business’. 1053

1047 Activision Internal Document; on this point, see also Activision Internal Document which is discussed more in
detail below.
1048 See for example: Activision Internal; Activision Internal Document; and Activision Internal Document.
1049 Activision Internal Document.
1050 Activision Internal Document.
1051 Activision Internal Document.
1052 Activision Internal Document.
1053 Activision Internal Document.

255
 (f) However, a subsequent internal Activision email from February 2022 from
an Activision employee to [] discusses the [].1054

(g) Similarly, a subsequent Activision document from February 2022 contains

a slide discussing the impact of the acquisition of Activision by Microsoft to
Activision’s current deals and pipeline. With respect to [], the document
notes that [].1055 The document appears to be an addendum to a
previous Activision presentation dated January 2022, titled ‘[]’, and
meant for review by the Activision CEO. This presentation states that one
of the []. The same document also notes with respect to launching CoD
Warzone on [] that [].1056 This version of the presentation appears to
have been reviewed by [],1057

8.257 Activision internal documents also suggest that Activision senior

management, including []. This was a follow-up to []. These documents
show that Activision was []. 1058 [].

8.258 Activision submitted that it has had the opportunity to consider []. 1059
However, Activision did not provide any evidence in support of these claims.

8.259 With respect to the February 2022 email exchange from an Activision
employee to [] presented above, Activision submitted that this document
was created by [] who relied upon outdated and unreliable information. 1060
However, as noted above, the information contained in that document is
confirmed in a subsequent document which appears to have been an
addendum to a presentation reviewed by a senior Activision employer and
meant for review by the CEO.

8.260 With respect to the discussions with [] in [], Activision submitted that the
communication with [] was never presented to [] to enter any cloud
streaming deal. 1061

8.261 However, as we noted above, certain Activision employees, including at a

senior level, were involved directly or indirectly in discussions with [] when
the Merger was announced in January 2022. 1062’ The Activision documents
presented above strongly suggest that these discussions were [].

1054 Activision Internal Document

1055 Activision Internal Document.
1056 Activision Internal Document.
1057 Activision Internal Document.
1058 Activision Internal Document; and Activision Internal Document.
1059 Activision response to working paper,.
1060 Activision response to the Provisional Findings, 2 March 2023, paragraph 50.
1061 Activision response to the Provisional Findings, 2 March 2023, paragraph 49.
1062 These senior Activision employees include the [], the [], the person who was [], and [].

256
 Therefore, if these discussions were indeed not presented to either [], []
or [], this may have been [].

8.262 We consider that these documents show that Activision was []. More
broadly, these documents show that, regardless of whether Activision senior
leadership would have ultimately [], Activision has had [], at a senior level
(including, [], the [] and [] themselves). We consider that these
documents are consistent with latency and other potential obstacles to
streaming Activision content as being surmountable.

[]

8.263 The [] between [] are discussed in the section on the availability of
Activision’s content on MGS services in Chapter 7. These [] included [].

8.264 The documents relating to these [] appear to show that Activision [].
From these documents, it appears that the main reasons for Activision’s [].

8.265 In relation to []:

(a) An Activision email from [] to himself in January 2020, which appears to
be a draft of talking points to [], notes in relation to a []: ‘[] concern
around Streaming
[]’.1063

(b) Another internal Activision email dated June 2020 from [] to []
regarding [] partnership discussions notes that Activision ‘[].’ The
document also notes that [] ‘reiterated [] understands [] may still
want to go become []’.1064 This suggests that Activision’s reticence to
[] was mainly linked to [].

8.266 In relation to []:

(a) In one internal Activision email exchange from August 2021, [] states,
responding to ‘whether/ how to engage in the [] or not’ that: ‘I don’t
think it is necessary at this time. We are still meaningfully far away on []
from our discussions. In addition, I don’t believe at this point []. I think
this will be good conversation []’. 1065

1063 Activision Internal Document.

1064 Activision Internal Document.
1065 Activision Internal Document.

257
 (b) An Activision email from [] to [] in April 2021, discussing an upcoming
meeting with [] states that [].1066

(c) One Activision document dated February 2021 [].1067

(d) Another draft Activision document from August 2020 discusses benefits
and considerations of adding Activision titles (in particular, CoD Warzone)
to []. Among the benefits, the document recognises: ‘[]’. Among the
considerations, the document notes: ‘[]’.1068

(e) Another Activision document from May 2020 contains an email chain
discussing the request of a CoD developer ([]) to []. In the email
chain, [] notes that ‘[I] assume we only want to do []. As [] says, it’s
not clear at all that [].1069

(f) Another Activision document from January 2020 notes that with respect to
[]. The document also notes that [], and that ‘additional vetting [is]
needed for []’.1070 This suggests that [].

8.267 We note that a Microsoft document dated May 2022 shows that []. The
document notes that [] 1071

8.268 We consider that these documents show that Activision was []. They also
show that even if Activision [], Microsoft was [].

Other providers

8.269 Activision’s internal documents show [].

8.270 One document dated April 2020 discusses a possible [] with []. 1072
However, the document notes that Activision remained available for
discussions, noting that it would hold a follow-up conversation with [] to
hear updates. 1073

1066 Activision Internal Document; See also an email from [] to himself (which appears to be a series of notes)

from January 2019 noting [] position on []: ‘[]. This relationship [between [] and Activision] continues to
go in the wrong direction.’ (Activision Internal Document).
1067 Activision Internal Document.
1068 Activision Internal Document.
1069 Activision Internal Document.
1070 Activison Internal Document.
1071 Microsoft Internal Document.
1072 [].
1073 Activision Internal Document.

258
8.271 Another of Activision’s internal documents dated March 2021 contains an
internal Activision email chain. The email chain discusses a possible []. In
the document, one Activision employee noted that ‘The biggest problem in the
space is that it is []. In any competitive situation, this puts players on the
streaming service at a disadvantage’. However, in the same document []
states ‘I don’t want to make any commitment to [] at this point, [].1074 This
document shows that Activision was cautious about a possible partnership
with [], outlining potential [].

8.272 This evidence suggests that Activision was open to discussing and [],
subject to [], even though the [].

Third party evidence

8.273 We received evidence from several cloud gaming service providers on their
discussions with Activision to bring Activision’s content onto their platforms
over the years up to 2022 before the Merger was announced.

8.274 [] submitted that it had held discussions with Activision to []. It submitted
that, from a technical perspective, it would be possible to do so. 1075 []
engaged in discussions with Activision several times over the last few years,
including after 2019, []. 1076 [] impression was that Activision saw itself as
a leading content provider that would ‘wait and see’ (ie, assess the success of
[] over time) before placing any content on []. 1077

8.275 []. 1078

8.276 One Activision internal document dated February 2021 suggests that the
reason why Activision was not interested in []. This document notes that
[]. The same document also states that the key barrier to []. 1079

8.277 [] submitted that it []. 1080

Conclusion on the availability of Activision’s content on cloud gaming services

8.278 On the basis of the evidence above, we consider that Activision would likely
have made its games – including day and date releases – available for cloud

1074 Activision Internal Document.

1075 [] call note.
1076 [] call note.
1077 []call note.
1078 [] call note.
1079 Activision Internal Document.
1080 [] response to the CMA’s RFI [].

259
 gaming in the next five years. We consider that this is more likely for cloud
gaming services which do not have an MGS-based model, ie those with a
B2P or BYOG model, and note that some []. We consider that this will make
it more attractive for Activision to add its games to those services.

8.279 The evidence suggests that technical obstacles such as latency have either
been overcome already, or they would be overcome in the near future. Cloud
gaming service providers explained that their respective platforms already
offer gameplay on a par to consoles. We consider the fact that Activision
allowed NVIDIA GFN to offer games – including CoD – on its streaming
service during its beta testing phase also suggests that it’s technically
possible to play these games on cloud gaming services.

8.280 We consider Activision’s interactions with cloud gaming service providers

showed [] rather than to any significant concerns with cloud gaming itself.
The evidence also shows an acknowledgment from Activision of the potential
of cloud gaming services to expand the number of gamers Activision can
reach. Indeed, several of Activision’s internal documents suggested [] in
identifying a commercially viable way to distribute its games via cloud gaming.
We consider it likely, in particular, that Activision would have [] in the near
future absent the Merger.

Ability to foreclose

8.281 In Chapter 7, we assessed whether the Merged Entity would have the ability
to foreclose rivals in downstream console hardware and distribution. In doing
that, we assessed both the upstream market power of Activision’s content,
and in particular CoD, to the console gaming market and its importance to
PlayStation in particular.

8.282 We follow a similar framework here, by assessing whether the Merged Entity
would have the ability to foreclose rivals in cloud gaming services. We do so
by looking at Activision’s market power upstream in game publishing and the
importance of Activision’s gaming content to the downstream cloud gaming
services market.

8.283 In doing so, we focus on the whole of Activision’s catalogue across console
and PC. That is because, while Microsoft’s and SIE’s cloud gaming platforms
run on console-based OS (ie, Xbox OS and Orbis OS, respectively), all the
other could gaming platforms run on a PC OS (predominantly Windows or

260
 Linux). We therefore consider the relevance of both console and PC games
for cloud gaming platforms. 1081

8.284 As noted in the section above, Activision content is not available on cloud
gaming services at the moment, but we consider it would likely become
available within the next five years absent the Merger (and that this would
include day and date releases). We also noted that this would be most likely
the case for cloud gaming rivals which have or will have in the future a BYOG
or B2P business model (either standalone or in in combination with an MGS
model). While we noted that older games are more likely to be added to MGS
services, we consider that removing older games would not give the Merged
Entity the ability to foreclose rivals, given that it is mostly new releases that
drive downstream competition.

8.285 Moreover, in the market definition section we noted that we do not think that
firms that are solely focused on bringing mobile-quality games to cloud are a
meaningful constraint on cloud gaming services that provide console-quality
and/or gaming PC-quality games (high-performance games). In the following
sections, we focus on the latter, that is games that would not typically run
natively on mobile. As is evident from the sections below, high-performance
games are the focus of competition within cloud gaming services and
therefore the focus of our competitive assessment.

8.286 In this section, we assess whether, absent the Merger, Activision’s content
would become an important input for cloud gaming as the market develops
and this content becomes available. In doing so, we focus primarily on the
importance of this content for rivals who have, or are likely to have, a BYOG
or B2P business model (although we note that content, and AAA content in
particular, will be no less important for a cloud MGS business model). Given
that the importance of Activision’s content must be considered relative to the
full range of available games (ie, any ‘alternatives’), we consider in the same
section the importance of Activision’s content and alternative games. 1082

Importance of Activision content

8.287 The Parties offer a range of games available on console and PC:

(a) We have discussed the Parties’ console games in the framework section
of the previous theory of harm. They include CoD, Overwatch (another

1081 We do not focus on mobile content as no cloud gaming platform currently runs on a mobile OS. This means
that the content offered on cloud gaming services (almost all of which is games designed for console and PC) is
generally very different from mobile content (games designed specifically for mobile).
1082 Ie we do not consider non-Activision games in a separate section. Rather they are considered alongside

Activision games when discussing evidence.

261
 FPS game), and Diablo (an Action RPG / MMORPG). These
games/franchises are all available on PC.

(b) In addition, Activision publishes PC-only AAA game franchises including

WoW and Hearthstone (an online card game within the Warcraft
franchise).

8.288 We consider that Activision’s full range of content is relevant to our

assessment of this theory of harm. We do not, however, consider it necessary
to carry out a detailed assessment of the importance of each individual game
for cloud gaming services. We focus primarily on the importance of
Activision’s most popular titles, such as CoD and WoW, which provide an
indication of the importance of Activision’s full gaming catalogue. We also
highlight other Activision titles where they are relevant to our assessment.

Parties’ views

8.289 Microsoft submitted that Activision’s content would not give it a material
advantage in cloud gaming services or deter entry. According to Microsoft:

(a) Activision’s reluctance to make its content available through cloud gaming
services has not deterred several companies from launching cloud
gaming services (such as PS+, Amazon Luna and NVIDIA GFN) in recent
years. 1083 The Parties also noted that Steam is the market leader in PC
distribution without having CoD. 1084

(b) The primary use case for cloud gaming today relates to ‘try before you
buy’ functionality, and there is no sense in which the addition of Activision
content could offer any advantages in this regard compared to rivals. 1085

(c) There are challenges to making CoD available for streaming. []. 1086 []
and that Microsoft’s tests show that []. 1087

(d) Rival cloud gaming providers were also successfully developing their own
first-party content, and that there was no suggestion that CoD was seen
as ‘unique’ or ‘special’ amongst Windows PC games. 1088 In relation to

1083 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.40; and Microsoft
response to working papers.
1084 Microsoft response to working papers.
1085 Microsoft response to working papers.
1086 Microsoft response to working papers.
1087 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.72.
1088 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.41.

262
 WoW, as a single-game subscription game, it is not an obvious candidate
for inclusion in a BYOG or B2P cloud gaming service. 1089

(e) Popular alternatives to CoD are available for cloud gaming. Microsoft
submitted that [] as well as seven of the top 10 most played games on
Steam are already available on GFN. 1090 It also submitted that the range
of games available for cloud gaming is only likely to increase in the future,
at least on the CMA’s interpretation of the evidence. It submitted that in
particular, the CMA’s counterfactual assumes that game publishers will be
increasingly incentivised to place their content on cloud gaming services
in the future as those services continue to grow. 1091

(f) The CMA’s approach to the evidence on the counterfactual fails to

recognise that it contains a logical flaw in that if it were the case that
without Activision content cloud streaming would be successful as a
means of delivering games, that would be evidence that Activision content
was not important for the development of cloud gaming. Thus, if the basis
of the counterfactual finding were in fact to be correct, it would show that
the importance of Activision games was wholly overstated and there was
no basis for a foreclosure theory in the first place. 1092

8.290 Microsoft also submitted that given Activision’s share of supply in PC game
publishing and combined PC and console game publishing is less than []%
on any basis, there is clearly a broad range of alternative games already in
existence that gamers play. 1093 Microsoft submitted multiple estimates for PC
and console shares of game publishing:

(a) Its own estimates based on a range of sources: Microsoft submitted that
Activision’s share in console game publishing in 2021 was very low, at [0-
10]% worldwide ([10-20]% in the UK). 1094 It also submitted that the
Merged Entity’s share of PC game publishing in 2021 was only [0-10]%
worldwide ([10-20]% in the UK), meaning that a vast range of other
content (accounting for over [80-90]% of PC game publishing revenue)
would remain potentially available for rival cloud gaming services to
use. 1095

1089 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.48.
1090 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.58.
1091 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.59.
1092 Microsoft response to the CMA’s Supplementary Evidence paper.
1093 Microsoft response to the CMA’s Supplementary Evidence paper.
1094 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 1.5; Parties FMN.
1095 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.41 and Parties FMN.

These are based on figures estimated by Microsoft using internal market intelligence data based on their-party
sources.

263
 (b) Microsoft also submitted shares from Newzoo: 1096 Microsoft submitted
that Activision’s share of game-time on PC in 2022 was limited both in the
UK and on a worldwide basis at []% and []%, respectively, and in
terms of MAUs, Activision’s respective shares were even smaller at []%
and []%. It submitted that on a combined console and PC game
publishing segment, Activision’s share of gametime in 2022 was only
[]% in the UK and []% worldwide, while in terms of MAUs, Activision’s
respective shares were, again, even smaller at []% and []%. 1097

8.291 Finally, since the Provisional Findings, Microsoft has entered into agreements
with NVIDIA, Boosteroid and Ubitus regarding the provision of content to their
cloud gaming services. These are discussed in further detail under the ability
and incentive assessments below. However, with regard to the appropriate
counterfactual against which to assess the Merger, Microsoft has made
submissions regarding the terms of these agreements which relate to [].
Microsoft submitted that this is a significant change in the counterfactual as
set out in the CMA’s Provisional Findings. Microsoft submitted that it is
important in the context of the CMA’s analysis under this theory of harm
where it is the range of available games that is important. Microsoft therefore
submits that the question for the CMA is the extent to which adding Activision
games to the range of content, [] which cloud game providers would
otherwise have. 1098 We take this as a submission that any provisions in the
three agreements that apply regardless of the Merger should form part of the
CMA's counterfactual – the result of this being Microsoft submitting that the
CMA should consider the importance of Activision content in a counterfactual
where []. 1099

Our assessment

8.292 In relation to the Parties’ submissions, we note the following:

(a) The competitiveness of rivals can be harmed even without deterring them
from entering the market altogether. This is particularly relevant for a
growing market like cloud gaming, where foreclosure may have an
adverse impact on the growth of rival cloud gaming services compared to
the Merged Entity’s.

(b) Even if Activision’s share were relatively small, it could still be significant
in a market where a range of inputs (rather than a single input) is an

1096 Microsoft response to the Remedies Working Paper.

1097 Microsoft response to the CMA’s Supplementary Evidence paper.
1098 Microsoft response to the Remedies Working Paper.
1099 This understanding is confirmed at [] of Microsoft response to the CMA’s Supplementary Evidence Paper.

264
 important driver of competition. In markets where downstream customers
buy only one product, a relatively small share of supply upstream
suggests that there are likely to be several alternatives to the necessary
input. Where downstream customers want to access a range of products,
reducing the range (or quality of that range) that rivals are able to offer
may have a significant impact on downstream competition. The impact is
particularly severe where the content making up that relatively small share
of supply is particularly important to that range/quality. This is the case for
console, PC, or cloud gaming services, in which the range and quality of
gaming content available is an important parameter of competition as
discussed in Chapter 7. As the MAGs explain, when assessing the
importance of an input, the CMA may have regard to the role an input
plays as a determinant of product quality. 1100 In any event, a narrow focus
on shares of supply in differentiated markets, such as game publishing
fails to recognise that games with an apparently modest share may have
a more significant role in shaping downstream competition.

(c) As discussed in the market definition section, evidence from Microsoft

suggests that consumers []. This suggests that the use-case for cloud
gaming is not limited to the ‘try before you buy’ functionality, and that most
customers choose cloud gaming for these other reasons.

(d) In relation to WoW, although access to expansions require a monthly

subscription through battle.net, the game currently requires users to have
a PC in order to run it. A cloud gaming service would still benefit from
serving customers who prefer to stream WoW from the cloud, rather than
having to play it on their local PC (just like any other PC game). Any
game-specific subscription payments could be processed through account
linkage to battle.net.

(e) We agree with Microsoft that the range of games available on cloud
gaming services is likely to increase in the future. In the evidence below
we consider Activision games alongside games from other publishers
(including those not currently on cloud gaming services). In relation to the
agreements with [], our assessment already takes this into account, as
our analysis generally assumes game publishers will have increasing
incentives to put their content on cloud gaming services, and that this is
not unique to Activision games. In addition, our analysis on the
importance of Activision's games looks at games across console and PC,
and is not limited to games already available on cloud. To the extent []
further spurs the growth of cloud (absent the merger), that will in turn

1100 CMA129, paragraph 7.14(b).

265
 increase the incentive for Activision to likewise make its games available,
as per our assessment above.

(f) We disagree with Microsoft that there is any logical inconsistency

between the counterfactual and importance of Activision content. The
withdrawal of an important input could harm a competitor’s
competitiveness without causing it to become so unprofitable such that it
goes out of business or impact the growth potential of the industry without
eliminating any growth potential whatsoever. Cloud gaming exists today
and has grown without Activision content and is projected to grow further,
but it is still possible that i) it could grow more rapidly if Activision content
is widely available across the market absent the Merger, and ii) limiting
access to Activision content post-Merger could seriously harm rivals’
competitiveness in circumstances where rivals could have had
widespread access to Activision content absent the Merger.

8.293 We address the rest of the Parties’ submissions by reference to the evidence
below.

Microsoft internal documents

8.294 Evidence contained in the Parties’ internal documents suggest that AAA
content in general, and Activision’s content more specifically, would be an
important input to cloud gaming services in the counterfactual.

8.295 In relation to popular AAA content:

(a) One email from [] suggests that []. 1101

(b) One email from [] discusses []. 1102

(c) One email from an external analyst to Microsoft dated []. 1103 A third
party report also notes that [] 1104

(d) One Microsoft internal document dated [] The document also shows
that []. 1105

8.296 In relation to Activision content:

1101 Microsoft Internal Document.

1102 Microsoft Internal Document.
1103 Microsoft Internal Document.
1104 [].
1105 Microsoft Internal Document.

266
 (a) One report by an investment advisor from January 2021 received by
Microsoft assesses investment opportunities in Activision, Electronic Arts,
and Take-Two. The document states that ‘Activision Blizzard is the best-
positioned company in our coverage space with regard to broader gaming
industry trends. The company’s broad and diverse portfolio of fully-owned
intellectual property across multiple platforms allows Activision Blizzard to
capitalize on industry trends’. The document also states that Activision’s
IP is a ‘must-have for any broad gaming library service, allowing them to
command advantageous terms’. 1106

(b) One internal Microsoft document dated []. The document suggests that
[]. 1107

(c) A number of Microsoft documents discussing customer feedback/survey

results in July 2020 report on the most-requested games for xCloud.
These survey results show that []. 1108 These results were then used in
a monthly programme update to the entire xCloud team. 1109

(d) An Activision internal document dated December 2021 also reports that
NVIDIA claimed it was expecting around [] NVIDIA GFN MAUs to play
[]. 1110 NVIDIA confirmed that it provided Activision with a rough
estimate []. 1111 [] MAUs would have been equivalent to around []%
to []% of NVIDIA GFN MAUs at the time. 1112

Third party evidence

8.297 Cloud gaming service competitors also highlighted the critical importance of
content, particularly AAA titles, to cloud gaming services. In particular:

(a) A competitor [] submitted that [] Google decided to close Stadia
down. 1113 This competitor [] also submitted that for it content represents
around []% of the total cost of its cloud gaming service. It also
submitted that a cloud gaming platform needs a sufficient number of top-

1106 Microsoft Internal Document.

1107 Microsoft Internal Document.
1108 Microsoft Internal Document; and Microsoft Internal Document.
1109 Microsoft Internal Document.
1110 Activision Internal Document.
1111 NVIDIA response to the CMA’s RFI
1112 []. The Parties submitted that [] estimates for the [] would attract was [] and as such are almost

certainly an overestimate of [] own internal estimates. Microsoft response to the Provisional Findings, 2 March
2023, paragraph 3.69 (b).
1113 [] call note.

267
 tier AAA titles from leading game developers and publishers to attract and
maintain a bigger user base. 1114

(b) Another competitor [] submitted that Google Stadia was not successful
because they did not have enough content, which was a result of running
their cloud gaming service on Linux, rather than Windows OS. 1115

(c) Another competitor [] submitted that the success of gaming services
(including cloud gaming) is critically driven by the quality of content. 1116

8.298 These views are supported by competitors’ internal documents:

(a) An internal document from a competitor []. 1117

(b) Another document from that competitor []. 1118

(c) Another document from that competitor []. 1119

(d) An internal document from another competitor [] dated October 2021
attributes []. 1120

(e) Another document from that competitor [] dated April 2021 states that
having enough content is critical to ‘land the content flywheel’. 1121

8.299 Competitors highlighted the importance of Activision’s content, in particular,

for cloud gaming services:

(a) One competitor [] described Activision games as ‘must-have’ for its
cloud gaming service with ‘no meaningful substitute’. 1122 The same
competitor stated that some gamers will not switch from CoD to anything
else because of their investment of ‘time, money, and friends into the
franchise’. 1123

(b) Another competitor [] submitted that CoD: Modern Warfare’s presence
on its cloud gaming platform would ‘have a material effect’ on the players
that its platform could acquire. 1124

1114 [] response to the CMA’s RFI.

1115 [] call note.
1116 [] response to the CMA’s RFI.
1117 [] Internal Document.
1118 [] Internal Document.
1119 [] Internal Document.
1120 [] Internal Document.
1121 [] Internal Document.
1122 [] response to the CMA’s RFI.
1123 [] response to the CMA’s RFI.
1124 [] call note.

268
 (c) Another competitor [] submitted that the failure of Google Stadia shows
the importance of CoD compared to other gaming franchises, and that
CoD’s role in attracting consumers to platforms is not directly
proportionate to engagement alone. This competitor submitted that
Google Stadia did in fact have a reasonable catalogue of games,
including successful franchises such as FIFA, Assassin’s Creed and NBA
2K, which accounted for at least []% of PlayStation engagement in
2021. It further submitted that, despite having these games, Google
Stadia did not have CoD, and that this prevented Google Stadia from
reaching a meaningful number of MAUs. 1125

(d) Another competitor [] noted that, based on their knowledge and
communication with customers the gaming community also asks for
certain titles (including CoD) to be widely available in the cloud. The other
games that are also most important in attracting customers according to
this provider are [] and []. 1126

8.300 These views are supported by competitors’ internal documents, which shows
that Activision content is expected to be particularly important relative to the
alternatives in the counterfactual:

(a) One internal document from a competitor [] dated October 2019 shows
that this competitor has targeted deals with publishers/franchises such as:
EA, Activision, Roblox and Fortnite. The same document described
Activision as having a ‘critical IP portfolio’. 1127 Another document from this
competitor dated December 2019 described Activision as a leading game
publisher and this competitor’s number one target for content acquisition
at that time. 1128

(b) Internal documents from another competitor []. 1129

(c) Two other documents from this competitor []. 1130

(d) Internal documents from a competitor []. 1131 []. 1132

8.301 We also received reports from a competitor [] on what it described as the
‘great success’ of Activision games on this competitor’s cloud gaming service

1125 [], submission to the CMA.

1126 [] response to the CMA’s RFI.
1127 []Internal Document.
1128 [] Internal Document.
1129 [] Internal Document; [] Internal Document; and [] Internal Document.
1130 [] Internal Document; and [] Internal Document.
1131 [].
1132 [] response to CMA’s RFI; and CMA analysis, [].

269
 during the time in which the service was in a testing phase. 1133 Activision
complemented this evidence by also submitting documents on how many
unique users of this competitor’s [] service were playing Activision’s games
during the testing phase. 1134 This evidence suggests that:

(a) On the one hand, []. 1135

(b) On the other hand, []. 1136

8.302 In relation to the evidence on [], Microsoft submitted []. 1137

8.303 Microsoft also submitted that the evidence on the [] during the []
demonstrates that CoD cannot be seen as an important game for cloud
gaming services. It submitted that if CoD was not popular amongst [] users
[], the CMA has no basis to assume that it would be in the future if it were
hypothetically made available for streaming. 1138 In particular, Microsoft
submitted that:

(a) the fact that [] did not suggest it was important. It submitted that, given
[], gamers did not []. The Parties also submitted that []. 1139

(b) given CoD was available on [], that there was no rational basis to
assume gamers would have been unwilling to ‘invest’ in it. This also
ignores []. 1140

(c) []. 1141

(d) the testing phase for [] likely contained fewer games than the full []
offering. As such it submitted that the evidence from the testing phase
therefore likely significantly overestimates the importance CoD would
have in the counterfactual if it was made available on [] again. 1142

8.304 We note that there are several limitations with the data and analysis above,
including that:

1133 This competitor explained that []. See [] response to CMA’s RFI; and CMA analysis [].
1134 Activision Internal Document; and Activision Internal Document.
1135 [] response to CMA’s RFI; and CMA analysis, [].
1136 [] told us that []. [] response to the CMA’s RFI.
1137 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.69 (a).
1138 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.66.
1139 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.67 (a); and Activision response to

the Provisional Findings, 2 March 2023, paragraph 58. See also Activision response to the Provisional Findings,
2 March 2023, paragraph 60 [].
1140 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.67 (b).
1141 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.67 (d).
1142 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.68.

270
 (a) Users from the testing phase may not be representative of current []
users. For example, access during the testing phase was free, so we
would expect to see a disproportionate number of gamers who want to
access a free service and play F2P games. 1143

(b) Some gamers might have decided not to invest in an expensive game like
CoD in order to use it on [], given there was no certainty the service
would have continued after the testing phase. In relation to Microsoft’s
submission that customers would have been willing to invest given it was
on the [] testing phase [], we note that customers would not have
known how long it would be available. The actual CoD gametime on []
that took place during the testing phase, therefore, would not adequately
capture potential demand from customers who would be more inclined to
stream CoD if they had more certainty that it would remain available the
service.

(c) The fact that [] is a BYOG service and users can download games to a
PC does not make gametime data during the [] testing phase more
reliable. Users without a gaming PC may have been reluctant to take the
risk of buying CoD to play on [] testing phase service (as set out
above). And users with a gaming PC would have been more likely to play
CoD on their PC in any event. As such, the behaviour of both cohorts on
[] testing phase service may underestimate potential gametime that
CoD would have on cloud gaming services as the market develops and
cloud gaming services become reliable alternatives to gaming PCs.

(d) []. This means that actual usage during the testing phase would likely
underestimate the potential importance of Activision content.

(e) Given the data is at an individual game level, it is possible the CoD
franchise overall would have been consistently in the top 10, even if
individual games were not. [] does not hold this data, so we are unable
to test this. However, the PC telemetry data discussed below confirms
that a number of CoD games account for a material proportion of total
CoD game time. For instance in July 2022, according to the PC telemetry
data, [] 1144 []. 1145 This would likely cause data on individual CoD
titles reaching the top 10 to underestimate the importance of CoD as a
franchise.

1143 [] response to the CMA’s RFI.

1144 The PC telemetry data [].
1145 CMA analysis of PC telemetry data. See paragraph 8.317 onwards and Appendix C.

271
8.305 We recognise that it is difficult to measure the importance of Activision’s
content based on the available data. We also recognise that the range of
games on cloud gaming services such as [] will continue to expand, which
will reduce the relative importance of individual games. We have, therefore,
taken account of the available evidence in the round. For example, as the
Parties submit, we recognise that measuring the popularity of a game by
unique users is less accurate than by MAUs. This is because MAUs give a
better indication of ongoing engagement with a game, whereas unique users
can include gamers who try a game but are then disengaged from it.
Nonetheless, in the absence of MAUs, we place some weight on the data on
most requested games on [], together with available information on unique
users (although we place less weight on the latter, given we are unable to
verify the exact methodology, for the reason discussed above). For the
reasons set out above, we also place limited weight on the gametime on []
when it was on its testing phase (though we note that even that data shows
that []).

Fortnite

8.306 Microsoft submitted that its experience with offering Fortnite on xCloud
[]. 1146 It submitted that []. 1147

8.307 Microsoft submitted that the Fortnite evidence is highly relevant to assessing
Microsoft’s ability to foreclose using CoD. 1148 In particular Microsoft submitted:

(a) Fortnite is not relevant only to mobile. It is available through Xbox Cloud
Gaming on console, PC and mobile. 1149

(b) The evidence presented in the Provisional Findings show Fortnite is one
the most popular games on [], and that on that service the []. 1150

(c) [] Fortnite was not []. 1151

(d) []. 1152

8.308 We discuss Microsoft’s experience with Fortnite in detail above. We recognise

that Fortnite was available on xCloud to play on console, PC and mobile, and
that any [] may be []. However, we note that Fortnite is []. On GFN

1146 Microsoft response to working papers.

1147 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.71.
1148 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.75.
1149 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.74 (a).
1150 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.74 (b).
1151 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.74 (c).
1152 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.74 (d).

272
 there are two versions of Fortnite, a mobile focused cloud version and a non-
mobile focused cloud version, []. Therefore, we consider this is clear
evidence that Fortnite is important (and successful) for cloud gaming services.
To the extent it [] for Microsoft, the evidence suggests at least a significant
factor is related to those playing on mobile and in particular to [] (we note
that Microsoft itself expressed Fortnite being [].) 1153 We also note that
despite the above there were still [] average MAUs of F2P Fortnite globally,
and [] in the UK, on xCloud between May and December 2022, 1154 [].
Whilst this is a [] of overall Fortnite MAUs across mobile, PC and console,
this reflects cloud gaming’s current small share of gaming generally. In our
view the data still demonstrates that Fortnite has [].

Evidence from console and PC gaming

8.309 Microsoft submitted that the CMA cannot rely on its findings in relation to
console gaming, which relate to CoD’s importance to SIE specifically in
attracting and retaining console customers, in determining whether CoD is an
important input for cloud gaming. 1155

8.310 Microsoft submitted this is wholly inconsistent with the CMA’s market
definition analysis which states that cloud gaming is attractive to a new pool of
customers and consumers consider different factors important for cloud
gaming as compared to consoles. 1156 It also noted that the CMA, in
Provisional Findings, cited data [], indicating it expects more cloud gaming
to take place on PC. 1157

8.311 Microsoft further submitted that, absent evidence specific to cloud gaming, the
most pertinent issue is CoD’s importance as a PC game. 1158

8.312 We agree with Microsoft that the most relevant evidence on the importance of
CoD would come from cloud gaming services, if available. However, given
CoD and other Activision games are not on cloud gaming services currently,
there is limited and imperfect available evidence. Given that we are focusing
on cloud gaming services capable of running graphically complex games that
currently run on high-performance consoles and PCs, we consider customers’
current preferences across console and PC gaming to be a good starting

1153 See paragraph above.

1154 Microsoft response to the CMA's RFI.
1155 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.51.
1156 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.52.
1157 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.53.
1158 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.54.

273
 point to understand preferences in the developing cloud gaming services
market.

8.313 We disagree with the Parties’ submission that we should not rely on console
evidence, and that the most pertinent evidence is CoD’s importance as a PC
game:

(a) Cloud gaming services aim to be device agnostic. Their purpose is to

achieve parity in terms of gameplay with consoles as well as gaming PCs,
without being tied to a specific piece of hardware. We recognise that there
is a wider range of PC games relative to console games. Nonetheless, in
assessing the relative importance of different games to cloud gaming
services as the market develops, it is important to take account of the
current preferences of all potential customers, which currently include
both console and PC gamers.

(b) We do not consider it relevant that []. We have repeatedly been told
that cloud gaming is device agnostic, 1159 while [] stated that the vast
majority of its users do not have a high-end PC and are new to
gaming. 1160

8.314 We consider that the most important games for cloud gaming services would
be the most popular games across consoles and PCs. In assessing the
importance of Activision’s content for cloud gaming services, therefore, we
consider its current importance to both consoles and gaming PCs. With
respect to consoles, we have already assessed as part of our previous theory
of harm the importance of CoD. We found that CoD consistently ranks as one
of the most popular console games. We found that CoD is so popular that a
significant number of PlayStation gamers would switch to Xbox if CoD were to
become exclusive to Xbox post-Merger, and that this would materially harm
PlayStation’s ability to compete. We consider, therefore, that Activision has
some of the most popular content for consoles.

8.315 With respect to PC, the number of available games is larger than on consoles.
Whilst the Parties’ internal documents suggest that CoD and WoW (which is
PC-only) are among the most played games on PC, CoD is less prominent on
PC than it is on console. These documents also suggest that there is a larger
number of games that reach a similar or greater level of engagement to CoD
on PC relative to console. They also suggest that shooters and Multiplayer

1159 See for instance paragraph 2.27, 8.52, 8.62 and 8.425.
1160 [] call note.

274
 Online Battle Arena (MOBA) games are particularly popular genres on PC. In
particular:

(a) A Microsoft internal document from 2021 looks at the top PC games
played by various categories of PC gamers []. The document shows
that []. The document also shows that [].

(b) This document also shows that []. 1161

(c) The same document also shows that []. 1162

(d) One Activision internal document dated February 2022 shows that both
CoD titles and WoW []. In particular, [], respectively whilst [] in
2021. WoW [] 2020 and [] in 2021. The other games which featured
in the [] list in 2020 were (in ranking order): []. In the 2021 list, the
other games were (in ranking order): []. 1163

(e) One Microsoft document from 2020 shows that CoD: Modern Warfare
was [] by PC gaming-hours in 2020 worldwide, preceded by []. When
looking at the [], CoD: Modern Warfare ranked [] by PC gaming-
hours in 2020 worldwide with a []% share of gaming-hours, whilst WoW
ranked [] with a []% share. Other Activision games in the list were
Overwatch ([] position, []% share), and WoW Classic ([] position,
[]% share). The remaining games in the list were (in ranking order):
[]. 1164 The document also shows that there are [] in the composition
of the top PC games list. For example, []. 1165

(f) One Activision document shows that on PC, Activision Blizzard games
achieved a total of []m average MAUs in 2020, corresponding to [] of
the total MAUs on Steam in the same year. 1166

8.316 In relation to the document discussed in (a)-(c) above, Microsoft submitted

that this document provides many examples of popular titles across multiple
genres, and it shows the broad range of content which is of interest to PC
gamers, which is much broader than the shooter genre. 1167

1161 Microsoft Internal Document; [].

1162 Microsoft Internal Document.
1163 Activision Internal Document.
1164 Microsoft Internal Document.
1165 Microsoft Internal Document.
1166 Activision Internal Document; in 2020, Activision games were only available on PC through Activision’s own

PC storefront Battle.Net.
1167 Microsoft response to working papers.

275
 • Microsoft PC telemetry data

8.317 Microsoft submitted that Windows telemetry data for June 2021 through to
December 2022 shows that []. Microsoft also submitted that the (global)
game-time share of the CoD franchise is low in comparison to the four highest
ranked games or franchises during 2021 and 2022. 1168 Microsoft submitted
that, in the UK, []. 1169

8.318 We assessed Microsoft’s telemetry data. 1170 We first looked at the share of
gametime on PC of Activision’s full range of games in terms of hours played
(based on the 12 months to Feb 2023). We then looked at the ranking of
Activision’s most popular games in terms of hours played (based on the same
data). In both of these assessments, we modelled different scenarios (the full
methodology is described in Appendix C):

(a) Sensitivity A. This includes all games in Microsoft’s telemetry data.

(b) Sensitivity B. This excludes very simple games (eg, []) from
Microsoft’s telemetry data. We consider that these games are less likely
to drive demand for cloud gaming services, given that cloud gaming
services are capable of serving as console and gaming PC replacements
in order to run graphically complex games such as CoD. We expect that
consumers are less likely to pay the fees of a cloud gaming service to
play simple games that can be played on almost any device (eg, old PCs,
work laptops, etc) 1171

(c) Sensitivity C. This excludes the same games excluded in the Sensitivity
B analysis, and it also excludes []. Microsoft’s PC telemetry data shows
that [] is [] the most popular game in terms of game hours []. 1172
However, it is a [] requirements, and it features [] in terms of most
requested games we have seen for cloud gaming providers (eg for []),
so we consider its relative importance for cloud gaming is likely overstated
by the PC telemetry data.

8.319 This is set out in tables 8.5, 8.6, 8.7, and 8.8.

1168 []. Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.56 (a).
1169 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.56 (b).
1170 Which is based on about []% of PCs running Windows 10 or 11, given that customers need to opt-in to

allowing Microsoft to collect telemetry data. This is explained further in Appendix C.

1171 Gamers may still play these games on cloud because they may move all their gaming to cloud, but these are

less likely to drive their choice.

1172 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.57 (b).

276
Table 8.5: Activision games’ shares of PC gaming in terms of hours played, March 2022—
February 2023, UK
Activision Franchise Sensitivity A Sensitivity B Sensitivity C

Hours (m) Percentage Hours (m) Percentage Hours (m) Percentage

Call of Duty [] [] [] [] [] []

World of Warcraft [] [] [] [] [] []
Overwatch [] [] [] [] [] []
Starcraft [] [] [] [] [] []
Diablo [] [] [] [] [] []
Hearthstone [] [] [] [] [] []
Sekiro [] [] [] [] [] []
Spyro [] [] [] [] [] []
Candy Crush [] [] [] [] [] []
Heroes of the Storm [] [] [] [] [] []
Total [] [] [] [] [] []

Source: CMA analysis of Windows telemetry data

277
Table 8.6: Activision games’ rankings in PC gaming in terms of hours played, 2022, UK, Sensitivity A
2022 monthly rank 2023 monthly rank Averages
ABK Franchise
March April May June July August September October November December January February Mean Median

Call of Duty [] [] [] [] [] [] [] [] [] [] [] [] [] []
World of Warcraft [] [] [] [] [] [] [] [] [] [] [] [] [] []

Source: CMA analysis of Windows telemetry data

Table 8.7: Activision games’ rankings in PC gaming in terms of hours played, 2022, UK, Sensitivity B
2022 monthly rank 2023 monthly rank Averages
ABK Franchise
March April May June July August September October November December January February Mean Median

Call of Duty [] [] [] [] [] [] [] [] [] [] [] [] [] []
World of Warcraft [] [] [] [] [] [] [] [] [] [] [] [] [] []

Source: CMA analysis of Windows telemetry data

Table 8.8: Activision games’ rankings in PC gaming in terms of hours played, 2022, UK, Sensitivity C
2022 monthly rank 2023 monthly rank Averages
ABK Franchise
March April May June July August September October November December January February Mean Median

Call of Duty [] [] [] [] [] [] [] [] [] [] [] [] [] []
World of Warcraft [] [] [] [] [] [] [] [] [] [] [] [] [] []

Source: CMA analysis of Windows telemetry data

278
8.320 Our assessment of Microsoft’s telemetry data shows that Activision’s games
account for between []% and []% of PC game time in the UK. The data
also suggests that both CoD and WoW are top [] games in terms of
ranking. CoD’s popularity varies across the year (most likely explained by the
date of new game releases), and, under sensitivity C, both CoD and
Overwatch were []. 1173

8.321 In relation to CoD, we also consider that these gametime figures

underestimate the popularity of the franchise. For two thirds of the year, the
latest CoD game was Vanguard, which turned out to be one of the least
popular games in the CoD franchise. This suggests that 2022 was not a
representative year of CoD’s popularity on PC gaming. This is consistent with
the significant difference in sales in the first few months of Vanguard
compared with Modern Warfare II. 1174 1175

8.322 Finally, we note that this PC telemetry data considers game time on all PCs.
Given that CoD is a relatively demanding game to run from a technical
perspective, only a subset of those PCs in the telemetry data will be capable
of running CoD well. 1176 This contrasts with cloud gaming services, which are
capable of serving as console and gaming PC replacements in order to run
graphically complex games such as CoD. While Sensitivity B and C remove
the most basic games, we consider that this telemetry data analysis still
underestimates the popularity of the CoD franchise under all sensitivities
given the impact on CoD will be larger relative to the less complex games
remaining in the analysis. We note that this issue does not arise in relation to
analysis of console telemetry data.

8.323 We also analysed global PC telemetry data (excluding China). 1177 We found
that under Sensitivity A, Activision games were slightly less popular than in
the UK ([]% vs []% respectively). We understand this is primarily because
CoD is less popular in Asia than in Europe/North America. We think it is most
relevant to consider UK data, given we have found the geographic market to
be the UK, and therefore place more weight on the UK figure.

1173 While we have not calculated the ranking for Overwatch throughout the period, we note that generally
Overwatch has had []. However, following the release of Overwatch 2 on 4 October 2022, Overwatch ranked
[]. We note that [], it still ranked [] under Sensitivity A, [] under Sensitivity B, and [] under Sensitivity
C in February 2023. See Appendix C.
1174 See for instance Call of Duty: Modern Warfare 2 UK launch sales are up 92% over Vanguard and Call of

Duty: Modern Warfare 2 is close to out-selling Vanguard after just 2 weeks, both accessed 11 April 2023.
1175 We expect the CoD ranking from March to September 2023 to be higher than in March to September 2022.

For instance, CoD ranks [].

1176 NVIDA’s pitch to Activision noted that []. Activision internal document.
1177 Given neither Microsoft nor its rivals operate cloud gaming services in China.

279
8.324 We also note that Microsoft submitted alternative PC game publishing shares
from Newzoo (see paragraph 8.290), which compared Activision with other
publishers on both gametime and MAUs. This suggests that at the publisher
level Activision has [] with []% share compared with the []% share.
Similarly, Activision had the [] of MAUs ([]%, with [] next at []%).

8.325 Microsoft made two submissions in relation to the PC digital storefront,

Steam:

(a) Steam, the largest digital storefront, did not carry any new releases of
CoD for three years until November 2022 following Activision’s
commercial decision to only sell its PC games on Battle.net, yet it still
maintained its leading position in PC game distribution and increased its
revenues. 1178

(b) Even following CoD returning to Steam, it did not appear in the top 10 list
of games played on the platform, while the peak number of players for
CoD: Modern Warfare II (including Warzone II) is half the figure for
Counter-Strike: Global Offensive, also a shooter game. 1179

8.326 We do not consider the evidence of CoD’s popularity on Steam to be strong

evidence of its importance to cloud gaming services. First, Steam is not a
cloud gaming service, it is a PC gaming store. As we noted above, we expect
demand for cloud gaming services to resemble not just current demand for
PC games, but some combination of demand for PC and console games.
Second, although significant, Steam represents a subset of all PC game
purchases and does not capture purchases of available games via other PC
stores, it is therefore at best indicative of a subset of gamers purchases.
Third, the data of the most played games on Steam does not represent an
improvement on Microsoft’s own PC telemetry data, which captures a greater
proportion of PCs and is able to produce a more representative sample of
users’ gametime. This can result in material differences in the data output: for
example, Microsoft submitted that Counter-Strike: Global Offensive had
double the peak players of CoD: Modern Warfare II. But Microsoft’s own PC
telemetry data shows that []. We therefore place limited weight on the table
of the 10 most played games on Steam by peak number of daily active users.

8.327 Microsoft also submitted that industry rankings show that CoD is consistently
outranked by other publishers’ PC games, given IGN’s ranking of the ‘25 best
PC games to play right now’ does not include CoD, Metacritic’s ranking does
not list CoD at the top of its rankings and PC Gamer’s ‘top 100 PC games’ list

1178 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.56 (c).
1179 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.56 (d).

280
 does not include CoD. 1180 We consider that these rankings are not an
adequate measure of CoD’s popularity, and in particular, its importance, given
there can be stark differences between critic’s reviews and actual game-time
and sales. 1181 We consider it better to measure importance more directly
through MAUs, share of gametime, and share of revenue across console and
PC.

8.328 We also reviewed evidence from the latest Newzoo report (Newzoo is a third
party provider of video games and gamer data) published on 21 December
2022. The report shows that CoD had a strong performance in 2022 across
PC and console (Xbox and PlayStation). In particular, the report shows that
two CoD games (Modern Warfare 2/ Warzone 2 and Modern Warfare/
Warzone) were the fourth and fifth largest in terms of MAUs across PC and
console (Xbox and PlayStation) in 2022, as based on Newzoo estimates. 1182

Microsoft’s agreements with NVIDIA, Boosteroid and Ubitus

8.329 Following the publication of our Provisional Findings, Microsoft entered into
cloud gaming agreements with NVIDIA, Boosteroid and Ubitus. In this section
we assess what impact these agreements have, if any, on the Merged Entity’s
ability to foreclose its cloud gaming service rivals.

Parties’ views

8.330 Microsoft made several submissions regarding the agreement it has entered
into with NVIDIA in response to the Provisional Findings and in response to
the Remedies Working Paper. 1183 Microsoft explained that it has entered a
legally binding agreement with NVIDIA for all of Activision’s PC games to be
made available on GFN for at least ten years post-Merger. 1184 Microsoft also
made submissions regarding the agreements it has entered into with
Boosteroid and Ubitus in response to the Remedies Working Paper. Microsoft
explained that each of these agreements contains [], to be made available

1180 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.56
1181 For instance, the number one game on PC Gamers ‘top 100 PC Games’ is Disco Elysium, which sold
2.6million units on Steam and made $7.4million in the first six months. PC Gamers top 100 and Disco Elysium
statistics, both accessed by the CMA on 12 April 2023. In comparison CoD: Modern Warfare II, which is not in the
top 100 list, made over $1billion in 10 days.
1182 The Games Market in 2022: The Year in Numbers | Newzoo, accessed by the CMA on 16 January 2023;

China and India are excluded from these estimates.

1183 See also Microsoft response to the Provisional Findings Addendum, 3 April 2023, paragraph 4.3. In addition

to those submissions discussed here and in the incentive section below, Microsoft also made submissions that
these agreements address the cloud SLC. These submissions are discussed further in Chapter 9 (countervailing
factors) and Chapter 11 (remedies). This is because the assessment in this Chapter deals with the question of
whether an SLC may be expected to arise in the cloud gaming market in the UK in the first place.
1184 Microsoft response to Provisional Findings, 2 March 2023, paragraph 3.3.

281
 for streaming on NVIDIA’s, Boosteroid’s and Ubitus’ cloud gaming platforms
for 10 years post-Merger. 1185

8.331 Microsoft submitted that, given the terms of these three agreements, which
provide for access to the content that the CMA considers to be ‘important’,
there is no basis to find it could have the ability to foreclose NVIDIA,
Boosteroid or Ubitus. Microsoft has submitted that the theory that it would be
able to foreclose cloud gaming rivals is undermined by the fact that it has
made legally binding commitments not to do so, and that it would have to
breach these agreements in order to foreclose these rivals and would face
[] claims if it did so. 1186

Our assessment

8.332 As explained in Chapter 7, the MAGs make it clear that the CMA’s
assessment of the ability of the merged entity to foreclose is unlikely to place
material weight on contractual protections. 1187 The considerations in our
guidance apply to the present case. In particular, contracts may be
renegotiated or terminated early, or may not be enforced depending on the
respective parties’ bargaining positions (and Microsoft’s overall strengths in
cloud gaming discussed already in this chapter is relevant in this regard, as
we would expect it to hold considerable leverage in relation to any subsequent
negotiation or contractual dispute). Nor can we be sure that NVIDIA,
Boosteroid or Ubitus (or any other third party) would be able to enforce the
terms of any relevant contracts should any of them need to do so.

8.333 In relation to the specific terms of these agreements, we note that Microsoft
has specifically acknowledged a degree of uncertainty in respect of the future
development of each contract by including an [] clause, which broadly sets
out that []. 1188

8.334 As well as demonstrating the difficulty of contracts adequately allowing for

different circumstances that could arise in a dynamic and growing market, we
consider this demonstrates the difficulties in placing any material weight on
contracts being able to constrain an ability to foreclose that otherwise exists,
where bargaining power in the event of such renegotiation or any related
disputes would not be equal. Further, in all three agreements, [].

1185 Microsoft response to the Remedies Working Paper. Microsoft also made similar submissions relating to all

three agreements in its response to the CMA’s Supplementary Evidence Paper.

1186 Microsoft response to Remedies Working Paper.
1187 CMA129, paragraph 7.15.
1188 See Microsoft, email to the CMA; and Microsoft response to the CMA’s Remedies Notice.

282
8.335 We note further uncertainty arising from the terms of these agreements
through [] provisions. For example, the Boosteroid and Ubitus agreements
each contain a provision allowing Microsoft to []. 1189 It is not clear the extent
to which such [] may arise, although the fact that this provision has been
included in these agreements indicates that Microsoft sees this as a material
risk.

8.336 In any case, we note that Microsoft has entered into contracts with three cloud
gaming providers, who either have a BYOG model (NVIDIA, Boosteroid) or a
B2B focus (Ubitus). We are assessing here whether the Merged Entity would
have the ability to use Activision’s games to foreclose cloud gaming service
rivals in general, not limited to any specific rivals. In the context of a nascent
and growing market, we cannot be confident that agreements with a limited
number of providers remove the Merged Entity’s ability to foreclose in the
cloud gaming services market more generally. This is the case even in
circumstances where cloud gaming providers with agreements in place
consider that any concerns they have are now resolved – as their incentives
are to ensure access to content for their own business, rather than ensuring
competition across the market more generally. 1190

8.337 Accordingly, while we recognise that these agreements may provide NVIDIA,
Boosteroid and Ubitus with some level of a protection against foreclosure to
some extent, given the uncertainty flowing from the terms of these
agreements which relate only to three cloud gaming providers, we do not
consider it appropriate to place any material weight on these agreements in
the overall assessment of the Merged Entity’s ability to foreclose its cloud
gaming service rivals.

Conclusion on Ability

8.338 Based on the available evidence, we consider that Activision’s games—

particularly CoD, WoW, and to a lesser extent Overwatch 1191—would be an
important input to cloud gaming services absent the Merger.

8.339 We recognise that the available evidence on the importance of Activision’s

games arising directly from existing cloud gaming services is limited. We

1189 Clause 4.4 of the Cloud Gaming License Agreement, [] March 2023; Clause 4.4 of the Cloud Gaming
License Agreement, 11 March 2023.
1190 We note that Microsoft pointed us to NVIDIA’s statement to the CMA that the agreement []. Further, each

of Microsoft’s agreements with NVIDIA, Boosteroid and Ubitus include [].

1191 We note that Activision is due to release a new game in the Diablo franchise (Diablo IV) on 6 June 2023

(Your guide to the Diablo IV open beta, accessed by the CMA on 10 April 2023). Given Diablo III is over 10 years
old, we expect this new release to increase interest in the Diablo franchise significantly. However, we have not
conducted any formal assessment of the future importance of Diablo IV specifically.

283
 consider that limited data is to be expected in a dynamic market that is in the
early stages of development.

8.340 We consider, however, that the popularity of Activision’s games on consoles

and gaming PCs is indicative of its importance for cloud gaming services now
and in the near future. On consoles, we found that CoD is an important game
and that alternatives with similar levels of gametime, revenue, and
engagement are limited. On PC, the evidence indicates that CoD and WoW
are among the most played games in the UK, and Overwatch is also
reasonably popular. We found that there are more games on PC than on
console, and CoD is relatively less popular on PC than on console (although
still comfortably a top 10 game). Overall, however, we found that Activision’s
content is only slightly less important on PC than on console.

8.341 This evidence indicates that Activision content would likely become an
important input to cloud gaming services absent the Merger. It also indicates
that any alternatives would not be a sufficiently good replacement for
Activision content, especially CoD and WoW, and that they would not be
enough to offset any loss of Activision content by cloud gaming rivals, nor to
compensate for the reduction in range and choice for customers of those
rivals.

8.342 This is particularly relevant in a nascent market like cloud gaming, wherein
economies of scale and network effects may play a significant role in
determining the success of a service, as discussed above. In turn, any harm
to cloud gaming rivals’ competitive strength is likely to be self-reinforcing and
may hinder these rivals from scaling quickly enough, which is especially
important to compete successfully when network effects play a role.

8.343 In the previous sections we also considered that multi-homing might be

stronger in cloud gaming than in console in the counterfactual, given the lower
switching costs. Multi-homing might in principle reduce the ability to foreclose
in cloud gaming if customers already use multiple platforms, eg to access
exclusive content. However, the evidence on multi-homing specific to cloud
gaming was mixed, mainly because the market is still developing.
Furthermore, any foreclosure might prevent the emergence of multi-homing in
the counterfactual, especially if such foreclosure prevented rival cloud gaming
services from reaching sufficient critical mass that would allow them to
succeed.

8.344 Therefore, as cloud gaming rivals are still negotiating to attract the larger
publishers to their platforms in order to be successful, we consider that this is
likely to be a crucial time. In such a moment for the cloud gaming market, we
consider that not having access to Activision, and particularly CoD titles as a

284
 result of foreclosure would have even more potential to significantly reduce
the chance of these cloud gaming services to succeed.

8.345 For the reasons set out above, we do not consider that the NVIDIA,
Boosteroid or Ubitus agreements have any material impact on the overall
assessment of the Merged Entity’s ability to foreclose cloud gaming service
rivals.

8.346 We therefore consider that, absent the Merger:

(a) Activision content, in particular CoD WoW and Overwatch, would be a

particularly important input to cloud gaming services.

(b) Any alternatives are likely not to be sufficient to offset the loss incurred to
cloud gaming rivals by foreclosure of Activision content, nor to
compensate for the reduction in range and choice for customers of those
rivals.

(c) Multi-homing is unlikely to limit ability to foreclose in cloud gaming. In fact

foreclosure could negatively impact the emergence of multi-homing itself
in the future.

8.347 Therefore, we conclude that the Merged Entity has the ability to foreclose
rivals in the cloud gaming market in the UK using Activision content.

Incentive to foreclose

8.348 In this section, we assess whether the Merged Entity would have the incentive
to use Activision’s content to foreclose downstream cloud gaming
competitors. These rivals will include NVIDIA, Amazon Luna, and potential
new entrants [].

8.349 The Merged Entity would have the incentive to engage in foreclosure
strategies if the benefits of doing so exceed the costs. 1192 In assessing what
constitutes a benefit or a cost, we consider not only the immediate financial
implications of a foreclosure strategy, but also the longer-term consequences
(the latter reflecting possible strategic costs and benefits that manifest over a
longer time horizon or affect different parts of the business).

8.350 In this case, given that cloud gaming is a developing market, we would expect
to place more weight on the long-run strategic objectives of the Merged Entity.
As set out in the MAGs, ‘in complex and dynamic markets, firms may not

1192 CMA129, paragraph 7.16.

285
 focus on short term margins but may pursue other objectives to maximise
their long-run profitability, such as ‘eliminating a possible long-term threat,
increasing the stickiness of existing customers, positioning themselves
strongly in high-growth markets, gaining customers to obtain direct or indirect
network effects, obtaining access to customer data or enabling cross-selling
within a broader ecosystem’. 1193

8.351 The evidence that we use to assess these strategic objectives is primarily
qualitative. As set out in the MAGs, ‘the CMA may undertake more extensive
quantitative analysis in simple markets with high quality data but focus on a
qualitative assessment in complex and dynamic markets, where firms’ current
positions and margins may not be a good guide to the future, and strategic
considerations may play a greater role.’ 1194 Given that cloud gaming is still in
its infancy, we do not believe that a quantitative analysis of short-run gains
and losses from foreclosure would be informative.

8.352 As such, we have focused our analysis on two types of evidence:

(i) Microsoft’s past behaviour regarding first-party content on rival cloud
gaming platforms, and (ii) Microsoft’s business strategy as set out in its
internal documents. As noted in the MAGs, the purpose of the incentives
analysis is to predict the merged entity’s behaviour, and it may be possible to
understand this directly from its past conduct and business strategy. 1195

8.353 We consider the evidence on Microsoft’s current and potential strengths in

cloud gaming to be relevant for the assessment of the incentives to foreclose.
As per the MAGs, the ‘gain in downstream sales’ resulting from foreclosure
‘will be greater if the merged entity has a more successful downstream
offering’. 1196 Therefore, other things equal, the more successful Microsoft is in
cloud gaming (now and in the future), the greater the incentives to foreclose
cloud gaming rivals.

Parties’ views

8.354 Microsoft submitted that it does []. It expects downloads to continue to

dominate the market, and it does not expect cloud gaming to replace native
mobile games purchased through mobile app stores. 1197

1193 CMA129, paragraph 7.19(e).

1194 CMA129, paragraph 7.18.
1195 CMA129, paragraph 7.19 (a).
1196 CMA129, paragraph 7.19 (b).
1197 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.49.

286
8.355 Microsoft submitted that the []. 1198 In particular, Microsoft submitted that it
has []. 1199 More generally Microsoft submitted that the CMA cannot
conclude on the balance of probabilities that cloud gaming services will
become profitable in the next five years. 1200

8.356 Microsoft also submitted that the possibility that cloud gaming might become
profitable in the long term is insufficient to find, on a balance of probability
standard, that it could have an incentive to foreclose in the next five years. It
also submitted that the CMA must first be confident that any such foreclosure
strategy would be profitable. 1201

8.357 Further Microsoft submitted that demand for cloud gaming has [], and that
it’s not in Microsoft’s interest to harm the consumer experience of a rival cloud
gaming provider. 1202

Our assessment

8.358 As a preliminary point, we note that console providers, including Microsoft,

place significant value in having exclusive content to differentiate their
platform and attract more users. Most first-party Xbox and PlayStation games
are exclusive to their respective platform, and almost every studio that
Microsoft has bought now makes games exclusive to Xbox. We have seen no
evidence to suggest that the importance that Microsoft places on having
exclusive content on its console would not extend to having exclusive content
on its cloud gaming service. Indeed, given Microsoft’s much stronger position
in cloud gaming services, we would expect its incentive to make its content
exclusive to be even stronger than on console.

8.359 As discussed above, we believe that no cloud gaming rival can match
Microsoft’s cost advantage arising from its ownership of Windows, Azure, and
the Xbox gaming catalogue combined. As such, we believe that Microsoft is
already in a uniquely strong position in the market for cloud gaming services.

8.360 As discussed above, while we do not put a lot of weight on market shares, we
do find them indicative of the current strength of different cloud gaming
providers, and our global market shares estimates presented above support
that Microsoft already holds a strong position in cloud gaming. These
estimates show that Microsoft’s share increased from [30-40]% in 2021 to [60-

1198 Microsoft response to working papers.

1199 Microsoft response to working papers.
1200 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.86 (a).
1201 Microsoft response to working papers.
1202 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 5.49.

287
 70]% in the UK in 2022, and that, in 2022, it had [] as many average MAUs
as the next biggest service, NVIDIA GFN.

8.361 In the reminder of this section, we consider Microsoft’s past behaviour in

making games available to rival cloud gaming platforms, ie whether it has (i)
placed recent first-party games on rival cloud platforms and (ii) taken games
off rival cloud platforms when it has acquired a publisher (in particular,
Zenimax). We also consider Microsoft’s internal documents setting out its
business strategy in cloud gaming.

Past behaviour

8.362 Microsoft submitted that it has no policy against including first-party games in
rival multi-game subscription or cloud gaming services. 1203 It submitted that its
titles were available on PS+ and Stadia. 1204 In particular, it submitted that
Rage 2, Wolfenstein: Youngblood, Doom 64, Doom and Doom Eternal were
available on Google Stadia, and that Elder Scrolls V: Skyrim Special Edition is
available on PlayStation Plus Extra and Premium. 1205

8.363 Microsoft also submitted a list of a further 23 games that it stated were
available on PS+. 1206 We understand, however, that the only games on this
list available on SIE’s cloud gaming service are published by Bethesda. 1207

8.364 To assess the extent to which Microsoft makes its AAA games (ie those more
comparable to CoD) available on rival cloud gaming platforms, we have
looked at a list of Microsoft’s biggest games and franchises, 1208 excluding
Bethesda games (which we discuss further below). 1209 Of seven games and
franchises on this list, six were available on xCloud through Game Pass

1203 Microsoft response to the CMA’s RFI.

1204 Microsoft response to working papers.
1205 Microsoft response to working papers. Elder Scrolls V: Skyrim Special Edition is available on PlayStation

cloud through PlayStation Plus Premium membership.

1206 The games listed were: Brink, Deathloop, Dishonoured, Doom (2016), Fallout 3, Fallout 4, Fallout 76, Fallout

New Vegas, Hunted, Hunted The Demons Forge, Prey, Rage, Rage 2, Rogue Warrior, The Elder Scrolls IV
Oblivion, The Elder Scrolls Online, The Elder Scrolls V Skyrim, The Evil Within, The Evil Within 2, Wet,
Wolfenstein II: The New Colossus, Wolfenstein: The New Order, and Wolfenstein: The Old Blood. See Microsoft
response to the CMA’s RFI.
1207 In addition, some of these Bethesda games were available on SIE’s cloud gaming service in the past, but no

longer are (such as Dishonoured, Rage 2, the Evil Within, The Evil Within 2, Wet, Wolfenstein II: The New
Colossus, and Wolfenstein: The Old Blood.)
1208 We combined a list of games from the Xbox Game Studios website, accessed by the CMA on 17 January

2023, (which are highlighted as ‘some of the biggest franchises in history’) and Microsoft’s top games and
franchises by revenue in 2021. Parties Internal Document. Excluding the Bethesda games, these were: [].
1209 We look at Bethesda games separately because firstly, Microsoft has only taken ownership of Bethesda

since March 2021 and so there may have been contractual obligations in place between Bethesda and other
cloud gaming platforms prior to the acquisition (Officially Welcoming Bethesda to Team Xbox - Xbox Wire,
accessed by the CMA on 31/01/2023). For instance, [] (Microsoft response to RFI). Secondly, unlike Activision
games, Bethesda games were available on Google Stadia, PlayStation cloud gaming and NVIDIA GFN prior to
the acquisition.

288
 Ultimate (only [] was not available on xCloud). 1210 When looking at rival
cloud gaming platforms, however, we have found that none of these games
were available.

8.365 We have also looked at Microsoft’s past behaviour in relation to Bethesda’s

games. We identified five major Bethesda game franchises (Doom, Elder
Scrolls, Fallout, Rage and Wolfenstein). As Microsoft submitted, 1211 the latest
titles in all franchises are or were available on a rival cloud gaming platform
(PS+ or Stadia). We note that all except two of these titles were released and
available on one or both of these cloud platforms prior to Microsoft’s
acquisition of Bethesda. 1212 The titles released after the acquisition (notably
Deathloop and Ghostwire) are not available on PS+ cloud as SIE’s cloud
offerings run on PS4 servers and these new titles were released on PS5 and
do not offer backwards compatibility.

8.366 The Parties submitted that if Microsoft were incentivised to foreclose rival
cloud gaming services, it would also have removed these Bethesda titles from
rival platforms. 1213

8.367 We note that many of the latest, but older, Bethesda games are on
PlayStation’s cloud gaming platform. Some Bethesda games were also
available on Google Stadia prior to its closure. However, none continued to
remain available on either GFN, Amazon Luna, or Boosteroid following
Microsoft’s acquisition of ZeniMax. As described elsewhere in this Chapter,
we consider these competitors at least as close competitors to xCloud as
PlayStation’s cloud gaming offering. We also note that Microsoft would be
aware that SIE cannot stream the latest games that only work on PS5.

8.368 More importantly, Bethesda games are only a small subset of Microsoft
games, and as described in the analysis above, none of Microsoft’s AAA
games are available on rival cloud gaming platforms. We also consider it
relevant that, unlike Bethesda games at the time of the ZeniMax acquisition,
none of Activision’s games are currently available on cloud gaming platforms.
This is relevant because as noted in Chapter 7, in our view, while Microsoft
strives to acquire and produce exclusive content, the financial and strategic
calculation of creating new and exclusive games for Xbox may be different
where Microsoft has to trade off certain quantifiable losses (from removing

1210 Age of Empires II was released on xCloud in Jan 2023, while Age of Empires IV will be released on xCloud
later in 2023, accessed by the CMA on 31/01/2023.
1211 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.80.
1212 Fallout 76 was added to PlayStation cloud in October 2021 (following its released in October 2018) and Elder

Scrolls V: Skyrim Special Edition was added to PlayStation cloud in November 2022 (following its release in
October 2016).
1213 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.81.

289
 from a rival) against uncertain gains (drawing more customers to Xbox). While
we found this was relevant in our incentive analysis in relation to CoD on
console, 1214 it does not apply to cloud gaming where there are no significant
quantifiable costs from removing a game from a rival.

Internal documents discussing Bethesda games on GFN

8.369 We have also considered internal documentary evidence in relation to the

previous acquisition of ZeniMax (which includes the publisher Bethesda).
Bethesda had two games (Wolfenstein: Young Blood and Quake 2 RTX) on
NVIDIA GFN at the time of Microsoft’s acquisition of Zenimax. Following the
acquisition, a Microsoft internal document shows []:

(a) In an internal email in March 2021 to [] notes that []. 1215

(b) In the same email chain, [] replies to [] suggesting they []. 1216

(c) Ultimately Microsoft decides to remove these games from GFN. In

another email, [] confirms that he has the ‘OK’ from [] to give []. In
this email thread, when discussing whether Microsoft should []. 1217 In
discussing [] with another colleague, [] replies to [] to say []. 1218

8.370 Microsoft submitted that it had a consistent policy not to allow a provider to
stream its games without a license. 1219 Microsoft also submitted that while the
Xbox Cloud Gaming team did not object to continuing to support these games
on NVIDIA GFN, Microsoft ultimately removed the titles in 2021 because no
valid licensing agreement was in place. 1220 Microsoft submitted that []. 1221

8.371 Microsoft submitted that []. 1222

8.372 Microsoft also submitted that one of these two titles, Wolfenstein: Young
Blood, continued to remain on Google Stadia, which is also a competitor to
Xbox Cloud Gaming, as a valid licence agreement was in place. 1223

1214 See Chapter 7 above.

1215 [] See Microsoft Internal Document.
1216 [] See Microsoft Internal Document.
1217 Microsoft Internal Document.
1218 Microsoft Internal Document.
1219 Microsoft response to the CMA’s RFI.
1220 Microsoft response to the CMA’s RFI; and Microsoft response to the Provisional Findings, 2 March 2023,

paragraph 3.84.
1221 [] Microsoft response to the CMA’s RFI.
1222 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.83; and Microsoft, Main Party

Hearing transcript.
1223 Microsoft response to the CMA’s RFI.

290
8.373 We consider that the evidence presented by Microsoft shows that it cares
about protecting IP. However, Microsoft’s submissions do not address the fact
that a Microsoft senior employee []. We have also seen no evidence of
[]. 1224

8.374 In any case, we consider that these motives ([] licensing) are not mutually
exclusive and [] are likely to have contributed to Microsoft’s decision to
remove the two Bethesda games. Our reading of Microsoft’s documents
suggest that competition with rivals is likely to play a role in not concluding
licensing agreements with them.

Microsoft’s agreements with NVIDIA, Boosteroid and Ubitus

Parties’ views

8.375 As noted above, Microsoft has made a variety of submissions regarding the
agreements it has entered into with NVIDIA, Boosteroid and Ubitus. 1225 A
number of these submissions relate to the impact of these agreements on the
Merged Entity’s incentives to foreclose post-Merger. Microsoft has submitted
that, as a result of these agreements, our assessment of incentives ‘has been
radically altered’. 1226 Microsoft submitted that if Microsoft were to foreclose
these three rivals, it would face [] claims from each of them. 1227

8.376 Microsoft submitted that the NVIDIA agreement proves conclusively that
Microsoft is incentivised to distribute Activision content widely. 1228 Microsoft
added that the agreements with NVIDIA, Boosteroid and Ubitus show the
strategy that Microsoft has maintained throughout that it wants to make its
games available widely. 1229 Microsoft submitted that the CMA’s analysis of its
incentives to foreclose relies primarily on Microsoft’s past conduct in making
its first-party content available on rival cloud gaming services. Microsoft
submitted that the NVIDIA Agreement provides more recent, and more
relevant, evidence which undermines any suggestion Microsoft would seek to
make Activision content exclusively available for streaming on Game Pass
Ultimate post-Merger. 1230

8.377 Microsoft submitted that the NVIDIA agreement comprehensively removes

any incentive on Microsoft’s part to withhold Activision content from NVIDIA.

1224 Prior to the agreement entered into during the course of this inquiry.
1225 In addition to the references cited in this section, see also Microsoft response to the Provisional Findings
Addendum, 3 April 2023, paragraph 4.3.
1226 Microsoft response to Remedies Working Paper.
1227 Microsoft response to Remedies Working Paper.
1228 Microsoft response to Provisional Findings, 2 March 2023, paragraph 1.2.
1229 Microsoft response to Remedies Working Paper.
1230 Microsoft response to Provisional Findings, 2 March 2023, paragraph 3.18.

291
 Microsoft submitted that it shows that Microsoft’s allegedly ‘uniquely strong
position in cloud gaming’ has not incentivised it to foreclose its leading rival,
NVIDIA, and that the opposite is in fact true. We assume these submissions
apply equally to the Boosteroid and Ubitus agreements. Further, Microsoft
submitted that it has agreed to make its first-party PC games available on
NVIDIA GFN [] and that this reflects Microsoft’s incentives to distribute its
content widely, including through alternative cloud gaming business models
(eg BYOG) to its own. 1231 We note Microsoft has agreed [].

8.378 Finally, Microsoft submitted that these agreements significantly reduce any
incentive on Microsoft’s part to withhold Activision content from other cloud
gaming providers. 1232 Microsoft submitted that, in particular, the CMA’s
assessment of incentives relies heavily on the notion that Microsoft’s strategy
is driven by making its first-party content exclusively available on Game Pass
Ultimate for cloud game streaming. Microsoft submitted that, having granted
streaming rights to Activision content to other cloud gaming providers this
alleged strategic driver is removed, ie Activision content will not be exclusively
available for streaming through Game Pass Ultimate regardless of any
foreclosure strategy. 1233 Microsoft further submitted that the potential gains
from any hypothetical foreclosure strategy would be significantly reduced (ie
as consumers would have at least one alternative choice). 1234

Our assessment

8.379 We consider the evidence discussed above demonstrates that, absent the
agreements, the Merged Entity would have an incentive to foreclose cloud
gaming rivals in the UK. In this section, we are therefore considering whether
the agreements with NVIDIA, Boosteroid and Ubitus impact this assessment.

8.380 As stated above in Chapter 7, the MAGs make clear that the CMA may
consider any financial or reputational costs of terminating contracts in its
assessment of foreclosure incentives. We consider it is possible in principle
that the financial or reputational impacts of breaching these agreements could
impact the Merged Entity’s incentive to foreclose. However, we consider this
depends on the likely consequences in the event the Merged Entity did adopt
a foreclosure strategy and therefore breached the agreements, and the

1231 Microsoft response to Provisional Findings, 2 March 2023, paragraph 3.21(b).

1232 Microsoft response to Provisional Findings, 2 March 2023, paragraph 3.21(c); and Microsoft response to
Remedies Working Paper.
1233 Microsoft response to Provisional Findings, 2 March 2023, paragraph 3.21(c); and Microsoft response to

Remedies Working Paper.

1234 Microsoft response to Provisional Findings, 2 March 2023, paragraph 3.21(c). Microsoft also made similar

submissions on incentive to foreclose in relation to all three agreements in its response to the CMA’s
Supplementary Evidence Paper.

292
 materiality of those consequences in the context of an overall foreclosure
strategy and the strength of the Merged Entity’s market power.

8.381 In terms of the likely consequences of breaching the agreements, we

acknowledge that breaching the agreements Microsoft has entered into with
NVIDIA, Boosteroid and Ubitus would risk actions potentially being brought
against Microsoft which could have financial and reputational impact.
However, we have already discussed above that certain clauses within the
agreements themselves introduce uncertainty, for example enabling [] in
the event of [], which we consider is particularly relevant in a nascent and
growing market like cloud gaming. In addition, for all three []. Further, as
already explained, contractual protections may more generally be
renegotiated or terminated early or may not be enforced depending on the
respective parties’ bargaining positions (and Microsoft’s overall strengths in
cloud gaming discussed in this chapter is relevant in this regard). Accordingly,
we consider there is considerable uncertainty as to what the consequences
would be for Microsoft in relation to these agreements if it was to nevertheless
engage in a strategy of total foreclosure and we have not received any
evidence to indicate this would result in consequences that would outweigh
the benefits we have identified above of pursuing a foreclosure strategy.

8.382 In respect of Microsoft’s submission that the agreements prove conclusively

that Microsoft is incentivised to distribute Activision content widely, we
consider it is relevant that these agreements were entered into in the context
of an ongoing merger review process. We consider that Microsoft entering
into these agreements does not provide us with reliable evidence regarding its
incentives in the same way as other past behaviour which is separate to the
Merger and Merger review process, or our general analysis of the Merged
Entity’s incentives above. Accordingly, we do not consider the fact that
Microsoft has entered into these agreements undermines our findings on its
post-Merger incentives. Microsoft may have short-term incentives to enter into
these agreements to seek to address the competition concerns arising from
the Merger, but this is not informative of its longer-term commercial
incentives.

8.383 Accordingly, we do not consider that these agreements materially impact the
Merged Entity’s incentive to foreclose cloud gaming rivals. In particular, we
consider there is there is uncertainty as to any consequences of breaching
these agreements. To the extent there were consequences, we have seen no
evidence to suggest this would outweigh the incentives that we consider
otherwise would exist to engage in a foreclosure strategy. Finally, we note
that these agreements only cover three cloud gaming providers, whereas we
are assessing whether the Merged Entity would have the incentive to use

293
 Activision’s games to foreclose cloud gaming service rivals in general, in what
is an otherwise nascent and growing market.

Conclusion on the incentive to foreclose cloud gaming rivals

8.384 As set out above, we believe that cloud gaming will continue to grow and be
profitable in the next five years. In a nascent market, particularly those
characterised by some element of direct and/or indirect network effects,
success is highly uncertain for new entrants, and there is a greater
opportunity (and stronger incentive) for incumbents to engage in foreclosure
strategies in a bid to acquire market power.

8.385 In this case, we believe Microsoft is already one of the strongest incumbents
in cloud gaming services as evidenced by current market shares, and its
multi-product ecosystem means that it is well positioned to compete in this
market as it continues to grow and develop. As set out above in our TOH 2
framework, Microsoft will have a stronger incentive to foreclose given its
uniquely strong position in cloud gaming.

8.386 These two factors suggest Microsoft would expect significant recapture,
especially in the long run – something we put considerable weight on. We
also consider, therefore, that even if Microsoft games were not entirely
exclusive to xCloud, Microsoft would still recapture many sales when
considering whether to foreclose further downstream competitors.

8.387 We also see that Microsoft’s internal documents reveal a strategy of not
making its first-party titles available on rival cloud gaming platforms. [].

8.388 With few exceptions (ie some Bethesda titles on some platforms), Microsoft’s
past behaviour shows that it has not made current AAA first-party content
available on any cloud gaming service other than Game Pass Ultimate.

8.389 We do not consider the agreements which Microsoft has entered into with
NVIDIA, Boosteroid and Ubitus materially alter our assessment.

8.390 As such, we find that Microsoft has the incentive to foreclose cloud gaming
rivals, including the likes of NVIDIA GFN, Amazon Luna, Boosteroid, and
potential new entrants []. 1235

1235 Therefore, we consider Microsoft has the incentive to foreclose all downstream rival cloud gaming services.

However, we do not conclude that the incentive to foreclose each rival only holds if all other rivals are also
foreclosed.

294
Effect of foreclosure

Introduction

8.391 This section considers the effect on competition in the market for cloud
gaming services of a foreclosure strategy by the Merged Entity using
Activision content. It considers whether any such strategy would result in
substantial harm to overall competition in the downstream market. 1236 To
assess this, we have considered the strength of Microsoft’s cloud gaming
service (including as a result of Microsoft’s broader ecosystem set out above)
and the strength of other cloud gaming service providers.

8.392 We have also considered barriers to entry and expansion. This includes the
possibility of new providers entering the market, and the impact that a
foreclosure strategy may have on their incentive to enter in the first place. Any
negative effect on competition is likely to be larger if there are barriers to entry
and expansion or if network effects are strong, as these may reinforce the
Merged Entity’s strength following a foreclosure strategy. 1237 Competition
concerns may be particularly likely to arise if one of the merger firms has a
degree of pre-existing market power in the downstream market, and already
faced limited competitive constraints pre-merger. 1238

Strength of different cloud gaming services

8.393 Understanding the strength of Microsoft and rival cloud gaming providers is
important for two related reasons. First, it shows how concentrated the market
is and how much competition Microsoft faces in cloud gaming. Second, it
shows the likely impact that a foreclosure strategy could have in the
competitive strength of rivals.

Parties’ views

8.394 Microsoft submitted that it does not hold a uniquely strong position in cloud
gaming services and noted that rivals have their own strengths and assets. As
described above, Microsoft submitted that it does not use Azure or the
Windows OS in its cloud gaming services today. It submitted that while
Microsoft and other providers such as SIE and Amazon have first-party
content, NVIDIA has become the leading cloud gaming provider without

1236 CMA129, paragraph 7.20.

1237 CMA129, paragraph 7.20.
1238 CMA129, paragraph 7.21.

295
 having any first-party content. It also submitted that Microsoft is behind both
Amazon and Google on cloud infrastructure. 1239

8.395 Microsoft submitted that its [] is evident from the fact that it has a []. 1240

8.396 As discussed in detail above, Microsoft also submitted that cloud gaming
reduces network effects as the cost of switching or multi-homing is less than
for console gaming.

8.397 Microsoft submitted that with the NVIDIA agreement in place no cloud gaming
SLC can arise from the merger, and that Merger will in fact enhance
competition in cloud gaming services. 1241

Our assessment

Internal documents

8.398 Internal documents describing Microsoft’s competitors in cloud gaming and

their respective strengths are discussed above in the section on Microsoft’s
potential strengths in cloud gaming. They show that Microsoft monitors and
benchmarks itself primarily against [], although it also considers the
potential competition from a wider group including [].

Shares of supply

8.399 As described above, the market for cloud gaming services is currently
concentrated with the top three competitors having over 90% share of the
market. 1242 It is, however, a developing market, and there has been recent
entry and expansion by rivals. For example, Amazon Luna has recently
entered the UK market and is only available in a small number of countries
currently. [].

Evidence considered in Microsoft’s potential strengths in cloud gaming section

8.400 As set out in the MAGs, ‘Competition concerns may be particularly likely to
arise if one of the merger firms has a degree of pre-existing market power in

1239 Microsoft response to working papers; and Microsoft response to working papers.
1240 Microsoft response to working papers.
1241 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.21.
1242 There are a small number of providers for whom we were not able to obtain MAU data to calculate shares of

supply. However we expect that these providers are very small and would not meaningfully change the
competitive assessment.

296
 the downstream market, and already faced limited competitive constraints
pre-merger.’ 1243

8.401 As described in detail above, we believe that Microsoft already has a strong
position in cloud gaming services as a result of its multi-product ecosystem.
This includes its strengths across PC OSs (Windows), cloud infrastructure
(Azure), and first-party content (Xbox). Whilst rivals have their own strengths,
for example Amazon on cloud infrastructure, NVIDIA on GPU technology, and
SIE on first-party content, Microsoft appears to be in the strongest position as
a result of its combined strength in these three key inputs for cloud gaming
services.

8.402 Before considering the effect of a potential foreclosure strategy by the Merged
Entity on the market for cloud gaming services, we consider the evidence we
have seen on entry and expansion in the market.

Entry and expansion plans

8.403 The evolution of the cloud gaming services market is relevant to our
assessment. Given our assessment is forward looking, entry and expansion
plans absent the merger are helpful in assessing the extent to which
competitive constraints are changing in this market, and also whether any
foreclosure strategy is likely to have an effect on the market. We have set out
the evidence we gathered on third parties’ entry and expansion plans below.

• Amazon Luna

8.404 Amazon launched Amazon Luna in the UK (as well as in Germany and
Canada) in March 2023, with the same game catalogue it has in the US. 1244

8.405 Internal documents indicate that []. 1245 []. 1246

8.406 []. 1247

1243 CMA129, paragraph 7.21.

1244 ‘Amazon Luna blog’ accessed by the CMA on 29 March 2023.
1245 []Internal Document.
1246 [] Internal Document.
1247 [] Internal Document.

297
 • NVIDIA

8.407 NVIDIA submitted that it continues to invest in technological innovations and

adding the latest technology to its cloud gaming hardware and software to
improve its service. 1248 An internal document from NVIDIA shows that []. 1249

• SIE

8.408 An internal document from January 2021 describes []. It shows that []. It
also shows []. It shows []. 1250

• Nintendo

8.409 Nintendo currently has a limited cloud gaming offer but in our view its history
in gaming and strong first party content suggests it could expand. It noted that
its []. 1251

• [ ]

8.410 []. 1252 []. 1253

• Boosteroid

8.411 Boosteroid submitted that it intends to reach 100 million MAUs 1254 by 2026.
This is expected to be primarily due to []. 1255

• Others

8.412 Other possible new entrants to cloud gaming services include Valve, Tencent,
Netflix, []. Valve and [] have not indicated current plans to enter or
expand in cloud gaming. 1256 Tencent also stated that it does not have any
specific expansion plans in cloud gaming. 1257 However, an internal document

1248 [] response to the CMA’s RFI.

1249 [] Internal Document.
1250 [] Internal Document.
1251 [] response to the CMA’s RFI.
1252 [] call note.
1253 [] call note.
1254 MAUs defined as customers per month with active subscriptions. We consider this figure is likely over

ambitious given other industry estimates described above eg the entire cloud gaming market is expected to have
around 100 million revenue driving users in 2025. Nonetheless it demonstrates Boosteroid’s plans for substantial
growth.
1255 Boosteroid response to the CMA’s RFI.
1256 Valve call note; [] response to the CMA’s RFI.
1257 Tencent response to the CMA’s RFI.

298
 from one potential competitor highlights the potential of these firms and how
they might enter into cloud gaming. 1258

8.413 Netflix is another potential entrant to cloud gaming. It indicated that it has
publicly stated that it is exploring adding cloud gaming to its service. []. 1259

8.414 We also note that some smaller current providers may have plans to expand,
such as Blacknut as described above.

Conclusion on strength of different cloud gaming services

8.415 We believe that Microsoft already has a strong position in cloud gaming
services. This can be seen from its current share of the market, and its short-
and long-term advantages derived from its multi-product ecosystem. We
consider we have seen ample evidence of Microsoft’s strengths in its own
documents, and in the assessment of third parties.

8.416 Besides Microsoft, the other key competitors appear to be Amazon, SIE,
Boosteroid, and NVIDIA, []. Whilst these competitors have their own
respective strengths, none appears to have the full breadth of capabilities
important to cloud gaming.

8.417 Amazon only recently launched in the UK, and is yet to launch in all other
countries except the US, Canada and Germany. It currently has a very low
market share worldwide []. It is one of the few players with a strong existing
cloud infrastructure. It lacks, however, first-party content or access to
Windows on equal terms as Microsoft.

8.418 NVIDIA has a relatively high current market share and has a reasonably
strong position in cloud infrastructure as a leading GPU manufacturer. Its
market share of paid users however is much smaller. Although it does not
have any first party content, it has been able to make a large library of third
party content available.

8.419 SIE currently only offers cloud gaming as a way to play older games from its
catalogue. It therefore does not currently offer a strong constraint, although
this may change in the future if it expands its cloud gaming offer. It is likely to
be weaker than other competitors such as Microsoft, Amazon, and NVIDIA on
cloud infrastructure, as it does not have a large cloud infrastructure network or
advantages in building one.

1258 [] Internal Document.

1259 Netflix response to the CMA’s RFI.

299
8.420 Boosteroid has achieved a small market share since entering the market. It
has developed its own cloud infrastructure and although it has no first party
content, its BYOG model allows users to access a wide range of content. It
also has plans for significant growth.

8.421 [] may also become a significant constraint [].

8.422 We consider that cloud gaming therefore appears likely to be a relatively

concentrated market in the UK, with Microsoft enjoying a leading position and
several advantages over its rivals. In a concentrated market, harm to rivals is
likely to constitute harm to competition, unless there is some way to avoid that
harm. 1260 We have considered which competitors might be affected by a
foreclosure strategy, and to what extent, in the conclusion on effect section
below.

8.423 In the section below, we assess barriers to entry and expansion in the market
for cloud gaming.

Barriers to entry and expansion

8.424 This section considers the evidence on barriers to entry and expansion in
cloud gaming. To the extent that barriers to entry and expansion increase the
likelihood of a concentrated market for cloud gaming services, this increases
the effect on competition of the foreclosure of any one individual potential
competitor or group of potential competitors.

Parties’ views

8.425 Microsoft submitted that cloud gaming reduces barriers to entry into gaming. It
stated that, as cloud gaming is device agnostic and does not require the
consumer to invest in hardware, gamers can easily switch and multi-home
across services, reducing network effects. It submitted that evidence on the
video streaming market suggests that many consumers multi-home thereby
reducing barriers to entry as users are not ‘captive’ and are willing to try new
services. 1261

8.426 Microsoft further submitted that a large number of companies have entered
cloud gaming in recent years indicating that barriers to entry are low,
particularly for providers using a BYOG model. 1262 It also submitted that the

1260 CMA129, paragraph 7.21.

1261 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraphs 5.51-5.52.
1262 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 3.92.

300
 existence of cloud gaming infrastructure providers such as Ubitus, further
reduces barriers to entry for new cloud gaming providers. 1263

Our assessment

Third party views

8.427 One competitor [] stated that barriers to entry in cloud gaming were high,
with the two most significant barriers being sourcing cloud infrastructure and
acquiring content. It stated that for its service cloud infrastructure represents
around []% of the total cost of providing a cloud gaming service. It stated
that content represents around []% of the total cost, with the average price
for licensing a AAA game being 10 million USD, often structured as minimum
guarantees. It added that delivering high-quality content requires building a
relationship of trust with game developers so that a platform can provide a
comprehensive library to consumers. It also noted that as consolidation
increases within the gaming sector, it will be difficult for any new or nascent
cross cloud gaming service to establish a compelling library of content. 1264
The same competitor described how a large user base is necessary to attract
developers to release games on a platform. It stated that game developers
want assurance that their games will reach a certain group and number of
users, and that if a cloud gaming platform already has an established user
base, it would therefore be easier to reach an agreement with a game
developer to license their games. 1265

8.428 Another competitor [] submitted that new entrants face several challenges
in providing a cloud gaming service, highlighting cloud infrastructure, OS
licence costs, and acquiring content as the most significant. First, it stated that
Microsoft and Amazon can leverage their large cloud infrastructure networks
to gain a cost advantage over any other competitor. It added that a competitor
using public cloud infrastructure eg Azure, AWS, or GCP, would face
challenges with high costs, and not having control over maintaining and
updating the infrastructure. Second, it stated that the requirement to pay a
licence fee for Windows, or use a less suitable alternative OS, was another
barrier. Third, it stated that cloud gaming services faced a challenge in
obtaining ‘top games’ for their service. Whilst it noted that technical barriers to
adding content to a service could be low, there were barriers on the business

1263 Microsoft response to Supplementary Evidence paper.

1264 [] response to the CMA’s RFI
1265 [] response to the CMA’s RFI.

301
 side in the form of exclusivity arrangements and requests for large upfront
payments. 1266

8.429 Internal documents from another competitor []. They describe []. It also
describes []. It further describes the []. 1267

8.430 Another competitor []submitted that barriers to entry in the wider gaming
market (not restricted to cloud gaming) are generally low. It stated that the
emergence of different business models in gaming had given developers
more options to monetise and bring content to market in a way that makes
sense to them and consumers. It submitted that this was reflected by the
recent entry of competitors such as Amazon Luna, Netflix Games, Apple
Arcade, and Epic Game Store. However, we note that of these, only Amazon
has launched a cloud gaming service. This competitor also stated that
restrictions that Apple imposes on gaming (including cloud gaming services)
on its mobile OS have hindered the growth of cloud gaming. 1268

Conclusion on barriers to entry and expansion

8.431 We believe that the evidence shows that there are significant barriers to entry
and expansion in cloud gaming, including the cost of cloud infrastructure, the
cost of acquiring content, and the need for economies of scale in order to
drive down costs. Most entry to date has been by large companies with
significant existing advantages in gaming or technology and the resources to
overcome these barriers. Although there has been some entry by smaller
companies, these have so far failed to gain any significant market share (with
the possible exception of Boosteroid).

8.432 This increases the potential effect on competition of the foreclosure by an

individual competitor. In a market characterised by network effects and a need
for scale the barriers to expansion for new and emerging entrants are high. If
entry and expansion barriers are already high, then any significant entry is
harder. In these circumstances foreclosing any likely entrant will have a
greater effect on competition.

8.433 The possibility of entry or expansion from game publishers to offset the
possible loss of Activision content is considered in Chapter 9.

1266 []response to the CMA’s RFI.

1267 [] Internal Document.
1268 []response to the CMA’s RFI.

302
Conclusion on effect

8.434 The evidence assessed on the effect of total input foreclosure using Activision
content on competition in cloud gaming services suggests that Microsoft has
significant strengths in the provision of cloud gaming services through its
ownership of Windows and Xbox OSs, its cloud infrastructure capabilities, and
its existing first party content. Whilst a few providers also have some
capabilities relevant to cloud gaming, none is as strong across all important
areas for cloud gaming as Microsoft.

8.435 Cloud gaming services are in their infancy, so current market shares do not
provide the most accurate indication of the relevant strengths of different
competitors. Nonetheless, current shares of supply also point in the direction
of Microsoft being in a leading position in the market.

8.436 As for possible entry and expansion, we note that several providers are
attempting or planning to enter or expand in cloud gaming. In particular
Amazon has recently entered into the UK market, while new entrants [] are
still at early stages and it is not clear what impact their entry may have. This
suggests that, absent the Merger, the market may be poised to benefit from
more competition.

8.437 There appear to be significant barriers to entry in the market, particularly the
cost of cloud infrastructure, the cost of acquiring content, and the need to
grow sufficiently to benefit from economies of scale. There are also material
direct and indirect network effects, although these may be mitigated to some
extent by the possibility of multi-homing.

8.438 The evidence therefore suggests that the market is relatively concentrated at
present. The biggest constraints to Microsoft come from NVIDIA, and
Amazon, both of which are significantly weaker than Microsoft. SIE could
present a greater constraint in the future as it expands its cloud gaming
service, although it has disadvantages on cloud infrastructure.

8.439 Given the small number of existing competitors and Microsoft’s strength, loss
of competition from any of these competitors would be concerning. A
reduction in competitiveness in a market characterised by network effects can
also raise the barriers to entry for others. Having concluded that, absent the
Merger, Activision’s content is likely to become available on cloud gaming
services using a B2P or BYOG model, we believe that the immediate effects
would be felt most strongly by players such as NVIDIA, Boosteroid, Amazon
[]. As discussed in our ability assessment, we think that Activision content is
sufficiently important for each of these competitors to such an extent that it
would materially reduce each of their competitiveness. While we consider that

303
 the case for all of these competitors, we note that not all competitors need to
be foreclosed for foreclosure to result in substantial harm to overall
competition in the downstream market. Furthermore, other competitors could
also be adversely affected by foreclosure, and it could raise barriers to entry
for potential new entrants, or smaller players wishing to expand into providing
AAA games.

8.440 With regard to the Parties submission that the NVIDIA agreement addresses
any concerns (which we understand also now extends to the Boosteroid and
Ubitus agreements), as described above in the ability and incentive to
foreclose sections, we do not consider it is appropriate to place material
weight on these agreements and we do not consider these agreements
materially alter the position as regards the Merged Entity’s overall incentives.
Further, we believe that the ability and incentive to foreclose other competitors
such as Amazon and SIE, as well as smaller players looking to expand
remains unaffected, as well as the ability and incentive to foreclose potential
new entrants, []. Even leaving aside the parties to these agreements with
Microsoft, whilst Amazon has only recently entered the market and [], as
described above both are large companies with advantages that could make
them competitive in the market for cloud gaming, and able to impose a
substantial constraint on Microsoft. Despite its expected disadvantages on
cloud infrastructure, SIE could also provide a constraint given its first party
content and reputation in gaming.

8.441 We have also concluded in the section on ability to foreclose, that in the
counterfactual, Activision content would be a particularly important input to
cloud gaming services, and that any alternatives are likely not to be sufficient
to offset the loss incurred to cloud gaming rivals by foreclosure of Activision
content. We therefore believe that Microsoft using Activision’s content to
foreclose rivals in cloud gaming services would distort the development of the
cloud gaming market and result in substantial harm to overall competition in
this market.

Conclusion on TOH2

8.442 We therefore believe that the Merger may be expected to result in a

substantial lessening of competition in the market for the supply of cloud
gaming services in the UK.

304
9. Countervailing factors
9.1 The CMA’s MAGs indicate that, in some instances, there may be
countervailing factors that prevent or mitigate any SLC arising from a
merger. 1269

9.2 There are two main ways in which this could happen:

(a) the entry and/or expansion of third parties in reaction to the effects of a
merger; or

(b) merger efficiencies. 1270

9.3 As we have found that an SLC may be expected to arise in the supply of
cloud gaming services in the UK, we consider countervailing factors in relation
to this SLC in this chapter. However, as we describe below, evidence on entry
and expansion is also relevant to our assessment of the ability to foreclose in
console gaming in Chapter 7.

Entry and expansion in game publishing

9.4 Entry, or expansion of existing firms, in response to a merger, can mitigate the
initial effect of an acquisition on competition, and in some cases may mean
that there is no SLC. In assessing whether entry or expansion might prevent
an SLC, the CMA considers whether such entry or expansion would be timely,
likely and sufficient to prevent an SLC. 1271

9.5 In terms of timeliness, entry or expansion being effective within two years of
an SLC arising would typically be considered by the CMA to be timely.
However, depending on the nature of the market, the CMA may consider a
period of time shorter or longer than this. In particular, a longer timeframe may
be appropriate in dynamic markets. 1272 In assessing likelihood, we consider
whether there are barriers to entry or expansion in game publishing services
in the UK.

9.6 Although we have found that the Merger may not be expected to result in an
SLC in console gaming services in the UK because the Merged Entity would
not have the commercial incentive to withhold CoD from PlayStation, we

1269 CMA129, March 2021, paragraph 8.1.

1270 CMA129, March 2021, paragraph 8.1.
1271 CMA129, March 2021, paragraph 8.31.
1272 CMA129, March 2021, paragraph 8.33.

305
 consider the evidence on entry and expansion in game publishing to be
relevant for the assessment of the Merged Entity’s ability to foreclose rivals in
console gaming services.

Introduction

9.7 Before moving on to our assessment of any countervailing entry or expansion,

we make some broad observations relevant to that assessment.

9.8 First, in assessing the likelihood of countervailing entry or expansion, it is

important to consider what responses to foreclosure are more likely to
effectively countervail an SLC. In response to foreclosure of Activision content
from cloud gaming providers, game publishers may choose to make new
games for foreclosed cloud gaming platforms. Users may also increase their
gametime or spend on other existing games (to the extent they continue to
use that service).

9.9 We consider that even if certain users were to increase their gametime or
spend on other existing games, this would not prevent a platform’s
competitive offering from being harmed as a result of a lack of access to
Activision’s games. As set out in Chapter 7, with respect to consoles,
Activision’s games are already available on Xbox and PlayStation, and
removing them from one console would harm that console’s competitiveness.
Some consumers would be expected to switch away, whilst those who decide
to stay on that console would no longer be able to play Activision’s games and
would therefore be harmed by the reduction in game range. 1273 We consider
that the same reasoning applies to cloud gaming services that would have
had access to Activision’s games in the counterfactual but would not have
access to those games as a result of the Merger. Fewer consumers would be
expected to choose the foreclosed cloud gaming provider, and there would be
a reduction in game range for those that do choose to use that service.

9.10 Second, as per our counterfactual assessment, we also expect other major
publishers to add their AAA games currently available on PC and/or console
to cloud gaming platforms in the next five years absent the Merger, in a
similar way to Activision games. Therefore, when assessing the importance of
Activision’s content to cloud gaming as part of our ability assessment in
Chapter 8, as well as when assessing entry and expansion of rival publishers,
we have assumed that these AAA games are already part of the competitive
offering of rival cloud gaming platforms. As such, under countervailing factors

In other words, consumers would no longer have access to their first-choice of game(s), and would therefore
1273

move to playing another game(s) in circumstances where they would prefer to play an Activision game.

306
 we are assessing the scope for new entry and expansion triggered by the
Merger to countervail foreclosure.

9.11 Third, in our assessment of entry and expansion, the relevant question is not
merely what action a rival ‘could’ theoretically take. It is also relevant to
consider whether entering or expanding would be a profitable strategy. The
cost to a rival publisher of developing a new franchise or improving an existing
franchise may be prohibitive to the extent it faces high barriers to entry, and to
the extent to which its likelihood of success is risky or uncertain (leading to
lower ‘expected’ or ‘probability-weighted’ returns). Our focus in this
assessment is on whether entry or expansion is likely to happen in response
to the Merger.

9.12 Finally, we note that the market for cloud gaming services is nascent. As part
of our counterfactual assessment, we found that this market is likely to
continue to grow absent the Merger, and that Activision would likely have
placed its content on cloud gaming services. In Chapter 8, however, we found
that the Merger would bring about a structural change in the market that
would give the Merged Entity the ability and incentive to foreclose rival cloud
gaming platforms by withholding Activision’s valuable content. We consider
that the result of this structural change would be to inhibit these platforms
from achieving the growth and scale that they may have reached absent the
Merger. This would significantly reduce or eliminate any incentive for game
publishers to develop new games for those platforms to countervail the effects
of the Merger.

9.13 In any event, and as explored in detail below, we have found that there are
significant barriers to entry and expansion in game publishing, which would
affect the timeliness, likelihood, and sufficiency of any potential entry or
expansion. Accordingly, for the reasons given below, we have found that it is
unlikely that any entry or expansion in terms of new games (whether console,
PC or cloud games) being made available for use on rival cloud gaming
platforms, would prevent or mitigate the SLC that we have identified in cloud
gaming services.

Barriers to entry and expansion

Parties’ views

9.14 In relation to developing console games the Parties submitted:

(a) That barriers to entry in the supply of console games in digital form are
low, reflecting the availability of mature software development tools, such
as game engines and audio and video middleware which make the game

307
 development process less burdensome. The Parties also submitted that it
is increasingly easy to port games developed for PCs to consoles. These
efficiencies allow for PC games to be adapted for consoles without the
need to re-write the game from scratch. 1274

(b) That the cost of developing and publishing console games varies,
depending on the type of game(s) being developed. They stated that the
development and publishing of console games may involve material
investments in terms of costs, resources and marketing, as they are
normally content heavy and offer an advanced gaming experience in
terms of graphics, music, available options, gameplay mechanics, scope
and depth of the storyline. 1275 The Parties also noted that some games
that were developed as low-budget and low-resource games have also
proven to be strong competitors (such as Minecraft and PUBG). 1276

9.15 We do not assess here the submissions by Microsoft regarding whether

Activision’s content is ‘irreplaceable’ or whether there are alternative to it, 1277
as this is addressed in our assessment of ability in Chapter 8.

Our assessment

9.16 In the assessment that follows, we have considered a range of evidence.

Some of that evidence is relevant directly to Activision’s largest titles – CoD
and WoW. Other evidence is related to entry and expansion to replace games
more generally or is of relevance to Activision’s broader library, including, but
not limited to, those two titles. 1278

Third party views

9.17 Third parties stated that it is difficult to develop an alternative to CoD because
the costs and time taken to develop a game like CoD are significant. For
example, one third party [] submitted that AAA titles take a long time to
develop and involve significant production costs. 1279 This third party submitted
that other publishers were unlikely to match CoD’s annual release schedule
and Activision’s expertise and stated that other companies developing a CoD
rival would have higher costs because they cannot build upon existing CoD
game frameworks and experience. 1280 The third party also stated that other

1274 Parties FMN.

1275 Parties FMN.
1276 Parties FMN.
1277 Microsoft’s response to the CMA’s Supplementary Evidence Paper, 12 April 2023.
1278 We note that Activision’s largest franchises (excluding Candy Crush, which is a mobile game) are CoD,

WoW, Overwatch and Diablo.

1279 [] response to the CMA’s RFI.
1280 [], submission to the CMA.

308
 experienced game developers like EA, who already have an incentive to
improve their offering to match CoD, have failed in rivalling CoD’s level of
success. 1281

9.18 A report by IDG submitted by a third party [] dated August 2021 notes that
development budgets are reaching unprecedented ranges. The report
observes that while 5 years ago most AAA console/PC releases had
development budgets between $50-150 million, on average, games that are
greenlit today, with a potential release in 2024-2025, are being approved for
development budgets of $200 million or higher. Also, the report says that
some AAA franchises like CoD have development budgets already exceeding
$300 million, and the next GTA and other future tent-poles are also expected
to hit $250 million or higher. 1282 Activision is also quoted in this report saying
with reference to CoD: ‘We have to make so much content for Call of Duty,
that we can’t even lean on one lead studio anymore. Now we need almost 1.5
lead studios for each annual CoD. That kind of bandwidth pressure is forcing
us to use outsourcers more and more. I don’t see that changing anytime
soon.’ 1283

9.19 The above figures are consistent with submissions we received from other
third parties. For instance:

(a) A publisher [] mentioned that the overall figures for development and
marketing costs for major brands and their recent instalments are
approximately Euro 150 million for pre-launch development costs and
approximately Euro 50 million for launch marketing campaign costs. 1284

(b) Another publisher [] reported development costs for its major AAA
franchises ranging between more than $80 million to almost $350 million
per title, and marketing costs reaching up to $310 million depending on
the franchise. 1285

(c) We received similar ranges from an additional publisher []. Specifically,

it reported a total of development and marketing costs between about
$110 million and almost $380 million for some of its latest major
releases. 1286

(d) A fourth publisher [] submitted that the costs related to developing and
regularly releasing new titles can vary significantly depending on the

1281[].
1282 [] Internal Document.
1283 []Internal Document.
1284 [] response to the CMA’s RFI.
1285 [] response to the CMA’s RFI.
1286 [] response to the CMA’s RFI.

309
 game type or business model of a particular studio. It provided an
example that for one AAA game the development budget value could
range between $90 million and $180 million, whereas the marketing
budget could range between $50 million and $150 million. This publisher
also submitted that for one of its major franchise’s development costs
reached $660 million and marketing costs peaked at almost $550
million. 1287

(e) Finally, another publisher [] estimated that its development costs for
seasonal updates for one of its first party titles range from approximately
$50 million to $65 million. 1288

Internal documents

9.20 The Parties’ internal documents discuss the difficulty in replicating large
franchises such as Activision’s, around which demand coalesces due to
network effects:

(a) One Activision document noted that demand continues to coalesce

around major franchises, with each of the top-10 grossing games
worldwide in the previous year being an established franchise that had
been active for a decade (out of which Activision owns [] franchises).
The document also added that there are [] in gaming, and highlights
Activision’s [] and the []. 1289

(b) One Microsoft internal document explains that [], and that [].
However, the document goes on to note that []. 1290

(c) An Activision presentation noted that ‘while the concept of [] franchises
has long been seen across media and entertainment, gaming franchises
enjoy attributes that are unique to the sector, including []. 1291

9.21 Along with strong network effects, the Parties’ internal documents analyse
various other barriers to entry and expansion in creating and sustaining a
successful AAA franchise. One Activision document lists a range of factors
that limit the threat from new entrants that attempt to build and sustain gaming
franchises. It discusses the broad and deep capabilities required, especially
as [] becomes more critical to []; the fact that [] given []; the [] of
game platforms requiring []; and challenges around []. 1292 The document

1287 [] response to the CMA’s RFI.

1288 [] response to the CMA’s RFI.
1289 Activision Internal Document.
1290 Microsoft Internal Document.
1291 Activision Internal Document.
1292 Activision Internal Document.

310
 states that the durability of game franchises is driven by factors such as
[]. 1293

9.22 Third party reports submitted by the Parties note that strong intellectual
property make franchises successful and durable, with very few big
successes being from new intellectual property:

(a) An analyst report prepared by a third party media and telecommunication

research company noted that ‘franchises are everything’; with the biggest
franchises in video games dominating the industry and with ‘longevity and
relative strength that eclipse even the Hollywood analogue’. The report
noted that these franchises can be built on owned and internally
developed intellectual property, as is the case with most of Activision’s IP,
or leverage others’ intellectual property, like Electronic Arts does with its
respective sports franchises. The report stated that as the industry has
evolved, the importance of franchises has steadily increased, as the
biggest games year-to-year tend to come from established franchises. For
example, of the top 20 best-selling games in the U.S. for 2020, only two
could be labelled as new IP or unrelated to an existing franchise. 1294

9.23 We have also seen evidence to suggest that the current state of the gaming
industry is driven by established franchises and their enduring brand loyalty:

(a) One Activision document noted that concerns over the gaming industry
being [] continue to [] as established franchises []. The top-
grossing game charts are dominated by established IP, with multiple
factors driving the durability of game franchises. The document further
noted that [] as Activision grows []. 1295

(b) An analyst report prepared by a third party media and telecommunication

research company also noted that, over the past decade, ‘three mega-
trends – digitization of distribution, in-game monetization, and franchise
management – have transformed the video game business from what was
a solid, albeit volatile, business, to what today is a very good, and much
less volatile, business’. The document also notes that ‘today’s video game
business is vastly better than the hit or miss driven industry of the
past’. 1296 Whilst Activision submitted that this document was prepared by
a third party research company, we consider that we can place weight on
this evidence given that the document’s purpose was to provide

1293 Activision Internal Document.

1294 Activision Internal Document.
1295 Activision Internal Document.
1296 Activision Internal Document.

311
 information to institutional or professional clients considering investing in
the video game industry. 1297

Other evidence on barriers to entry and expansion

9.24 In this section, we first consider whether the presence of network effects at a
game and/or platform level has an impact on the barriers to entry and
expansion for games on cloud gaming and console services. Second, we
consider the following additional barriers faced by a publisher to build a new
game/franchise or reposition an existing game to make it better and more
attractive to cloud and/or console gamers: (i) development costs; (ii)
developer personnel; (iii) time taken for development; and (iv) other potential
barriers such as existing content, brand awareness, and the likelihood of
success.

• Network effects

9.25 As evidenced above, we consider there are strong direct network effects at a
game level prevalent in both console and cloud gaming, particularly for multi-
player games. This means that multi-player games create higher switching
costs than single-player games because groups of gamers (eg friends) will
have to coordinate their switch to a different franchise if they want to keep
playing together. This therefore increases barriers to entry and expansion for
new games. For instance, a report by IDG dated August 2021 and submitted
by a third party [] mentioned that multiplayer sessions are more engaging
and tend to retain players longer. 1298 Microsoft submitted that there is no
evidence that network effects are a barrier to entry as some 1,500 multi-player
PC games were published on Steam in 2022. 1299 In this context we note that
Activision games, and particularly its largest franchises – CoD, WoW,
Overwatch, and Diablo, are multiplayer games.

9.26 In consoles, there are also network effects at a platform level – some gamers
might be keen to play with gamers on the same console. This implies that any
hypothetical withdrawal or degradation of Activision games on rival platforms
may generate an opportunity for other games to capture some of the players
not willing to switch away from those platforms and increase their user base.
The persistence of these network effects at a platform-level post-Merger
would lower barriers to entry and expansion for new games.

1297 Activision response to working paper.

1298 [] Internal Document [].
1299 Microsoft response to the CMA’s Supplementary Evidence paper, based on Steam stats, accessed by the

CMA on 17 April 2023.

312
9.27 Therefore, in consoles, assessing the impact of network effects on any
barriers for games to expand would require a comparison of the strength of
network effects at the platform level relative to those at a game level. The
Parties submitted that network effects at a platform level are diluted by the
presence of cross-play for large multi-player games and instead exist at a
game level; 1300 however, we note that it is difficult to measure the effect of
network effects on barriers to entry and expansion empirically. We consider
network effects in conjunction with other barriers to entry in the assessment
below.

9.28 In relation to cloud gaming, we consider platform-level network effects (ie

network effects for a specific cloud gaming provider) to be less relevant (see
further Chapter 8).

• Development costs

9.29 Data from the Parties and third parties shows that Activision games overall,
and in particular CoD and WoW, had significantly higher development costs
than other games, including SIE’s games and third party publisher games
(including shooters).

(a) Activision data showed that Activision spent more than [] in
development costs on the CoD franchise in 2021, with an annual
development cost of [] for each title. The annual development cost was
similarly approximately [] in 2020, and [] in 2019. For WoW Activision
spent [] in 2021, [] in 2020 and [] in 2019, while the sum of
development costs for all top 12 Activision franchises was over [] in
2021. 1301

(b) By comparison, [] data on the development costs of four of its AAA
games – [] – indicates that each of these titles cost [] to develop
overall, with annual costs of []. 1302 Similarly, the costs of other games
and shooters have been lower than CoD as per game industry reports; for
example, while we do not have annual spending figures, it has been
reported that Battlefield 4, the game closest to CoD in terms of gameplay
features, is reported to have cost $100 million in total. 1303

9.30 As noted by the Parties, development costs vary significantly depending on

various factors such as the type of game being created, the size and

1300 Microsoft response to working papers.

1301 Parties Internal Document.
1302 [] Internal Document.
1303 The 12 Most Expensive Games Ever Made & How Much It Cost To Make Them’, accessed by the CMA on

23 January 2023.

313
 experience of the development team, the length of the game development
process, whether or not the game is offered as an ongoing live service, the
amount of marketing and promoting, as well as the distribution channels. 1304

9.31 We note that CoD in particular has various features noted by the Parties
above that lead to it having significantly higher development and ongoing
maintenance costs than most other games, such as:

(a) Multi-player game with complex storyline: CoD is a large multiplayer

game, with high graphical intensity and complex storylines that take time
and resources to develop. CoD is also offered as a live-service game.

(b) Frequent (annual) releases: Activision releases a CoD game every year.
This contributes to CoD’s success as it draws gamers to the franchise due
to its familiarity.

(c) Developers: As evidenced further below, CoD has a large number of

developers working on it to enable the frequent releases and offer the
complex multi-player experience noted above.

(d) Multiple rotating studios: Each release in the CoD franchises takes 3-5
years to develop given the complexity of the game and frequency of
release. Activision has four rotating lead studios (Sledgehammer, Infinity
Ward, Treyarch and Raven Software), each supported by several other
studios. For example, more than 11 studios reportedly contributed to
developing Call of Duty: Modern Warfare 2. 1305 Evidence from Activision
showed that in the UK alone, [] worked on ‘development, localization,
and tech’ of CoD games. 1306

(e) Marketing and distribution: CoD also has high marketing and
promotional budgets relative to most other games to support its frequent
releases, and is distributed through various channels.

9.32 A report by IDG submitted by a third party [] dated August 2021 also
emphasised the increasing cost of developing games. For instance, the report
mentioned AAA development is becoming more complex with more moving
parts, including live operations and multiplayer feature sets. AAA development
also requires a larger number of platforms to create content for, which also
increases development budgets. 1307 As regards other factors that play a role
in these increasing costs, the report listed increasing labour costs, gamers

1304 Parties FMN.

1305 ‘Call of Duty Leaker Claims 11 Studios are Working on Modern Warfare 2’, accessed by the CMA on 30
December 2022.
1306 Activision site visit.
1307 [] Internal Document.

314
 asking for more content both at launch and through live operations later, talent
shortages, rapid growth of the games market, and cross-platform
development. 1308

9.33 The evidence above indicates that developing a multi-player franchise with a
complex storyline and frequent releases such as CoD would require
significantly high development and promotional budgets.

• Developer personnel

9.34 The Parties stated at the site visit and the Main Party Hearing that []. 1309
The Parties’ and third parties’ data shows that CoD had more developers
working on the game than most other titles, including Microsoft and SIE’s first
party titles.

(a) Activision data shows that CoD utilised more than 2000 full-time
employees (FTE) in 2021. 1310 Activision’s Annual Report states that, of its
c. 9000 employees, over 3000 people are now working on the
franchise. 1311

(b) The Activision data also shows that WoW had over [] FTEs in 2021,
Overwatch had over [] and Diablo had over []. 1312

(c) Also, as noted above, a report by IDG submitted by a third party []
quoted Activision Blizzard as saying that it has so much content for CoD
that it needs almost 1.5 lead studios for each annual CoD. Further,
Activision is claimed in the report to have said that it typically uses about
15 outsourcing/co-development partners for a $100 million game, and 20-
30 different partners for a larger game. 1313

(d) By comparison, in 2021 Halo utilised [], Minecraft [] and Forza used
[] developers only, which included both FTE and external
developers. 1314

(e) Similar figures were submitted by another third party [] for [] and [].
Specifically, roughly [] to [] FTEs were estimated to support the
development and release of any given seasonal update of []. [] is

1308 [] Internal Document.

1309 Activision, Main Party Hearing transcript.
1310 Parties Internal Document.
1311 Activision Annual Report 2021, page 2, accessed by the CMA on 30 January 2023.
1312 Parties Internal Document.
1313 [] Internal Document.
1314 Parties response to the CMA’s RFI.

315
 currently supported by approximately [] to [] personnel (both FTEs
and part time). 1315

(f) Four of [] top first-party AAA titles by revenue from the past five years
had a combined peak workforce of [] FTEs. 1316

9.35 The same report by IDG dated August 2021 cited above and submitted by a
third party [] also emphasised the importance and costs of developing
personnel. Specifically, the report observes that in-house labour costs for
game developers, artists, engineers are going up significantly because of
talent shortages and the rapid growth of the games market. 1317

9.36 We therefore consider that a new entrant or an existing publisher would

require a significant number of developers to build a game that is equivalent
to CoD. We consider that building a game equivalent to Activision’s other
major franchises (WoW, Overwatch, and Diablo) would also require a
significant number of developers, given in 2021 the number used was [].

• Development time

9.37 We also note that large multiplayer games like CoD, WoW, and Overwatch
takes significant time to develop. Evidence submitted by third parties and from
public sources indicates that large AAA games take about 3 to 5 years to
develop. 1318

9.38 For instance, a report by IDG dated August 2021 and submitted by a third
party [] reported that each CoD release has a 4 to 5 year development
cycle that is compressed into 3 years. 1319

9.39 Also, one publisher [] submitted that for the titles of its major franchises the
duration of the pre-release development phase is typically 3 years, however it
may take even longer (up to 8 years) in case of substantial changes to the
game mandate in the course of the development. 1320

9.40 We note that similar to development costs above, the time taken to develop a
game can vary substantially depending on the type of game, its gameplay
features, budget, etc. However, as above, given CoD is a multi-player game

1315 [] response to the CMA’s RFI.

1316 [] Internal Document.
1317 [] Internal Document.
1318 See for example How Long Does It Take To Develop a Video Game? (geekygamingstuff.com), accessed by

the CMA on 27 January 2023.

1319 [] Internal Document.
1320 [] response to the CMA’s RFI.

316
 with a complex storyline and extensive gameplay features, we are of the view
that it takes longer to develop a game like CoD relative to many other games.

• Content, brand awareness and likelihood of success

9.41 The evidence above also indicates that, in addition to the costs, time and
developer resources required to build large AAA franchises such as CoD,
WoW, Overwatch, and Diablo, there are additional barriers to entry and
expansion:

(a) Existing content: Internal documents and third party views mentioned
above indicate that strong ‘intellectual property’ (which we understand to
mean content in this context) makes franchises successful and durable,
with very few big successes being from entirely new content/franchises.

(b) Brand awareness: We further note that large games such as many
Activision franchises have a high level of gamer awareness and loyalty to
the brand, which may be difficult to replicate for a new game. These
barriers would be lower for existing games wanting to expand as they
would already have an existing user base; however, specifically in relation
to CoD the evidence in Chapter 7 shows that CoD’s closest alternatives
are significantly behind CoD in terms of their consumer spend and
gameplay time on PlayStation, and are likely to still face some barriers to
expansion in terms of their existing user base.

9.42 Likelihood of success: The evidence above also indicates that the
probability of success is relatively low for a new game (or for the expansion of
an existing small game). For example, as discussed in Chapter 7, our
assessment of data submitted by the Parties shows that only []% of titles
have made it to the top 25 games over the past five years. 1321 While not
Microsoft’s own view, we have also seen an email from the [] mentioning
[] difficulties in entering the gaming market. Specifically, the email notes
‘[].’ 1322 This suggests that trying to emulate large AAA franchises is very
risky and poses an additional barrier to entry and expansion in game
publishing.

9.43 Games like Fortnite and PUBG have been successful with lower development
costs and no existing brand recognition. However, while successes such as
these do emerge, Activision’s success and importance has been durable over
time. Whilst some publishers may enjoy significant success, individual

1321 Microsoft response to the phase 2 Issues Statement, 31 October 2022, paragraph 3.43f.
1322 Microsoft internal document.

317
 examples of such success are not sufficient to demonstrate that barriers to
entry and expansion are ‘low’.

Conclusion on barriers to entry and expansion

9.44 For cloud gaming the relevant question to our assessment here is whether
new entry or expansion by rival games will occur to such an extent that the
impact of making Activision content unavailable to rival cloud gaming service
providers would be reduced to a level where any SLC is prevented or
mitigated (relative to a scenario where Activision content had become
available). It is in this context that we assessed barriers to entry and
expansion. Based on the above evidence, our view is that there are some
significant barriers to entry and expansion pre-Merger in publishing a large,
successful, and enduring franchise such as CoD and other Activision content,
particularly in terms of budget and skilled employees.

9.45 We consider the evidence presented above to be relevant for the assessment
of the ability of the Merged Entity to foreclose in console gaming services as
well. Based on the evidence above, we find that there are some significant
barriers to entry and expansion pre-Merger in publishing a large, successful,
and enduring franchise such as CoD for console.

9.46 Moreover, in both console and cloud gaming services, established games
have a competitive advantage relative to newly conceptualised games. While
there are examples of successful AAA games that were developed with a
relatively low budget and in a short period of time (eg Fortnite, PUBG), we
consider that this is not indicative of the barriers to entry and expansion for an
exceptionally successful AAA game such as CoD (or other Activision games)
more generally.

Recent entry or expansion and prospects of further entry and expansion

9.47 In this section, we consider whether there have been any examples of entry or
expansion in the markets for game publishing (whether for PC or console) in
recent years which may indicate whether this may also be likely post-Merger
in the event of rival cloud gaming services being foreclosed from accessing
Activision’s content. We also consider the evidence relating to the prospects
of entry or expansion following the Merger. We note that most of the evidence
is relevant to console gaming as well, and therefore is relevant to our
assessment of the Merged Entity’s ability to foreclose in console gaming
services discussed in Chapter 7 above.

318
Parties’ views

9.48 Microsoft submitted that there have been various tech and entertainment
entrants in recent times in game publishing, such as Amazon (which has
published two games in the top 10 most-played games on Steam—Lost Ark
and New World); Warner Bros. Games (Lego, Mortal Combat, Harry Potter);
Netflix; and Facebook. 1323

9.49 Microsoft also submitted that there has been consistent entry and a reduction
in concentration in game publishing over the past decade, as the industry has
expanded and become fragmented. It submitted that academic research
shows that, while the combined revenues of the 10 largest game publishers
has grown from $23 billion in 2002 to $150 billion in 2021, their total market
share in the global games industry has dropped from 90% to 57% over that
same period. 1324

Our assessment

Third party views

9.50 In relation to our assessment of the potential foreclosure of SIE in the console
gaming market, we asked publishers of other games on PlayStation whether
their strategy and incentives would change in the event that CoD were to be
no longer available on PlayStation.

9.51 Publishers stated that their strategy would likely remain unchanged. For
example, one publisher [] told the CMA that it believes that its strategic
incentives regarding video game development and publication would remain
unchanged and that CoD no longer being available on PlayStation would be
‘unlikely to materially affect the opportunities available to grow [their games’]
user base.’ 1325 Another publisher [] stated that it does not believe that it
would present them with a meaningful opportunity to grow and expand. 1326
We note, however, that the CEO of one rival publisher stated that, if CoD
were no longer available on certain platforms, it would provide a tremendous
opportunity for their rival shooter franchise. 1327

1323 Microsoft response to working papers.

1324 Microsoft response to working papers.
1325 [] response to the CMA’s RFI.
1326 [] response to the CMA’s RFI.
1327 ‘Electronic Arts, Inc. (EA) Goldman Sachs Communacopia + Technology Conference Call Transcript’

accessed by the CMA on 18 January 2023.

319
9.52 We also asked SIE to provide internal documents that refer to or discuss
plans, contingency plans, and/or strategies aimed at reacting to and/or
reducing any perceived impact of the Merger.

(a) SIE emphasised the importance of CoD to PlayStation and the impact that
losing it would have on PlayStation, []. SIE also submitted that if the
Transaction were allowed to proceed, []. 1328

(b) A review of SIE’s internal documents was consistent with the above
statements [], albeit the CMA considers limited weight should be placed
on this given these strategies and documents were developed in the
context of the ongoing merger review. In any event, given the evidence
set out above regarding the importance of CoD to PlayStation, we
consider that SIE would not be able to effectively mitigate any SLC that
arises as a result of the Merger through expansion. Although we have not
found an SLC in console gaming services, this evidence is nevertheless
relevant to our assessment of countervailing entry or expansion in game
publishing for cloud gaming services.

9.53 We consider that if SIE and other game publishers were unlikely to change
their strategy if CoD were removed from PlayStation, they would be even less
likely to do so as a result of Activision’s content being unavailable on cloud
gaming providers, given that the market for cloud gaming services is nascent,
and most platforms currently lack the scale or user base necessary to
incentivise game publishers to develop new games specifically for those
platforms (which, we have found, will only be exacerbated by the Merger).

Internal documents

9.54 The Parties’ internal documents discuss the failure of large [] companies
previously not present in game publishing in building successful game
software businesses.

(a) One Activision document notes how many new entrants in gaming,
despite being [], have failed over the years. For example, it discusses
how: 1329

(i) Between 2005 and 2016, [], before exiting the game publishing
business.

1328 See, for instance, SIE response to the CMA’s s109 notice; SIE response to the CMA’s s109 notice; SIE
response to the CMA’s s109 notice; SIE, follow-up response to the CMA’s s109 notice.
1329 Activision Internal Document.

320
 (ii) [] in the category in 2014.

(iii) [].

(b) Another Activision document notes that [] continue to get bigger, and
invest in gaming [], while lacking []. The document cited examples of
[]. 1330

9.55 While we note that the documents also discuss [], we consider the evidence
overall indicates the difficulty in creating high-quality successful content [].

Conclusion on recent entry of expansion and prospect of further entry or

expansion

9.56 Overall, while there has been entry and expansion of games published on PC
and console, the above evidence suggests that even large companies can
struggle to make large, successful games. In particular, we note that, with the
exception of Fortnite, there are few games that (i) are at a similar size and
scale to CoD; and (ii) are sufficiently similar to CoD to demonstrate the
feasibility of developing an effective alternative to CoD.

9.57 Additionally, we have not seen evidence to suggest that SIE or third party
publishers would enter or expand by creating a new game that can replace
CoD, or other significant Activision content. As such, as discussed in above,
we consider they would be even less likely to do so as a result of Activision’s
content being unavailable on cloud gaming providers. Overall, this suggests it
would be unlikely publishers would develop or expand games to a sufficient
extent to replace Activision’s major franchises such as CoD and WoW.

Conclusion on entry and expansion in game publishing

9.58 We have considered the timeliness, likelihood and sufficiency of any entry or
expansion from existing players and third parties in game publishing on cloud
gaming services, and the extent to which this could prevent or reduce the
impact of any SLC from arising as a result of the Merger.

9.59 Our assessment has focused on the extent that games will be developed or
expanded that can replace Activision major franchises, such as CoD and
WoW. We have given careful consideration to the range of games currently
available, the evidence that demand coalesces around big, familiar
franchises, and that those franchises (particularly CoD and WoW along with
other such as Overwatch and Diablo) have demonstrated significant staying

1330 Activision Internal Document.

321
 power. We also note that the skills, assets, and resources that have given rise
to Activision’s position across its games are the same factors that will
determine its ability to come up with new successful content in future. While
we realise that the list of games is not static and publishers frequently release
new games, some of which go on to be very successful, it is in this context
that we consider whether developed or expanded games will be timely, likely
and sufficient to replace Activision content.

(a) In relation to likelihood, we consider the evidence above (in particular on

third parties’ plans to expand and the significant barriers to entry) shows
that it is unlikely that publishers would create a new game or expand an
existing game that becomes a close substitute for Activision’s major
franchises such as CoD and WoW, to mitigate the impact of any SLC in
cloud gaming services.

(b) Given the evidence above on the time taken to develop a major franchise
such as CoD or WoW, we consider unlikely that an effective alternative to
Activision’s content, would emerge in the near future on cloud gaming
services.

(c) In relation to sufficiency, the evidence shows that, whilst it is likely that
there will be new games that are developed, they are not likely to be an
effective alternative to Activision’s major franchises such as CoD or WoW.

9.60 Therefore, our view is that it is not likely that entry or expansion of sufficient
scale would occur in a timely manner in order to prevent or reduce the impact
of the SLC in cloud gaming services in the UK from arising as a result of the
Merger.

Entry and expansion due to cloud gaming agreements

9.61 We also note that Microsoft submitted that, to the extent that the CMA takes
the view that the NVIDIA Agreement does not represent an efficiency (which
is discussed further below), the CMA must take into account the agreement
as a countervailing factor offsetting any lessening of competition identified.
Microsoft referred to paragraphs 8.28-8.30 of the MAGs when making this
point, so we take this as a submission on entry and expansion. Microsoft
presented this on the basis that, by making additional popular content
available on the NVIDIA platform, the Merger will result in NVIDIA expanding
its offering and exerting a stronger competitive constraint on the Merged

322
 Entity. 1331 We assume the same submission applies to the Boosteroid and
Ubitus agreements.

9.62 We have already considered these agreements in the context of the Merged
Entity’s ability and incentive to engage in foreclosure in the cloud gaming
market in the UK in Chapter 8, and we consider the same points apply here.
In particular, contracts may be of limited duration and, over time, may be
renegotiated or terminated and firms may waive their rights to enforce any
breaches in light of their overall bargaining position. There is also specific
uncertainty as a result of certain clauses in those agreements, as discussed in
Chapter 8 and below in Chapter 11. We consider this is further exacerbated
by our findings that Microsoft more generally would have the incentive to
foreclose cloud gaming rivals. We consider, as a result of these limitations,
that these agreements are insufficient to demonstrate that they will result in
expansion as a result of the Merger that will be timely, likely and sufficient to
prevent an SLC from arising.

Efficiencies

9.63 Efficiencies arising from a merger may enhance rivalry with the result that the
merger does not give rise to an SLC. In order for us to take efficiencies into
account, they must:

(a) enhance rivalry in the supply of those products where an SLC may
otherwise arise;

(b) be timely, likely and sufficient to prevent an SLC from arising;

(c) be merger-specific; and

(d) benefit customers in the UK. 1332

9.64 The MAGs make it clear that merger firms who wish to make efficiency claims
are encouraged to provide verifiable evidence to support their claims in line
with the CMA’s framework. 1333 The MAGs note that it is for the merger firms to
demonstrate that the merger will result in efficiencies. 1334

9.65 In its response to the Provisional Findings, Microsoft put forward efficiencies
in relation to both the supply of console gaming and in the supply of cloud
gaming services. 1335 As we have found that an SLC may not be expected to

1331 Microsoft response to the provisional findings, 2 March 2023, paragraph 4.21.
1332 CMA129, paragraph 8.8.
1333 CMA129, paragraph 8.7.
1334 CMA129, paragraph 8.15.
1335 Microsoft response to the Provisional Findings, 2 March 2023, section 4.

323
 arise in the supply of console gaming services in the UK, we do not consider
efficiencies relating to that market further.

Parties’ views

9.66 Microsoft submitted that its agreement with NVIDIA constituted a rivalry-
enhancing efficiency in cloud gaming. Microsoft explained that currently
Activision content is not available on any cloud gaming service and that, as a
result of the NVIDIA Agreement, Activision content will be made available on
NVIDIA GFN post-Merger. 1336 Microsoft submitted that this would enhance
competition in cloud gaming and result in a greater choice of goods and
services. 1337

9.67 Microsoft submitted that the benefit was timely and likely, as under the
NVIDIA agreement, should the merger proceed, Activision content will be
made available on multiple cloud gaming services. Microsoft added that
[]. 1338

9.68 Microsoft further submitted that the benefits of the efficiency would be
substantial. Microsoft explained that by making this content available for
streaming, customers will face a lower cost of accessing Activision’s high-
performance games as they will no longer need to purchase a console or
high-performance PC. Microsoft explained that it estimated that there were
over 1 billion PC gamers globally and 18 million UK gamers that could not
currently play CoD as they owned a low-performance PC, but would be able
to do so post-Merger. 1339

9.69 Microsoft also claimed that the efficiency was Merger-specific. Microsoft
explained that the Parties disagreed with the CMA’s provisional conclusion
that, absent the Merger, []. Microsoft also explained that even if this
provisional conclusion was correct, there is []. Microsoft submitted that
following the signing of the NVIDIA agreement, the Merger will ensure that
Activision content is available on multiple cloud gaming providers, thereby
enhancing rivalry compared to the counterfactual and delivering significant
benefits to consumers. 1340

9.70 Microsoft further submitted that the benefits of the efficiency will offset the
SLC identified by the CMA. Microsoft submitted this was because the cloud
gaming segment is de minimis and is expected to remain so and that, as a

1336 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 4.11.
1337 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 4.12.
1338 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 4.12(b).
1339 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 4.12(c).
1340 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 4.12(d).

324
 result, any SLC will be limited. Microsoft submitted this was even more so
given that in reality the SLC []. 1341

9.71 In addition to its efficiency claim, in its response to the Remedies Notice,
Microsoft submitted that the NVIDIA agreement qualified as an RCB, to be
taken into account as part of our remedies assessment. We consider the
question of the claimed Cloud Gaming RCB in Chapter 11 (Remedies) below.

Our assessment

9.72 We consider below the claimed efficiency arising from the NVIDIA agreement
in relation to the factors set out above in line with our published guidance.

9.73 We note that, subsequent to its response to the Provisional Findings,

Microsoft also signed agreements with Boosteroid and Ubitus. While we did
not receive specific efficiencies submissions on these agreements, we have
also included these in our assessment for completeness, as we consider that
similar considerations apply.

Merger-specificity

9.74 In line with our guidance, we consider here whether the claimed efficiency is
reliant on the Merger, or whether it would be brought about by other
means. 1342

9.75 In Chapter 8 above, we concluded that Activision would likely have made its
games – including day and date releases – available on cloud gaming
services in the next five years. This is not [], nor is it limited to Activision
only making its content available to a single cloud gaming service, as
Microsoft has suggested.

9.76 In this context, we therefore consider it likely that any benefits that may arise
as a result of Activision content being made available to rival cloud gaming
service providers under the NVIDIA, Boosteroid and Ubitus agreements would
accrue absent the Merger. Further, the evidence does not suggest the
Merged Entity would have a greater incentive to enter into these agreements
than Activision would absent the Merger. 1343 Rather, we consider Activision’s
incentives to make its content available on cloud gaming services absent the
Merger are greater than Microsoft’s incentives post-Merger. We note also our
assessment of merger-specificity in relation to the Cloud Gaming RCB in

1341 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 4.13.
1342 CMA129, paragraph 8.16.
1343 CMA129, paragraph 8.17.

325
 Chapter 11 (Remedies) below, which we consider equally applies here. As a
result, we consider that any efficiency arising from the NVIDIA, Boosteroid or
Ubitus agreements cannot be considered to be Merger-specific.

9.77 As any efficiencies relating to the agreements with NVIDIA, Boosteroid and
Ubitus are not Merger-specific, we do not consider that they constitute
efficiencies that may prevent an SLC by offsetting any anti-competitive
effects. However, even if these agreements were Merger-specific, they would
not have a material impact on the SLC for the reasons given below.

Timeliness/likelihood

9.78 We note Microsoft’s claim referenced above that []. Should this claim be
realised, the claimed efficiency would be felt within the timeframe covered by
the CMA’s competitive assessment. If similar timeframes could be achieved in
relation to the agreements with Boosteroid and Ubitus, any efficiencies arising
from those agreements may also be felt within the timeframe covered by the
CMA’s competitive assessment.

9.79 However, in terms of likelihood, our guidance is clear that the merger
efficiencies must be likely to be realised. 1344 We note that contracts in general
contain an inherent degree of uncertainty. For instance, over time contracts
may be renegotiated or terminated, and firms may waive their rights to
enforce any breaches in light of their overall bargaining position. Further, in
relation to these specific contracts, there are certain clauses which we
consider introduce further uncertainty, which we have already discussed
above in Chapter 8 (and we also refer to our analysis below of the
agreements in Chapter 11, which is also relevant here). As such, we consider
there is considerable uncertainty as to whether these agreements are likely to
result in rivalry-enhancing benefits within a reasonable timeframe.

Rivalry-enhancing/sufficiency and benefits to UK customers

9.80 We accept that any efficiencies resulting from the agreements with NVIDIA,
Boosteroid and Ubitus would be capable at least in principle of benefiting
some customers in the UK, in the sense that they may bring Activision’s
content to these rival cloud gaming providers. However, we have already
discussed above the uncertainties which exist in relation to these agreements
and the analysis in this regard in Chapter 8 and Chapter 11 is also relevant
here.

1344 CMA129, paragraph 8.13.

326
9.81 In Chapter 8 above, we identified an SLC in relation to the supply of cloud
gaming services in the UK. The SLC arises due to vertical effects resulting
from input foreclosure. As our concern relates to the withholding of Activision
content generally by Microsoft, and the impact this would have on competition
in the market, provision of this content that is limited to certain providers (who
are all either BYOG or B2B providers) would not enhance rivalry across the
market and would rather put select market participants at an advantage
compared to the rest of the market. Even if these participants were to account
for a large portion of the market, an advantage given to certain market
participants alone would not in our view be rivalry enhancing, particular in the
context of a nascent market with other participants using a variety of business
models (also not always involving Windows OS) and where future
entry/expansion and changes in the competitive landscape are to be
expected.

9.82 Further, any efficiency benefit must be considered in the context of our
conclusion that the Merged Entity would have the ability and incentive to
pursue a foreclosure strategy in the supply of cloud gaming services in the UK
where Microsoft has a strong position, reducing the constraint from a range of
actual and potential competitors. We have concluded that Activision content,
including day and date releases, would likely have been made available on
cloud gaming services absent the Merger. As such, and in the context of a
nascent and growing market, 1345 we find that any additional benefit of
Activision content being available on NVIDIA, Boosteroid, and Ubitus’ cloud
gaming services under the terms of these agreements would be small and
transitory, and is likely to be considerably less significant in magnitude than
the harm of it being withheld from the rest of the market more generally.

Conclusion on efficiencies

9.83 Taking into account the available evidence and the considerations set out
above, our view is that it is not likely that any efficiencies arising from the
Merger will be of sufficient magnitude and benefit to UK consumers to prevent
the SLC we have found in the supply of cloud gaming services in the UK.

See further Chapter 8 for our assessment of the growth prospects of the cloud gaming services market which
1345

we consider contradict Microsoft’s submission that the market is expected to remain de minimis in future.

327
10. Conclusion on SLC
10.1 As a result of our assessment, we conclude that the anticipated acquisition of
Activision by Microsoft constitutes arrangements in progress or in
contemplation which, if carried into effect, will result in the creation of an RMS.

10.2 We also conclude that the creation of that situation may be expected to result
in an SLC in the supply of cloud gaming services in the UK, due to vertical
effects resulting from input foreclosure.

328
11. Remedies

Introduction

11.1 Having found that the Merger may be expected to give rise to an SLC in the
market for cloud gaming services in the UK, we must decide whether, and if
so what, action should be taken to remedy, mitigate or prevent that SLC or
any adverse effects resulting from the SLC. 1346

11.2 This chapter sets out our assessment of, and final decision on, the
appropriate remedy to the SLC and resulting adverse effects we have found.
In particular, this chapter discusses:

(a) our remedy consideration process;

(b) framework for the assessment of remedies;

(c) overview of remedy options;

(d) effectiveness of prohibition of the Merger;

(e) effectiveness of partial divestiture;

(f) effectiveness of behavioural remedies;

(g) consideration of the proportionality of effective remedies, including

assessment of potential RCBs put forward by Microsoft; and

(h) our final decision on remedies.

Our remedy consideration process

11.3 On 8 February 2023, we notified the Parties that we had provisionally

identified an SLC in two markets in our Provisional Findings: the market for
console gaming services in the UK and the market for cloud gaming services
in the UK. Alongside the Provisional Findings, we published a Notice of
Possible Remedies (the Remedies Notice). In the Remedies Notice, we
sought views on possible remedies to these SLCs. In particular, we sought
views on prohibition of the Merger, partial divestiture of the Activision
business, and behavioural remedies. The Parties responded to our Remedies
Notice on 22 February 2023, and separate response hearings were held with
Microsoft on 27 February 2023, and with Activision on 1 March 2023. The

1346 Section 36(2) of the Act.

329
 Parties provided further submissions following their response hearings (on 6
March 2023 in the case of Microsoft and 7 March 2023 in the case of
Activision).

11.4 On 24 March 2023, we notified the Parties of the Addendum to the Provisional
Findings in which we set out our provisional view that additional evidence
received since the publication of our Provisional Findings meant that we
considered there to be only one provisional SLC, ie in cloud gaming services
in the UK. On the same date, we shared a working paper with the Parties (the
Remedies Working Paper), which set out our provisional decision that
prohibition of the Merger would be the only effective and proportionate
remedy to the provisional SLC in cloud gaming services based on our
consideration of the written and oral responses received from Microsoft,
Activision and third parties following the Remedies Notice.

11.5 On 31 March 2023, we received a response to the Remedies Working Paper

from Microsoft, and held a call with the Parties on 4 April 2023 to discuss its
response to the Remedies Working Paper.

11.6 In addition, we contacted a number of third parties to discuss potential remedy

options. We held calls with seven third parties and received questionnaire
responses from eight third parties. 1347 On 30 March 2023, we received a
submission from SIE setting out its observations on Microsoft’s published
responses to the Remedies Notice, and a further submission from SIE on 11
April 2023 setting out its views on Microsoft’s revised remedy proposal of 3
April 2023.

11.7 As the Addendum to the Provisional Findings was not issued until 24 March
2023, many of the earlier submissions, responses and discussions refer to
both the provisional SLC in the supply of cloud gaming services in the UK,
and the provisional SLC in the supply of console gaming services in the UK
(which was subsequently provisionally cleared). On the basis that we have

1347 We received written responses to our Remedies Notice from Market participant B [] (20 February 2023;
noting this was a response to both the Remedies Notice and the Provisional Findings) and SIE (22 February
2023). We held calls with the following third parties: [], [], [], [], [], [] and []. We also held a
hearing with []. We received responses to our remedies questionnaire from: [], [], [], [], [], [],
[] and []. We also received responses from members of the public to our Remedies Notice. It was necessary
for us to apply caution in interpreting these responses from the public. There is no practical way for us to verify
that views provided through an open channel are submitted by genuine consumers, nor that those submitting
views are representative of all relevant consumers (and in particular, UK consumers). In relation to
representativeness in particular, open calls for comment may attract greater attention from particular cohorts of
the market than others, either through random chance, or as a result of varying degrees of awareness across
different cohorts (which can be affected by press coverage, internet commentary and social media coverage—
including coverage by interested parties). As a result, we were unable to place any particular weight on these
email submissions, or to use them to make inferences about ‘average views’ of consumers. However, we note
that many respondents raised issues and concerns that were aligned with those under consideration in this
chapter.

330
 only found an SLC in the supply of cloud gaming services in the UK in this
Final Report, this chapter considers the evidence and views that are relevant
to this SLC finding.

CMA remedies assessment framework

11.8 Pursuant to section 36(2) of the Act, where the CMA decides that an
anticipated merger may be expected to result in an SLC, we must decide the
following:

(a) whether the CMA should itself take action under section 41(2) of the Act
for the purpose of remedying, mitigating or preventing the SLC concerned
or any adverse effect which may be expected to result from the SLC;

(b) whether the CMA should recommend the taking of action by others for the
purpose of remedying, mitigating or preventing the SLC concerned or any
adverse effect which may be expected to result from the SLC; and

(c) in either case, if action should be taken, what action should be taken and
what is to be remedied, mitigated or prevented.

11.9 The Act requires that the CMA, when considering possible remedial action,
‘shall, in particular, have regard to the need to achieve as comprehensive a
solution as is reasonable and practicable to the substantial lessening of
competition and any adverse effects resulting from it’. 1348

11.10 To fulfil this requirement, the CMA will seek remedies that are effective in
addressing the SLC and any resulting adverse effects. The effectiveness of a
remedy is assessed by reference to its: 1349

(a) impact on the SLC and the resulting adverse effects – the CMA views
competition as a dynamic process of rivalry between firms seeking to win
customers’ business over time – restoring the process of rivalry is a key
aim of a remedy;

(b) duration and timing – remedies need to be capable of timely

implementation and address the SLC effectively throughout its expected
duration;

(c) practicality in terms of implementation, monitoring and enforcement; and

1348 Section 36(3) of the Act.

1349 Merger remedies guidance (CMA87), December 2018, paragraph 3.5.

331
 (d) risk profile, relating in particular to the risk that the remedy will not achieve
its intended effect.

11.11 We note that Microsoft has made submissions regarding the appropriate legal
framework that the CMA should apply in assessing remedies. In particular,
Microsoft submitted that the Act does not require full prevention of any SLC
identified and instead recognises that mitigation may be sufficient. Microsoft
also submitted that the CMA need only ‘have regard’ to achieving as
comprehensive a solution to the SLC and any adverse effects arising as is
‘reasonable and practicable’, and that the CMA is therefore not bound by an
obligation to fulfil the requirement. 1350

11.12 As explained above, the Act requires the CMA to have regard to the need to
achieve ‘as comprehensive a solution as is reasonable and practicable’ to an
SLC and any adverse effects. In assessing this, the CMA first assesses
whether a remedy will be effective. The CMA considers that a remedy will only
be effective (ie a comprehensive solution) if it fully remedies or prevents the
SLC and its adverse effects (not just mitigates them). This approach, and the
‘high duty’ imposed on the CMA by the statute, 1351 has been endorsed by the
Courts. In particular, the Court of Appeal has explained that, once the CMA
has reached a conclusion on the SLC question, “then the action which it has
to take must be such as to remedy or prevent the SLC concerned. It is not at
that stage in the exercise concerned with weighing up probabilities against
possibilities but rather with deciding what will ensure that no SLC either
continues or occurs”. 1352 The Competition Appeal Tribunal (CAT) has also
found that it is reasonable for the CMA to not favour a remedy for which it
could not feel a “high degree of confidence of success”. 1353

11.13 As explained in the CMA’s Merger Remedies Guidance, having decided which
of the remedy options would be effective in addressing the SLC and resulting
adverse effects, the CMA will then consider the costs of those remedies. The
CMA may have regard, in accordance with the Act, 1354 to the effect of any
remedial action on any RCBs arising from the merger. In order to ensure that
any remedy is proportionate (ie reasonable), the CMA will seek to select the
least costly remedy, or package of remedies, of those remedy options that it
considers will be effective. 1355 The CMA will also seek to ensure that it does
not select a remedy that is disproportionate in relation to the SLC and its

1350 Microsoft response to the Notice of possible remedies (Remedies Notice), 22 February 2023, paragraph 5.2.
1351 Ecolab Inc. v CMA [2020] CAT 12, at [74].
1352 Ryanair Holdings PLC v CMA [2015] EWCA Civ 83, at [57]. See also Ecolab Inc. v CMA [2020] CAT 12, at

[74-75].
1353 Ecolab Inc. v CMA [2020] CAT 12, at [83].
1354 Section 36(4) of the Act.
1355 CMA87, paragraphs 3.4 and 3.6.

332
 adverse effects. 1356 In the event that the CMA considers fully remedying or
preventing an SLC and any adverse effects is not possible, or where no
effective remedy would be proportionate, the CMA will consider whether it is
nonetheless possible and proportionate to mitigate the SLC and its adverse
effects.

11.14 Accordingly, we agree with Microsoft that, in certain circumstances, it may be

appropriate for the CMA to take action which only mitigates an SLC and its
resulting adverse effects. However, we consider the appropriate first step in
our assessment is to determine what, if any, remedies are available which
would remedy or prevent the SLC and its resulting adverse effects, and then
to assess the proportionality of any such effective remedy.

Overview of remedy options

11.15 As set out in the CMA’s Merger Remedies Guidance, 1357 remedies are
conventionally classified as either structural or behavioural:

(a) Structural remedies, such as divestiture or prohibition, are generally one-

off measures that seek to restore or maintain the competitive structure of
the market by addressing the market participants and/or their shares of
the market.

(b) Behavioural remedies are normally ongoing measures that are designed
to regulate or constrain the behaviour of merger parties with the aim of
restoring or maintaining the process of rivalry absent the merger.

11.16 The choice of remedy will reflect the particular circumstances of each case,
though the CMA generally prefers structural remedies over behavioural
remedies, because: 1358

(a) structural remedies are more likely to deal with an SLC and its resulting
adverse effects directly and comprehensively at source by restoring
rivalry;

(b) behavioural remedies are less likely to have an effective impact on the
SLC and its resulting adverse effects, and are more likely to create
significant costly distortions in market outcomes; and

1356 CMA87, paragraph 3.6.

1357 CMA87, paragraph 3.34.
1358 CMA87, paragraphs 3.46 and 3.34.

333
 (c) structural remedies rarely require monitoring and enforcement once
implemented.

11.17 In the Remedies Notice, we set out the following structural remedy options:

(a) prohibition of the Merger; and

(b) requiring a partial divestiture of Activision Blizzard, Inc. We noted that this
could involve:

(i) Divestiture of the business associated with CoD (Call of Duty);

(ii) Divestiture of the Activision segment of Activision Blizzard, Inc. (the

Activision segment), which would include the business associated
with CoD; or

(iii) Divestiture of the Activision segment and the Blizzard segment (the
Blizzard segment) of Activision Blizzard, Inc., which would include
the business associated with CoD and WoW (World of Warcraft),
among other titles.

11.18 We also invited views on aspects of remedy design which might be needed to
make a divestiture remedy effective and ensure that no new competition
concerns would arise. These could include requirements relating to the scope
of any divestiture package, the process of selecting the assets to be divested,
the identification of suitable potential purchaser(s), and the divestiture process
including the timing of divestiture.

11.19 In the Remedies Notice, we explained that the CMA will generally only select
behavioural remedies as the primary source of remedial action where: 1359

(a) divestiture and/or prohibition is not feasible, or the relevant costs of any
feasible structural remedy far exceed the scale of the adverse effects of
the SLC;

(b) the SLC is expected to have a short duration; or

(c) behavioural measures will preserve substantial RCBs that would be

largely removed by structural remedies.

11.20 In the Remedies Notice, we set out a preliminary view that certain divestitures
and/or prohibition are, in principle, feasible remedies in this case. We noted
that the provisional SLC was not time-limited, and while RCBs had not yet

1359 CMA87, paragraphs 3.48, 7.1, and 7.2.

334
 been assessed in detail, evidence on efficiencies received up to that point did
not suggest that RCBs might be substantial.

11.21 We also noted in the Remedies Notice that, while none of the circumstances
in which the CMA would typically select a behavioural remedy as the primary
source of remedial action in a merger investigation appeared to be present,
we would nonetheless consider a behavioural access remedy as a possible
remedy.

The Parties’ responses to the Remedies Notice and Remedies Working Paper

11.22 In response to the Remedies Notice, Microsoft submitted its view that the
criteria for consideration of behavioural remedies were met in this case. It
submitted that the relevant costs of a structural remedy exceeded the scale of
the adverse effects of the provisional SLCs, and that RCBs were likely to be
substantial compared with any adverse effect of the Merger. It noted that the
claimed RCBs would be largely preserved by behavioural remedies but not by
structural remedies. 1360 Microsoft submitted that structural remedies – both
prohibition of the Merger and a partial divestiture of (part of) Activision – would
not be appropriate remedies and would result in lost benefits to customers
(the detail of these submissions is set out below). 1361 Instead, Microsoft
proposed that a behavioural remedy would be the most appropriate remedy to
the provisional SLCs identified, noting a number of RCBs which it submitted
would otherwise be lost as a result of a structural remedy. Microsoft proposed
two licensing remedies: (a) a content licensing remedy for console gaming
(the Console Remedy); 1362 and (b) a content licensing remedy for cloud
gaming (the Microsoft Cloud Remedy). 1363

11.23 As explained in this Report, we believe that an SLC may not be expected to
arise in the console gaming market in the UK as a result of the Merger.
Accordingly, we do not consider the proposed Console Remedy further in this
chapter. We consider that Microsoft’s representations on the relevant costs of
a structural remedy and RCBs remain relevant in the context of the SLC in
cloud gaming services in the UK and are considered further below. We
consider the proportionality of effective remedy options later in this chapter.

11.24 Activision did not provide a written response to the Remedies Notice or the
Remedies Working Paper, but echoed Microsoft’s view on structural remedies
and a preference for a behavioural remedy during its response hearing.

1360 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 1.5.
1361 Microsoft response to the Remedies Notice, 22 February 2023, paragraphs 5.1-5.23.
1362 Microsoft response to the Remedies Notice, 22 February 2023, paragraphs 3.1-3.12.
1363 Microsoft response to the Remedies Notice, 22 February 2023, paragraphs 4.1-4.10.

335
11.25 Microsoft provided a response to the Remedies Working Paper, as part of
which it made a number of amendments to the proposed Microsoft Cloud
Remedy with the aim of addressing the concerns set out in the Remedies
Working Paper. We summarise Microsoft’s views and our assessment of
those in the relevant sections below.

11.26 In following section, we discuss the effectiveness of different remedy options,

and conclude on those which we consider would represent an effective
remedy to the SLC and/or any of its adverse effects which we have identified.
In doing so, we consider the effectiveness of:

(a) Prohibition (see paragraphs 11.27 to 11.38);

(b) Partial divestiture (see paragraphs 11.39 to 11.42); and

(c) Behavioural remedies, in particular the Microsoft Cloud Remedy (see

paragraphs 11.43 to 11.132).

Effectiveness of prohibition

11.27 First, we consider the effectiveness of a prohibition remedy.

Description of remedy

11.28 This remedy option would involve prohibiting the Parties from completing the
Transaction. The Merger between Microsoft and Activision would thus not
take place and the current competitive dynamics in the market would not
change.

11.29 Prohibition would be effected by accepting undertakings under section 82 of

the Act or making an order under section 84 of the Act, prohibiting the Merger
and preventing the Parties from attempting to merge for a further period: our
normal practice would be to prevent a future merger between the Parties for
the next ten years, absent a change of circumstances.

Views of Parties and third parties

Views of the Parties

11.30 Microsoft submitted that this is a case in which prohibition of the Merger does
not preserve the market structure absent the Merger and prevent the SLC
and/or any of its adverse effects from arising in the way that might be
expected of a standard merger. It told us that Activision [] and that
Activision content is not available on any cloud gaming service today, and that

336
 the counterfactual finds that []. Microsoft submitted that based on the CMA
findings, prohibition would only result in Activision content becoming available,
at best, []. It told us that if prohibition could be considered an effective
remedy in that context, then the Microsoft Cloud Remedy must be equally
effective to remedy the SLC as it will result in Activision content becoming
available to a much larger range of cloud gaming providers, []. 1364

11.31 Microsoft submitted that weight should be placed on a third party’s ([]) view
that prohibition would be detrimental for the growth of cloud gaming (as noted
at paragraph 11.35 below). It told us that []. 1365

11.32 Further, Microsoft submitted that the ‘two key complainants’ against the deal
(SIE and Google) have many reasons to preserve the status quo or
undermine the Merger in order to extract higher profits to the detriment of
game developers and gamers. It submitted that, on this basis, the Microsoft
Cloud Remedy protects competition in a way that prohibition will not. Microsoft
submitted that ‘through prohibition, the CMA is ensuring the ongoing success
of entrenched competitors who have no incentive to expand competition’. 1366

Views of third parties

11.33 SIE told us that prohibition would be ‘clear-cut’ and ‘would safeguard against
the foreclosure strategies Microsoft could employ to withhold or degrade
access to Activision content’. 1367 In response to our remedies questionnaire,
an additional four third parties ([], [], [] and []) told us that they view
prohibition as an effective remedy. 1368

11.34 Another third party ([]) told us that prohibition of the Merger could be
effective in the case of a behavioural remedy not applying. However, it told us
that prohibition would not ensure the availability of games with respect to
cloud gaming in the same way as a behavioural remedy. 1369

11.35 One third party ([]) told us that prohibition would not be an effective remedy
because it disagreed with the SLC finding, and that blocking the Merger could
risk Activision entering into long-term exclusivity deals with Microsoft which
would have the same effect, in its view, as divestiture of the CoD
franchise. 1370 Another third party ([]) commented that prohibition would be

1364 Microsoft response to the Remedies Working Paper.

1365 Microsoft response to the Remedies Working Paper.
1366 Microsoft response to the Remedies Working Paper.
1367 SIE response to the Remedies Notice, 22 February 2023, paragraph 9.
1368 Third party responses to the remedies questionnaire: []; []; []; []; and [].
1369 [] call note.
1370 [] response to remedies questionnaire.

337
 ‘detrimental’ for the growth of cloud gaming and [] players as they would not
have access to the CoD title. 1371

Effectiveness of prohibition

11.36 Our Merger Remedies Guidance states that prohibition of an anticipated

merger is an effective remedy as it necessarily maintains the competitive
structure of a market that would have otherwise been changed by the
merger. 1372 This remedy option therefore preserves the market structure
absent the merger and prevents the SLC and/or any of its adverse effects
from arising.

11.37 In relation to Microsoft’s submission above that the Microsoft Cloud Remedy
would result in Activision content becoming available to a much larger range
of cloud gaming providers, when compared to prohibition, we consider that:

(a) in seeking to remedy the SLC, our starting point is to return the market to
the status quo ante, which prohibition would achieve – we are not seeking
to intervene to go beyond that requirement, which itself could have a
distortive impact on the market;

(b) in assessing the effectiveness of a remedy, we do not seek to recreate an

expected counterfactual outcome at a particular point in time but rather
seek to recreate the conditions of competition present in the
counterfactual. 1373 In this respect, Microsoft mischaracterises our findings
on the competitive conditions absent the Merger. We have found that the
cloud gaming market is nascent and dynamic and would continue to grow
absent the Merger, with providers testing a range of different business
models. We have also found that whilst Activision content has only been
available to cloud gaming platforms to limited extent so far, it would likely
become increasingly available within the next five years absent the
Merger (including day and date releases). As such, in assessing the
effectiveness of a remedy in addressing the SLC and its resulting adverse
effects, it is important to consider the impact of any remedy on preserving
that dynamic process in the market, taken as a whole; and

(c) Microsoft’s submission is predicated on the Microsoft Cloud Remedy

being effective in addressing the SLC we have identified – we consider
the effectiveness of the Microsoft Cloud Remedy later in this chapter.

1371 [] call note.

1372 CMA87, paragraph 3.35.
1373 See eg CMA87, paragraph 3.5.

338
11.38 In summary, if Microsoft and Activision are not brought under common
ownership or control, the competitive structure of the market (that would have
otherwise been changed by the Merger) would be maintained. As a
consequence, the vertical effects that we anticipate arising in the supply of
cloud gaming services in the UK would not arise. We therefore conclude that
prohibition would be an effective remedy which would comprehensively
address the SLC that we have identified, by preventing it and consequently
preventing any of its adverse effects from arising.

Effectiveness of partial divestiture

11.39 In the Remedies Notice, we set out a potential remedy option involving the
divestiture of the Activision business associated with CoD, or divestiture of
one or two of Activision’s operating segments. We investigated this option
further, consulting with the Parties and third parties.

Effectiveness of partial divestiture

11.40 Some third parties told us that a partial divestiture option had the potential to
be effective. 1374 Other third parties told us that partial divestiture would involve
effectiveness risks relating to scope and composition of the divestiture
package, availability of a suitable purchaser, shared intellectual property (IP)
assets and other development resources. 1375 One third party told us that it
could not see how partial divestiture could be better for competition and for
gamers than the Merger completing. 1376 SIE told us that a partial divestiture
remedy would be effective if the divested entity was able to compete viably,
on a standalone basis, and without the support of those parts of Activision's
business that remain with Microsoft (eg the King business). 1377

11.41 However, Microsoft told us that any partial divestiture scenario would be
[]. 1378 Activision acknowledged Microsoft’s position on partial divestiture,
noting its understanding that []. 1379

11.42 We are of the view that a partial divestiture could, in principle, represent an
effective remedy to the SLC we have found. However, a number of

1374 Third party response to the remedies questionnaire: []; []; and [].
1375 Third party response to the remedies questionnaire: []; and [].
1376 [] call note.
1377 SIE response to the Remedies Notice, paragraphs 10 and 11. See also third party call notes: []; []; [];

and [].
1378 Microsoft response hearing transcript.
1379 Activision response hearing transcript.

339
 composition and purchaser risks 1380 would be associated with such a
divestiture. These risks include identification of the assets and operations
necessary to effectively remedy the SLC; separation of assets and operations
common to the divested and retained parts of Activision; and the availability of
suitable purchasers for a divestiture package of this size. Given our initial risk
assessment, we consider that substantial additional evidence would be
required from the Parties to enable us to assess whether the divestiture of
part of Activision would, in practice, represent an effective remedy. We
explained to the Parties that, without additional evidence from them, we would
be unlikely to be able to conclude whether a partial divestiture is an effective
remedy. [], 1381 []. 1382 []. Accordingly, we do not assess the
effectiveness of any partial divestiture remedy further in circumstances where
[]. However, we have considered the possibility of this alternative remedy
option in the proportionality assessment, where relevant, below.

Effectiveness of behavioural remedies

Description of remedy

11.43 The Microsoft Cloud Remedy was initially submitted in Microsoft’s response to
the Remedies Notice on 22 February 2023. Further detail was provided during
Microsoft’s response hearing on 27 February 2023. Following this, Microsoft
presented a follow-up submission on 6 March 2023 which clarified some
points set out in the original version of the Microsoft Cloud Remedy and
revised other elements of the remedy. The Parties discussed the Microsoft
Cloud Remedy with the CMA staff team on 14 March 2023, with Microsoft
providing a response to the CMA’s questions on 15 March 2023, and a further
response to clarificatory questions on 17 March 2023. Microsoft provided a
summary of its Microsoft Cloud Remedy proposal on 22 March 2023. 1383

11.44 Microsoft subsequently made further amendments to the Microsoft Cloud

Remedy in its response to our Remedies Working Paper on 31 March 2023.
The CMA staff team discussed these amendments with the Parties on a call

1380 Composition risks are risks that the scope of the divestiture package may be too constrained or not
appropriately configured to attract a suitable purchaser or may not allow a purchaser to operate as an effective
competitor in the market. Purchaser risks are risks that a suitable purchaser is not available or that the merger
parties will dispose to a weak or otherwise inappropriate purchaser. See also CMA87, paragraph 5.3.
1381 Microsoft response hearing transcript; and

Activision response hearing transcript.

1382 We note that in its response to the Remedies Notice, Microsoft submitted its view that it []. It also noted

that it would extinguish RCBs and not be proportionate (Microsoft response to the Remedies Notice, 22 February
2023, section 5). We note these high-level comments, but take the view that the submissions from Microsoft are
not sufficient in themselves to allow for a detailed assessment of the effectiveness of a partial divestiture. We
would need significant additional evidence, for example on the practicalities of splitting up the Activision business.
1383 Microsoft response to the Remedies Working Paper.

340
 on 4 April 2023, and Microsoft provided responses to our follow-up questions
on 6 April 2023. In relation to Microsoft’s proposed remedy commitments to
the European Commission, on [] April 2023, Microsoft told us that it [],
and that []. On [] April 2023, the Parties []. We have reviewed
Microsoft’s amendments and consider that these extra elements (if they were
to be incorporated into the Microsoft Cloud Remedy to address the SLC we
have identified) do not change our assessment of the Microsoft Cloud
Remedy set out in this chapter.

The Microsoft Cloud Remedy

11.45 Under the Microsoft Cloud Remedy, Microsoft would commit to license
Activision games, including CoD and WoW, royalty-free to certain cloud
gaming providers with a B2P or BYOG offering for a period of ten years.
Microsoft has proposed that the Microsoft Cloud Remedy would incorporate
the following features:

(a) Scope: The remedy will apply to all current and future PC and console
franchises, titles in these PC and console franchises, and any other PC
and console games that: (i) have been developed in the past or will be
developed in the future, either in part or in full, by any of the ‘Activision
Blizzard Studios’ (defined in the footnote); 1384 or (ii) are based, either in
part or in full, on IP rights of any PC or console franchises, titles in these
PC or console franchises, and any other PC or console games that
Activision Blizzard Studios have developed in the past or will develop in
the future (the Eligible Games). 1385 Microsoft submitted that PC games
do not include only Windows OS games, but any games published by the
Activision Blizzard Studios for any PC OS and that it will not seek to
frustrate the use of compatibility layers to play Activision games on non-
Windows OS. 1386 In particular, Microsoft clarified that it was prepared to
commit that the [], [] and other PC OS versions of the Eligible Games
will be included in the Microsoft Cloud Remedy. 1387 We understand that
this remedy will apply globally, and not just to consumers in the UK. 1388

Microsoft defines Activision Blizzard Studios as: []. Source: Microsoft response to the Remedies Working
1384
Paper.
1385 Microsoft response to the Remedies Working Paper.
1386 Microsoft response to the Remedies Working Paper.
1387 Microsoft response to the Remedies Working Paper.

1388 As discussed with the Parties’ advisers on the call with the CMA staff team on [].

341
 (b) Term: the remedy will apply for a period of ten years from the
commencement date of the final undertakings. 1389,1390

(c) Consumer Licence: Microsoft will unilaterally grant a licence to any

consumer who has purchased or obtained a free licence to play a PC
game from an authorised third-party PC digital storefront (Consumer
Licence). 1391 Microsoft submitted that consumers will have the right to
play Eligible Games on Eligible Streaming Services (defined below)
regardless of whether the Consumer Licence was obtained prior to, or
after, the Merger. It submitted that the updated Consumer Licences will be
published on Microsoft’s website, and that the streaming right is therefore
‘portable’ in that it can be used to access a game which the consumer has
acquired on any Eligible Streaming Service. 1392 Microsoft submitted that
the Consumer Licence will be perpetual, meaning that consumers will
have a perpetual right to stream the Eligible Games for so long as such
games are available. 1393

(d) Streaming Provider Licence: Microsoft will grant a royalty-free licence to

Eligible Streaming Services (defined below) to stream Eligible Games
(Streaming Provider Licence). The Streaming Provider Licences will be
granted []. Streaming Provider Licences will also be made perpetual to
match the Consumer Licence. 1394

(e) Eligible Streaming Service: A cloud game streaming service that allows
consumers to play from the service provider’s cloud-based servers, PC or
console games for which the consumers have already obtained a licence
for download of the game (including through either B2P, free-to-play, or
subscription), which either: (i) is permitted by an Authorised Game Store
(defined below) to provide access to Eligible Games; or (ii) offers access
to Eligible Games through applications which do not require integration
with an Authorised Game Store (Eligible Streaming Service). 1395

(f) Authorised Game Stores: Authorised Game Stores will be: []. 1396
Microsoft also offered to authorise [] PC storefronts to distribute the PC

1389 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 4.5(b).
1390 Microsoft response to the Remedies Working Paper.
1391 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 4.5(c).
1392 Microsoft supplemental response to the Remedies Notice, 6 March 2023, paragraph 3.6.
1393 Microsoft response to the Remedies Working Paper.
1394 Microsoft response to the Remedies Working Paper.
1395 Microsoft response to the Remedies Working Paper.
1396 Microsoft response to the Remedies Working Paper.

342
 versions of Eligible Games available on Battle.net []. This solution
would involve []. 1397

(g) Access to the [] 1398 []. Microsoft submitted that the CMA will have the
ability to amend and/or expand the definition of Major Game Publisher
during the term of the remedy, following a report of the Monitoring Trustee
setting out a comparison of the Major Game Publishers as of the
‘Commencement Date’ and a list of other publishers that the Monitoring
Trustee considers suitable to be included in the definition of Major Game
Publisher. It noted that Microsoft would have the opportunity to comment
on the Monitoring Trustee’s report before the CMA takes a decision. 1399

(h) Pricing: The Consumer Licence and the Streaming Provider Licence will
be []. 1400

(i) Parity: For new releases, the Streaming Provider Licence covers Eligible
Games, including publicly available beta versions and early access
releases[]. 1401

(j) Terms: Licences granted under the Streaming Provider Licence

commitment will be available on Microsoft’s website and subject to the
following terms:

(i) Unless otherwise agreed with Authorised Game Stores, []. 1402 In
response to the CMA’s question on how this provision would operate
in relation to a console game sold, eg by SIE through the PlayStation
Store or a MGS, Microsoft clarified that []. 1403

(ii) Eligible Streaming Services will be responsible for: (a) securing any
third-party public performance or similar licences that are not owned
by Microsoft to the extent necessary to support the Eligible Streaming
Service; and (b) compliance with relevant laws, including consumer,
data protection and online safety legislation, and privacy
standards. 1404

1397 Microsoft submission to the CMA.

1398 Microsoft defines the Major Game Publishers as Tencent, Valve Corporation, Nexon, NetEase, EA,
SMileGate, Embracer (THQ) – Perfect World, Roblox Corporation, Take Two, Epic Games, and Ubisoft.
1399 Microsoft response to the Remedies Working Paper.
1400 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 4.5(e).
1401 Microsoft response to the Remedies Working Paper.
1402 Microsoft response to the Remedies Working Paper.
1403 Microsoft submission to the CMA.
1404 Microsoft response to the Remedies Working Paper.

343
 (iii) Microsoft will not invoke the existence of third-party IP rights, if any,
as a reason to revoke the Streaming Provider Licence. 1405

11.46 Microsoft submitted that the Microsoft Cloud Remedy would include the
following mechanisms for monitoring compliance and dispute resolution:

(a) Monitoring Trustee: a Monitoring Trustee will be appointed to, primarily: (i)
monitor and seek to ensure Microsoft’s compliance with the remedy and
to report to the CMA on a periodic basis; (ii) under the fast-track dispute
resolution procedure, present proposals for resolving disputes before they
are referred to binding arbitration; and (iii) issue opinions in the event that
Microsoft wishes to terminate a Streaming Provider Licence. 1406 The
identity of the Monitoring Trustee will be agreed with the CMA and
remunerated by Microsoft. The Monitoring Trustee may appoint advisers
as it and the CMA consider reasonably necessary for the performance of
the Monitoring Trustee’s functions, with any such advisers remunerated
by Microsoft. 1407

(b) Dispute Resolution Mechanism: A fast-track dispute resolution carried out

under the International Chamber of Commerce Rules of Arbitration. The
identity of the Arbitral Tribunal would be agreed between and
remunerated by the ‘Requesting Party’ and Microsoft. 1408 The CMA will be
enabled to participate in any arbitration process, and shall retain the
power to direct changes to the dispute resolution process where
necessary to ensure the effective operation of the remedy. 1409

11.47 Microsoft will appoint [] to be responsible for monitoring and certifying the
company’s ongoing compliance with the undertakings (the Compliance
Director) under the Microsoft Cloud Remedy. 1410 The Compliance Director will
be responsible for preparing an annual report certifying the company’s
compliance with the undertakings and will be responsible for monitoring
compliance with the terms of the undertakings, facilitating requests for
information, identifying and rectifying any instances of non-compliance, and
maintaining staff awareness of the requirements of the undertakings. 1411

1405 Microsoft response to the Remedies Working Paper.

1406 Microsoft response to the Remedies Working Paper.
1407 Microsoft, Annex 1 to the response to the Remedies Working Paper.
1408 Microsoft, Annex 1 to the response to the Remedies Working Paper.
1409 Microsoft, Annex 1 to the response to the Remedies Working Paper.
1410 Microsoft, Annex 1 to the response to the Remedies Working Paper; and Microsoft response to the

Remedies Working Paper.

1411 Microsoft, Annex 1 to the response to the Remedies Working Paper.

344
Views of the Parties and third parties

Views of the Parties – the Microsoft Cloud Remedy

11.48 Microsoft submitted that the Microsoft Cloud Remedy would address the
CMA’s concerns in relation to the SLC. In particular, Microsoft submitted that:

(a) The Eligible Games will not be exclusive to Xbox Cloud Gaming: Microsoft
submitted that this goes beyond what would occur if [] (as noted in the
assessment of the counterfactual in the Provisional Findings). Microsoft
submitted that the Microsoft Cloud Remedy goes ‘significantly beyond’
this scenario by enabling any Eligible Provider to stream the Eligible
Games, []. 1412

(b) The Eligible Games will not be timed exclusive to Xbox Cloud Gaming:
Microsoft submitted that the Eligible Games will be made available to
other cloud gaming providers at the same time as they are released for
sale on PC and []. 1413

11.49 Microsoft submitted that the Microsoft Cloud Remedy is a clear and self-
executing solution. It told us that it is self-executing because it authorises
cloud gaming providers to stream the game from their services, and
consumers to play the game on their devices. Further, Microsoft submitted
that because of the remedy’s simplicity, it is agnostic to how cloud gaming
providers may develop and innovate in the future. 1414 Microsoft submitted that
it considers the openness of the remedy to be a key strength of its proposal
and submitted that by granting a non-exclusive royalty-free, perpetual licence
to any Eligible Streaming Service, it is comprehensively addressing the SLC
in a manner which minimises implementation risks, particularly as there are
no payments from the Eligible Streaming Service to Microsoft. It told us that
the Microsoft Cloud Remedy is based on existing commercial practices. 1415

11.50 In its response to our Remedies Working Paper, Microsoft told us that the
assessment of the SLC which underpinned the Remedies Working Paper was
flawed and that the Remedies Working Paper continued to work on the basis
of various assertions or assumptions which were wrong. In particular,
Microsoft told us that it used a counterfactual that Activision would have
placed its games on cloud gaming services absent the Merger, which
Microsoft told us was wrong and contrary to the available evidence; and that it
ignored the agreements Microsoft had reached with NVIDIA, Boosteroid and

1412 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 4.8(a).
1413 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 4.8(b).
1414 Microsoft response to the Remedies Working Paper.
1415 Microsoft response to the Remedies Working Paper.

345
 Ubitus under which the distribution of Activision content on multiple cloud
gaming services was provided for, should the Merger proceed. Microsoft told
us that these errors meant that the SLC finding was unsustainable. 1416 Our
consideration of, and conclusions on, Microsoft’s arguments in relation to the
counterfactual and Microsoft’s post-Merger agreements are covered in
Chapter 8. 1417 We also consider these agreements further in Chapter 9 and
when we consider RCBs later in this chapter.

11.51 Microsoft told us that, in its view, the cloud gaming market is de minimis and
will remain so and therefore any SLC will be limited. On this basis, Microsoft
submitted that the Microsoft Cloud Remedy is proportionate to the size of the
SLC while also preserving and enhancing RCBs. 1418 We set out more detail
on the Parties’ views on, and our assessment of, RCBs and proportionality
from paragraph 11.135 below.

11.52 Microsoft submitted that accepting the Microsoft Cloud Remedy would signal
the CMA’s ability to design and implement remedies that recognise the
specific features of the digital economy and afford it the ability to foster
emerging market segments like cloud gaming. It submitted that the UK
Government had been clear that the new merger control rules overseen by
the CMA’s Digital Markets Unit (DMU) must avoid disproportionate burdens
on businesses, and that rejecting the Microsoft Cloud Remedy would call the
practicability of the new regime into question, harm consumers and have a
negative impact on innovation and investment in the UK. 1419

11.53 Microsoft submitted that the ten-year timeframe is driven by the standard
period for remedies that the European Commission has adopted. 1420 It
submitted that it is not aware of any streaming agreements in the market with
a longer duration and that the period is sufficiently long for cloud gaming to
establish itself as a consumer service and for providers to secure a range of
popular games. 1421 Microsoft submitted that the Microsoft Cloud Remedy will
immediately address the competitive concerns and that a period of ten years
allows sufficient time for the market to mature and for alternative sources of
content to emerge. Further, Microsoft submitted that a ten-year timeframe is in
line with CMA guidance, noting some previous CMA cases. Microsoft referred

1416 Microsoft response to the Remedies Working Paper.

1417 Chapter 8, paragraphs 8.278 to 8.280; and 8.332 to 8.337.
1418 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 4.10(a-b).
1419 Microsoft response to the Remedies Working Paper.
1420 Microsoft response hearing transcript.
1421 Microsoft supplemental response to the Remedies Notice, 6 March 2023, paragraph 3.3.

346
 to the perpetual nature of the Consumer Licence, which would continue for so
long as games are available. 1422

11.54 In relation to the design of the Microsoft Cloud Remedy, we set out below
Microsoft’s views on how it addresses the four main risks normally associated
with behavioural remedies (as set out in our Merger Remedies Guidance),
namely: specification, circumvention, distortion and monitoring and
enforcement risks 1423 (see footnote for details and also paragraph 11.90
below).

11.55 On the specification of the remedy, Microsoft submitted that the structure of
the Microsoft Cloud Remedy is based on industry practice, commonly used by
publishers, whereby licences are granted to each consumer and cloud gaming
streaming provider on PC, referring to its recent agreement with NVIDIA 1424
as an example. On this basis, it submitted that the remedy does not require
complex implementation or technical adaptation, or any significant monitoring
or supervision. 1425

11.56 On circumvention, Microsoft submitted that it currently distributes its PC

games through a number of third-party PC digital storefronts and will commit
to distribute Eligible Games through at least one third-party PC digital
storefront throughout the term of the remedy. Microsoft submitted that this
eliminates any risk of circumvention of the remedy by ensuring that Microsoft
distributes Eligible Games outside its owned and operated PC digital
storefronts. Microsoft also subsequently offered to []. Further, Microsoft
submitted that it commits to make the games available []. 1426 On
circumvention via third-party IP rights, Microsoft submitted that it is prepared
to commit not to invoke third-party IP rights as a reason for seeking to
invalidate or revoke a Streaming Provider Licence to manage the risk of
Microsoft including IP in its games which could undermine the purpose of the
remedy. It told us that the monitoring provisions put in place will assist in
managing this risk. Further, Microsoft submitted that the list of titles within
Eligible Games cannot be used to circumvent the remedy. It told us that the
remedy is broad in scope and captures all current and future titles developed
in part or full by the Activision Blizzard Studios or using the IP in the games

1422 Microsoft response to the Remedies Working Paper.

1423 The four main risks associated with behavioural remedies as set out in our guidance are: (a) specification
risks, which arise if the form of conduct required to address the SLC or its adverse effects cannot be specified
with sufficient clarity to provide an effective basis for monitoring and compliance; (b) circumvention risks, which
arise if other adverse forms of behaviour arise when particular forms of behaviour are restricted; (c) distortion
risks, which arise where behavioural remedies create market distortions that reduce the effectiveness of these
measures and/or increase their effective costs; and (d) monitoring and enforcement risks, which arise if there are
risks of ineffective monitoring and enforcement. Source: CMA87, paragraph 7.4.
1424 Microsoft, Annex 2 to the response to the Remedies Notice, 22 February 2023.
1425 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 4.9(a).
1426 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 4.9(b).

347
 they created and that it has a consistent track record of operating the studios
it acquires as separate entities. 1427 It submitted that an anti-circumvention
measure – ie involving the Monitoring Trustee in overseeing the Eligible
Games – will ensure that all Eligible Games that have been developed in the
past or would have been developed in the future by Activision Blizzard
Studios absent the Merger would be available. 1428

11.57 Further, Microsoft submitted that it is prepared to commit to not implement

any features or functions in the Eligible Games after the undertakings
commencement date with the deliberate purpose of preventing or negatively
affecting the operation or performance of any compatibility layer or of the
Eligible Games running on a compatibility layer. However, Microsoft noted
that the suggestion that a remedy should ensure future versions of Eligible
Games would be able to run on other OSs through compatibility layers is not a
reasonable requirement as Microsoft cannot ensure that a game operates
through a compatibility layer as compatibility layers are software interfaces
developed by third parties. It noted that if Windows develops a new feature,
the developer of the compatibility layer needs to update its own software,
which Microsoft cannot be responsible for. Microsoft noted that Activision
does not officially support compatibility layers for this reason (ie Activision
cannot guarantee that its games will run on a compatibility layer developed
and maintained by a third party). Further, Microsoft noted the complexities
that can arise from anti-cheat software meaning that some game titles cannot
run on certain open-source OS and compatibility layers because of security
concerns. 1429

11.58 In its response to our Remedies Working Paper, in relation to the CMA’s
broader concern that, under the Microsoft Cloud Remedy, there may be a
variety of ways in which user experience might worsen on competing
platforms even with some form of licensing remedy in place, Microsoft told us
that there was no basis for this concern as the titles which would be available
for streaming would be the same which are available to purchase from the
Microsoft Game Store or another Authorised Game Store on a royalty-free
basis. Moreover, it told us that Microsoft was offering streaming rights in
relation to titles irrespective of whether Microsoft itself decides to stream the
titles. It added that the CMA relied on evidence from one third party ([]) that
had consistently refused to hold any discussions with Microsoft in order to
address its concerns, and noted that the concerns raised have been rejected

1427 Microsoft submitted that none of the studios acquired by Microsoft since 2010 have been absorbed by
another Microsoft studio. It told us that one studio was later sold to the original team (Twisted Pixel Games) and
another studio (Press Play) was closed, and the remainder continue to be distinct entities.
1428 Microsoft response to the Remedies Working Paper.
1429 Microsoft response to the Remedies Working Paper.

348
 by the CMA in relation to console gaming. It told us that the Provisional
Findings also presented no evidence to suggest that partial foreclosure of
Activision content could foreclose cloud gaming rivals. 1430

11.59 Microsoft submitted that the Microsoft Cloud Remedy does not give rise to
any market distortion because it will remain in place for a targeted period and
is based on existing market-based terms commonly used by publishers.
Microsoft submitted that BYOG is the most successful business model on
cloud gaming and has the lowest barriers to entry, and that its remedy
addresses the concerns raised in relation to this model. It submitted that the
Microsoft Cloud Remedy also addresses the B2P business model to avoid
distortion of the market – ie to avoid favouring only the BYOG model and
potentially contributing to its growth by not capturing B2P. Further, Microsoft
submitted that cloud providers do not require an extensive catalogue of
content to launch, and many offer a range of different payment options. 1431

11.60 Microsoft submitted that the Microsoft Cloud Remedy is not limited in scope to
BYOG and B2P. It told us that the definition of Streaming Service includes all
business models, including MGS. It submitted that it is sufficient that MGS is
treated in the same way as other potential business models and submitted its
view that we have presented no evidence to suggest that Activision would
have contemplated placing its content on a MGS cloud gaming service. It
submitted that we have not found Activision content to be an important input
for MGS. Microsoft submitted that the evidence is limited to CoD and that
WoW is only ‘notable’. It told us that the fact that future games may have been
placed on cloud services in the future does not mean that they would be an
important input. Microsoft submitted that the Provisional Findings are
consistent with MGS growing without access to Activision titles and that the
CMA ruled out the concept of Activision content playing a role in the growth of
cloud gaming MGS. On this basis, Microsoft submitted that this cannot be
seen as a distortion. Further, Microsoft submitted that it would be
disproportionate and unnecessary for the scope of the remedy to include a
commitment to make Eligible Games available via MGS in order to address
the SLC identified by the CMA. 1432

11.61 In terms of monitoring and enforcement, Microsoft submitted that the

Microsoft Cloud Remedy is ‘self-executing’ and its implementation does not
require significant monitoring or supervision by the CMA. 1433 However, it
noted that it has included specific mechanisms to allow for monitoring and

1430 Microsoft response to the Remedies Working Paper.

1431 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 4.9(c)
1432 Microsoft response to the Remedies Working Paper.
1433 Microsoft response to the Remedies Working Paper.

349
 dispute resolution. Microsoft submitted that the Eligible Streaming Services
would have all of the information they need to: (a) request a licence; and (b)
detect potential breaches of Microsoft’s obligations and bring a dispute under
the disputes resolution procedure. Further, Microsoft submitted that it will
provide the Monitoring Trustee with full and complete access to the
information required to certify compliance with the remedy, 1434 and that
monitoring and enforcement costs should be modest due to the simplicity of
the remedy and its transparency to customers. 1435

11.62 Activision told us that it thought the Microsoft Cloud Remedy would be
effective. It noted its view that the bar to qualify as an Eligible Provider is low
– Activision said it was simple and easy to monitor. It told us that the remedy
‘gives a lot of flexibility, but it also ensures that there is an enormous amount
of competition’. 1436

Views of third parties – the Microsoft Cloud Remedy

11.63 In relation to the effectiveness of a behavioural remedy, the evidence from

third parties was mixed, with some telling us that a behavioural remedy for
cloud gaming could be an effective remedy, 1437 and others submitting that it
would not be effective.

11.64 A third party ([]) told us that the behavioural commitments in the Microsoft
Cloud Remedy would result in ‘positive outcomes to the market and that the
end-users who will have more choice as to where they play the games’. The
same third party also told us that it thinks a behavioural remedy in cloud will
also create a number of additional benefits for the industry and end-customers
that would not otherwise have existed – it told us that Activision content will
become widely available across the cloud, which means end-customers will
have access to this content on almost any device, platform and OS via cloud
gaming services. 1438 Another third party ([]) referred to the agreement that it
has entered into with Microsoft (which it considers to be very similar to the
Microsoft Cloud Remedy) and told us that it resolves its concerns around the
SLC. 1439

11.65 One third party ([]) told us that a behavioural remedy would need to
overcome ‘effectiveness challenges’, as Microsoft could use tools for full or

1434 Microsoft response to the Remedies Working Paper.

1435 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 4.9(d)
1436 Activision response hearing transcript.
1437 For example, [] ([] call note). Third party responses to the remedies questionnaire: ([]; []; []; [];

and []. See also third party call notes: []; []; and [].
1438 [] response to the remedies questionnaire.
1439 [] call note.

350
 partial foreclosure, which may vary with market developments and changes in
consumer preferences. This third party told us that the user experience for
Activision Blizzard content could be degraded by: 1440

(a) Microsoft limiting the availability of Activision Blizzard content on consoles

and cloud gaming services platforms (particularly for new types of
consoles and cloud gaming services platforms);

(b) content exclusivity by offering lower quality options (such as not having
access to features, upgrades, expansion packs, bonus content, or in-
game content such as skins or weapons) for Activision Blizzard content on
other consoles and cloud gaming service platforms;

(c) timed exclusivity by delaying new versions, updates, content packages, or

in-game content for Activision Blizzard titles from appearing on other
consoles and cloud gaming platforms at the same time as on Microsoft
devices;

(d) experience degradation on rival platforms by delaying and failing to

complete bug testing and remediation; having lower quality customer
responses; degrading key gaming qualities such as high-end graphics and
auto-saving reliability; increasing gameplay latency or deprioritising
latency issues; ignoring or deprioritising form factor and other user
experience changes needed to optimise game content; and impeding
gameplay matchmaking services; and

(e) discriminatory pricing and terms by raising prices for Activision Blizzard
content on other console or cloud gaming platforms or otherwise applying
less favourable terms and conditions to users accessing that content on
non-Microsoft platforms.

11.66 Another third party ([]) made a submission setting out its views on the latest
iteration of the Microsoft Cloud Remedy (of 3 April 2023, based on Microsoft’s
proposed commitments to the European Commission, which largely track
Microsoft’s modified Microsoft Cloud Remedy proposal made to the CMA in its
response to our Remedies Working Paper with subsequent amendments
made to it since then). 1441 This third party told us that the modified Microsoft
Cloud Remedy was designed to remedy shortcomings in the original Microsoft
Cloud Remedy, but that these changes were limited to ‘correcting’ prior

[]. [] response to the remedies questionnaire.

1440

Microsoft told us that the revised details of the Microsoft Cloud Remedy set out in its response to the
1441

Remedies Working Paper were consistent with the revised remedy proposal which had been put forward to the
European Commission, but that Microsoft’s proposal to the CMA went beyond the proposed commitments to the
European Commission in certain respects in order to address the CMA’s specific concerns set out in the
Remedies Working Paper. Source: Microsoft response to the Remedies Working Paper, paragraph 1.4.

351
 oversights to allow gamers that obtained a licence to Activision games to use
a cloud game streaming service in order to play those games on a console as
well as a PC. This third party told us that the revised Microsoft Cloud Remedy
was ‘inadequate and insufficient’ for the following reasons: 1442

(a) the modified Microsoft Cloud Remedy does not contain an obligation on
Microsoft to license the Activision games – in order to remedy the concern
that Microsoft has an incentive to withhold Activision content from
distributors of games via cloud game streaming services, Microsoft should
be required to ensure that all such distributors of games, including stores
that sell games to be streamed via the cloud, such as PlayStation Plus,
Amazon Luna, and (before it closed) Stadia, have access to such content
on fair, reasonable, and non- discriminatory terms;

(b) the modified Microsoft Cloud Remedy does not require Microsoft to
license Activision content to distributors of games via cloud gaming
streaming services – this third party told us that Microsoft’s obligations
arise only where a consumer has secured a licence to play Activision
games and wants to stream that game on a third-party cloud streaming
service. It added that operators that do not secure a licence to Activision
games would be foreclosed. This third party told us that, accordingly,
should Microsoft decide not to license Activision games, the only cloud
streaming services that would benefit from the modified Microsoft Cloud
Remedy would be BYOG services, such as NVIDIA, which rely on
consumers buying games elsewhere. It added that even then, the way in
which the modified Microsoft Cloud Remedy had been structured, make it
highly unlikely that even BYOG operators would benefit;

(c) the modified Microsoft Cloud Remedy does not remedy other ‘significant
defects’ in the original Microsoft Cloud Remedy, in particular:

(i) first, the modified Microsoft Cloud Remedy would entitle Microsoft to
retain all revenue from sales of Activision games, in-app purchases,
and any other future game-related transactional revenues – as a
result, cloud game streaming services would not become rivals to
Microsoft’s Game Pass Ultimate, but rather, they would become
Microsoft’s customers, with only the ability to stream Activision games
and send all the associated revenues to Microsoft. This third party
told us that this would be unattractive for any cloud game streaming
service, including BYOG operators like NVIDIA, which are the only

1442 [] submission to the CMA.

352
 operators that would benefit from the modified Microsoft Cloud
Remedy should Microsoft decide not to license Activision games;

(ii) second, the modified Microsoft Cloud Remedy does not include any
quality, content, or technical parity obligations, and therefore,
Microsoft could release Activision content to rival cloud streaming
service at a lower level of quality or with features or upgrades only
available to its own service;

(iii) third, the duration of the modified Microsoft Cloud Remedy is too
short – given the nascency of cloud game streaming, a ten-year
duration is insufficient to enable alternative cloud game streaming
services to establish themselves as credible competitors to Microsoft
in this nascent space; and

(iv) fourth, the modified Microsoft Cloud Remedy still does not allow for
swift redress. In this regard, this third party told us that the ‘Fast-Track
Dispute Resolution’ mechanism is long and burdensome and would,
at best, enable redress around one year after a licensee had become
aware of a potential breach. It added that in the meantime, the
licensee would have suffered irreparable harm, ‘as the market would
tip further to Microsoft’. Therefore, this third party considered that
swift and effective redress was essential for any access commitment
of this kind.

11.67 This third party ([]) also told us that the modified Microsoft Cloud Remedy
would ensure that Microsoft was the only cloud game streaming service able
to distribute irreplaceable Activision content to cloud gaming streaming
services. It told us that Microsoft could choose not to license Activision games
to cloud rivals or to license degraded versions without sanction or constraint.
Further, this third party told us that Microsoft’s rivals would be required to give
Microsoft all revenues generated from Activision games. Therefore, this third
party considered that the modified Microsoft Cloud Remedy would not remedy
the concerns it is intended to address, but instead, it would strengthen
Microsoft. 1443

11.68 Other third parties told us that they did not think the Microsoft Cloud Remedy
would be an effective remedy to the SLC. SIE and one other third party ([])
told us that behavioural remedies are not suited to this case. 1444 In particular,
SIE told us that the gaming industry is a ‘dynamic and evolving market’,

1443[] submission to the CMA.

1444[] response to the remedies questionnaire; and SIE response to the Remedies Notice, 22 February 2023,
paragraph 5.

353
 particularly in MGS services and cloud gaming, and any access commitment
that Microsoft might make to competitors is unlikely to capture these
developments with sufficient specificity to ensure competition in the future. 1445

11.69 On the timing and duration of the remedy, a number of third parties told us
that a ten-year timeframe would be appropriate for the Microsoft Cloud
Remedy. 1446 For example, one third party ([]) told us that []. 1447 However,
another third party ([]) told us that it is unclear how and when cloud gaming
services will evolve and that this uncertainty means a significantly longer
duration of a remedy needs to be factored in. 1448

11.70 On specification risks, two third parties ([] and []) told us that the remedy
package should consist of all Activision Blizzard content, including future
releases during the timeframe of the remedy. 1449 [] noted that [], which it
believes is a strong additional commitment from Microsoft. One third party
([]) told us that it was satisfied with the scope of its own agreement with
Microsoft, which covers []. 1450 We note, however, [] (and therefore that
this third party’s views are of more limited relevance to the assessment of the
scope of the Microsoft Cloud Remedy).

11.71 Another third party ([]) told us that specification risks would arise in this
case given that cloud gaming’s development was at a ‘nascent stage’. This
third party told us that it was difficult to predict the future of cloud gaming,
including because cloud gaming service providers were testing a range of
different business models to monetise their service with no fixed business
models yet, and therefore, Microsoft’s Microsoft Cloud Remedy was unlikely
to capture these developments with sufficient specificity to ensure competition
in the future. The same third party also told us that cloud gaming required a
technically complex infrastructure whose requirements could not be specified
with enough certainty to construct a remedy. It added that allowing games to
be accessed on remote servers and streamed on various devices at a
satisfactory quality would pose challenges including the construction of data
centres, and system requirements such as bandwidth and latency. It told us
that these challenges were ‘inflated for highly reactive, fast-paced shooter
games’ like CoD, making it ‘nearly impossible’ to define those requirements
precisely. 1451

1445 SIE response to the Remedies Notice, 22 February 2023, paragraph 16.
1446 Third-party responses to the remedies questionnaire: []; []; and [].
1447 [] call note.
1448 [] response to the remedies questionnaire.
1449 Third-party response to the remedies questionnaire: []; and [].
1450 [] call note.
1451 [] submission to the CMA.

354
11.72 Third parties also commented on Microsoft’s proposal to make available the
PC version of the game (running on Windows OS). Some third parties noted
that this may not be sufficient to capture future developments in the market.
For example, one third party ([]) noted that while PC hardware is generally
used to stream cloud games now, it is not certain that this will always be the
case; therefore, it may be the case that different versions of games (eg other
than PC versions) become more important in future. 1452 Similarly, another
third party ([]) told us that it does not think the Windows version of
Activision games would be appropriate to remedy the SLC as it forecloses
other business models by forcing cloud gaming providers to use Windows OS
and making gaming providers reliant on Windows Virtual Machines. This third
party ([]) told us that a potential solution could be Microsoft ensuring that it
will provide equivalency for game play features and experience through a
Windows version of the game that is compatible with Proton (a compatibility
layer which allows Windows games to run on non-Windows (eg Linux) OSs)
providing support for their open-source review. However, this third party ([])
noted that this also raises a number of parity concerns. 1453 Another third party
([]) told us that it []. 1454

11.73 With regard to circumvention risk:

(a) Some of the third parties we spoke to told us that behavioural remedies
would need to be appropriately specified so as not to create
circumvention risks. 1455

(b) One third party ([]) told us that it did not think that it would be likely that
Microsoft would want to circumvent the remedy. It told us that Microsoft
has made a public commitment to this, and that it was ‘highly unlikely that
they would turn their back on the player community’. 1456

(c) Another third party ([]) told us that Microsoft could foreclose access to
Activision content in multiple ways, including partial foreclosure of rival
cloud gaming services by degrading the quality of games on those
services, or simply not releasing the games on PC, circumventing any
obligation under the Microsoft Cloud Remedy to provide the game to rival
cloud gaming services. This third party told us that it would be ‘nearly
impossible to tailor a remedy that would address each option available to

1452 []response to remedies questionnaire.

1453 [] call note.
1454 [] call note.
1455 Third party response to the remedies questionnaire: []; []; and [].
1456 [] call note.

355
 Microsoft’, and added that Microsoft’s Microsoft Cloud Remedy did not
provide any ‘technical or content parity’ for cloud gaming. 1457

(d) Another third party ([]) told us that to ensure that under a behavioural
remedy, third parties received the same quality of game as that used by
Microsoft, the merging parties should ensure that built-in functionality in
the game is not tied to specific Microsoft products (such as Azure cloud
functionality) that are not publicly available. 1458

11.74 With regard to distortion risks, in response to our question of how a

behavioural remedy could remain relevant and appropriate as MGS and
cloud-gaming services continue to develop (eg ensuring that new entrants
would be captured by the behavioural remedy), one third party ([]) told us
that the remedy could apply to an existing or potential competitor in a broadly
defined market, allowing new entrants (including those with alternative
business models) to take advantage of the remedy. It told us that content can
be broadly defined, ensuring competitors have access to that content, and
that access rights can apply before public release to allow the third parties
time to perform quality assurance and launch on equal footing as the merging
parties. The same third party told us that the remedy could apply to all current
or future products, and that it could require periodic checkpoints to allow
adjustments to fit new market dynamics. 1459

11.75 Another third party ([]) told us that there was a risk that information sharing
that would be required between Microsoft and any potential new entrant could
distort the market. It told us that the new entrant would need to either disclose
to Microsoft the details of its console or cloud game streaming service or
would have to launch its service without CoD. It told us that neither option
would be attractive as it could damage the new entrant’s prospects and
further strengthen Microsoft. 1460 This third party also told us that the Microsoft
Cloud Remedy risked creating distortions that reduce competitors’
effectiveness and increase costs in the market. It added that the Microsoft
Cloud Remedy was silent on the revenue share split between Microsoft and
eligible providers of cloud gaming services, and that if Microsoft proposed to
keep all revenues, eligible providers of cloud gaming services could not
recoup costs associated with including Activision content on their services,
still less generate any margin from distributing Activision content. Moreover,
this third party told us that the Microsoft Cloud Remedy risked forcing the
market to develop via PC game storefronts as opposed to other types of

1457 [] submission to the CMA.

1458 [] response to the remedies questionnaire.
1459 [] response to remedies questionnaire.
1460 [] submission to the CMA.

356
 business models that cloud gaming service providers might have otherwise
developed, given that they were ‘open to exploring different ways of
monetising their services’ and their ‘current business models were not
fixed’. 1461

11.76 Another third party ([]) submitted that the obligation to make CoD available
to other platforms and consoles should also extend to other subscription
services and cloud gaming services. It told us that this obligation to release in
parity and simultaneously should apply to subscription services and not just
buy to play only. 1462

11.77 On the monitoring and enforcing of the Microsoft Cloud Remedy, one third
party ([]) told us that behavioural remedies could be enforced with an
‘effective monitoring program’. 1463 However, another third party ([]) told us
that while behavioural remedies were generally difficult to appropriately
monitor and enforce, this was ‘even more so in an industry like cloud gaming’,
which was evolving at a fast pace, towards an unpredictable direction. It told
us that these factors, coupled with the range of foreclosure strategies
available to Microsoft, made it challenging to monitor Microsoft’s compliance
with its proposed access remedy. It told us that Microsoft’s statement that the
Microsoft Cloud Remedy was ‘self-executing and its implementation does not
require significant monitoring or supervision’ significantly understated the
complexities of its enforcement. This third party added that it was clear that
while no significant monitoring of the Microsoft Cloud Remedy was in
Microsoft’s interest, it would not serve the wider interests of competition or
consumers. 1464

Assessment of the effectiveness of a behavioural remedy

11.78 We have assessed the effectiveness of the Microsoft Cloud Remedy in

comprehensively addressing the SLC that we have found. We consider this
remedy proposal to be predominantly behavioural in nature as, while it
involves a licence (which CMA guidance notes may be viewed under certain
limited circumstances as a specialised form of asset divestiture 1465), the most
significant aspects of the remedy involve ongoing measures designed to
regulate or constrain the behaviour of the merger parties. 1466

1461 [] submission to the CMA.

1462 [] response to remedies questionnaire.
1463 [] response to remedies questionnaire.
1464 [] submission to the CMA.
1465 CMA87, paragraph 6.2.
1466 CMA87, paragraph 3.34.

357
11.79 We have considered the suitability, in principle, of a behavioural remedy in
this case to address the competition concerns that have been identified,
taking into account the nature of competition in the relevant market. We then
looked in more detail at risks to effectiveness concerning specification,
circumvention, market distortion and monitoring of the proposed remedy. We
set out our assessment of these risks below.

Appropriateness in principle of a behavioural remedy in this case

Application of CMA’s Merger Remedies Guidance

11.80 The CMA’s Merger Remedies Guidance sets out the established position that
behavioural remedies are, due to their overall risk profile, unlikely to deal with
an SLC and its adverse effects as comprehensively as structural
remedies. 1467 However, the Merger Remedies Guidance 1468 also states that
‘[b]ehavioural remedies can operate satisfactorily in limited circumstances,
especially where the company operates in a regulated environment and
where there are expert monitors’. As noted above, our Merger Remedies
Guidance states that we will generally only select behavioural remedies as the
primary source of remedial action in a phase 2 merger investigation where:

(a) structural remedies are not feasible 1469 or the relevant costs of any
feasible structural remedy far exceed the scale of the adverse effects of
the SLC; 1470

(b) the SLC is expected to have a short duration; 1471 or

(c) behavioural measures will preserve substantial RCBs that would be

largely removed by structural measures. 1472

11.81 In the present case, cloud gaming is a nascent, rather than technologically
mature, market. The market is not regulated, so no expert sectoral regulator is
present to monitor and enforce a behavioural remedy. Instead, the Microsoft
Cloud Remedy includes provisions for monitoring and arbitration by
independent third parties (overseen by, or involving the participation of, the
CMA). We consider this further from paragraph 11.126 below. In relation to

1467 CMA87, paragraph 3.5.

1468 CMA87, paragraph 3.48.
1469 CMA87, paragraph 7.2.
1470 CMA87, paragraph 3.48.
1471 CMA87, paragraph 7.2.
1472 CMA87, paragraph 7.2.

358
 the other general factors set out in our guidance about the use of behavioural
remedies in merger control, we note that:

(a) As this is an anticipated merger, the structural remedy of prohibition is

clearly feasible and partial divestiture may also be feasible, although (as
explained in paragraph 11.41 above) we have not been able to consider
this in the depth needed to form a definitive view. This is because
Microsoft told us [].

(b) The SLC is not time-limited, and there is no reason to expect it will have a
short duration.

(c) The Parties have submitted that there are substantial RCBs that would be
lost if the Merger was prohibited but could be preserved with behavioural
remedies.

11.82 Our analysis of the RCBs claimed by the Parties is set out as part of our
proportionality assessment at paragraphs 11.135 to 11.307 below. The
remainder of this section considers the effectiveness of the Microsoft Cloud
Remedy. As explained in this chapter, we have engaged in detail with the
Parties and third parties to inform our assessment of the effectiveness of the
Microsoft Cloud Remedy.

Experience of the application of behavioural remedies in previous cases

11.83 The CMA’s 2019 update to its programme of evaluations of merger

remedies 1473 found that ‘the circumstances in which behavioural remedies are
the right outcome of merger control are rare’, and that ‘where there is no
expectation that the need for the remedy is itself in some way time-limited, the
case for behavioural remedies is weaker still, as there is a greater likelihood
that the remedy will either become ineffective or start to distort outcomes’. 1474

11.84 While the CMA’s 2019 evaluation report found that ‘if sufficient care is taken
over the design and implementation of behavioural remedies, and if active
and informed monitoring arrangements are put in place, behavioural remedies
can be at least partially effective for a limited period of time in narrowly
defined circumstances’, 1475 it also found that behavioural remedies ‘are more
complex and carry significantly higher risks than structural remedies and
generally require more work both in upfront design and implementation’. 1476

1473 Merger Remedy Evaluations – report on case study research (CMA109), June 2019.
1474 CMA109, paragraph 5.27.
1475 CMA109, paragraph 5.27.
1476 CMA109, paragraph 5.26.

359
11.85 Although the CMA will assess whether any proposed behavioural remedies
may be effective in each individual case, by reference to the specific
circumstances of that case, the CMA’s past experience of using behavioural
remedies is informative in highlighting the risks that can arise and how those
risks can impact the effectiveness of a remedy. This experience is also
reflected in the CMA’s Merger Remedies Guidance. We have therefore taken
account of this experience in informing our assessment of the remedies
proposed by Microsoft in this case. We nevertheless carefully considered the
possibility of the Cloud Remedy being potentially effective in this case, as
reflected by our in-depth assessment and engagement with the Parties and
third parties.

The nature of competition in the relevant market and of the SLC

11.86 Cloud gaming is a nascent market, which we have found will likely continue to
grow significantly in terms of revenues and users and become profitable in the
next five years. 1477 The market is currently characterised by incumbent firms
such as SIE and Microsoft offering cloud gaming as an extension to their
existing services, and new entrants seeking to gain market share through
innovative business models. We expect competition in this market to continue
to be dynamic and unpredictable, with significant uncertainty in the way that
the market may develop in the future.

11.87 Behavioural remedies are, by their nature, static rules restricting the conduct
of firms and are correspondingly limited in the extent to which they can adjust
effectively to changes in competitive conditions. They rely on obligations on
firms that are predominantly framed by the current conditions of competition.
While a behavioural remedy may be capable, at least in theory, of being
designed to be flexible to foreseeable changes in market and competitive
conditions (for example, the enactment of proposed legislation), it cannot be
drafted to take into account unforeseen and in some cases unforeseeable
changes in market and competitive conditions. The market we are concerned
with is dynamic and fast developing, which makes it particularly difficult to
accommodate the changes that may occur in the market within the
specification of a behavioural remedy. The CMA’s Merger Remedies
Guidance makes reference to this situation, stating ‘[w]here a market is likely
to be subject to frequent technological change or other wide-ranging market
developments, there is likely to be a significant risk that an access remedy [ie

1477 Chapter 8, paragraphs 8.14 to 8.52.

360
 a specific form of behavioural remedy] will become ineffective if the terms of
the access commitment do not accommodate these changes’. 1478

Assessment of the specific risks relating to the Microsoft Cloud Remedy

11.88 As well as having regard to the general considerations set out above
regarding the likely suitability of behavioural remedies in the circumstances of
this case, we have also considered in detail the specific risks arising from the
Microsoft Cloud Remedy and its potential effectiveness in addressing the SLC
that we have found.

11.89 We first look at the extent to which the design of the remedy would enable it to
be an effective remedy to the SLC. In line with CMA guidance, 1479 this
assessment includes the extent to which the remedy addresses the SLC, the
duration and timing of the remedy, and its practicality.

11.90 We also consider its risk profile and look at the four main risks associated with
behavioural remedies as set out in our guidance: 1480

(a) Specification risks: these risks arise if the form of conduct required to
address the SLC or its adverse effects cannot be specified with sufficient
clarity to provide an effective basis for monitoring and compliance.

(b) Circumvention risk: as behavioural remedies generally do not deal with

the source of an SLC, it is possible that other adverse forms of behaviour
may arise if particular forms of behaviour are restricted. Therefore, to
avoid or reduce these risks, behavioural measures need to deal with all
the likely substantial forms in which enhanced market power may be
applied.

(c) Distortion risks: these are risks that behavioural remedies may create
market distortions that reduce the effectiveness of these measures and/or
increase their effective costs. Distortion risks may result from remedies
overriding market signals or encouraging circumvention behaviour.

(d) Monitoring and enforcement risks: even clearly specified remedies may be
subject to significant risks of ineffective monitoring and enforcement. This
may be due to a variety of causes, such as the volume and complexity of
information required to monitor compliance; limitations in monitoring
resources; asymmetry of information between the monitoring agency and

1478 CMA87, paragraph 7.19.

1479 CMA87, paragraph 3.5.
1480 CMA87, paragraph 7.4.

361
 the business concerned; and the long timescale of enforcement relative to
a rapidly moving market.

Effectiveness of the design of the Microsoft Cloud Remedy

• Does the remedy address all aspects of the SLC?

11.91 We have identified two significant limitations in scope for the Microsoft Cloud
Remedy. These relate to the extent to which the remedy is able to address
concerns relating to certain existing and future business models, and the
extent to which the remedy would be effective in supporting cloud gaming
services using non-Windows OSs.

11.92 In assessing the effectiveness of a remedy, we do not seek to recreate an

expected counterfactual outcome at a particular point in time but to recreate
the dynamic conditions of competition present in the counterfactual. The CMA
views competition as a dynamic process of rivalry between firms seeking to
win customers’ business over time. 1481 As such, in assessing the
effectiveness of a remedy in addressing the SLC and resulting adverse
effects, it is important to consider the impact of any remedy on preserving that
dynamic process in the market, taken as a whole.

o Does the remedy cover alternative business models?

11.93 We have found that Activision would be likely to have made its games
available – including day and date releases – on cloud gaming services in the
next five years. We considered that this was more likely for cloud gaming
services which do not have an MGS-based model, ie those with a B2P or
BYOG model. 1482 However, we did not rule out that cloud gaming services
with different business models would arise absent the Merger, and we note
that it is difficult to predict with any certainty how an emerging and dynamic
market will continue to evolve.

11.94 The Microsoft Cloud Remedy offers a licence to stream Eligible Games to
consumers who have bought a copy of the game. It also licenses certain
cloud gaming services to stream Eligible Games, for which ‘consumers have
already obtained a licence for download of the game’. 1483

11.95 The design of the remedy does not make any provision for a direct
commercial relationship between the cloud gaming service and the publisher

1481 See eg CMA87, paragraph 3.5.

1482 Chapter 8, paragraph 8.278.
1483 Microsoft response to the Remedies Working Paper.

362
 of the Eligible Games (ie Activision). This limitation of the remedy restricts the
ability of the cloud gaming service to employ competitive strategies and
business models that we currently observe in the cloud gaming market, and
other strategies which might also be seen in the future in the absence of the
Merger. Such strategies and business models include, but are not limited to,
joint marketing arrangements, negotiation with games publishers to provide
exclusive or early access content (both of which are current features of
Activision’s commercial relationship with SIE) or to provide competitive
differentiation around game access or content, and MGS deals (such as
Amazon’s deal with Ubisoft).

11.96 The strategies listed in the paragraph above are only a few examples. Given
the dynamic nature of the market, we consider that there may be further
innovative competitive strategies that could be employed in the future that
cannot currently be predicted. This reflects our broad concern that, by seeking
to control outcomes in a market for a long period of time, the Microsoft Cloud
Remedy would be unable to adequately replace normal market-driven
incentives and strategies.

11.97 In its response to the Remedies Working Paper, Microsoft told us that there
was no requirement for the Consumer Licence to be paid for, and that the
definition of the licence does not specify the type of payment. It gave an
example, that ‘if a player is entitled to play an Eligible Game via a subscription
service that player is paying for, that player also has the right to play that
game via an Eligible Streaming Service’. 1484

11.98 We do not consider that this example demonstrates how the remedy
effectively addresses alternative business models to BYOG, such as
subscription services. In particular, it is not clear how a consumer would gain
the entitlement to play an Eligible Game via a subscription service. In this
example, the consumer would not necessarily have obtained a Consumer
Licence to stream, as they would not have purchased a copy of the game.
This would suggest that the consumer’s entitlement would derive from their
relationship with the cloud gaming service providing the subscription. In turn,
this would mean that the cloud gaming service would need to have a licence
from, or some other contract with, Microsoft to allow it to include the Eligible
Games in its subscription service. However, there is no provision for this in
the Streaming Provider Licence. The Streaming Provider Licence permits
streaming of Eligible Games ‘for the sole benefit of Consumers in accordance
with the Consumer Licence’. 1485 As we have observed earlier in this

1484 Microsoft response to the Remedies Working Paper.

1485 Microsoft, Annex 1 to the response to the Remedies Working Paper.

363
 paragraph, the subscription service consumer would not necessarily have a
Consumer Licence.

11.99 The Streaming Provider Licence also says that []. 1486 This appears []. It
suggests that there would be no financial incentive for a cloud gaming service
to []. It also prevents cloud gaming services from [], which we found to be
a feature of the console market and may become a feature of monetisation in
cloud gaming.

11.100 The Streaming Provider Licence also provides that []. 1487

11.101 In our view, this provision further demonstrates the limitations in scope
of the Microsoft Cloud Remedy. If Eligible Streaming Providers wish to offer
Eligible Games directly to their customers, our understanding is that they
would need to enter into a separate agreement with Microsoft, outside the
terms of the remedy. Our analysis in Chapter 8 has shown that, as a result of
the Merger, Microsoft would have the ability and incentive to foreclose rival
cloud gaming services in relation to Activision’s games. It follows from this
that Microsoft may not have the incentive to enter into such agreements
(either at all or on terms commercially acceptable to the third party provider).
Given this, we find that the Microsoft Cloud Remedy does not have a
sufficiently broad scope to cover these agreements, leaving this aspect of the
SLC without an effective solution.

11.102 We concluded that the Microsoft Cloud Remedy would not fully
address the adverse effects of the SLC on providers of cloud gaming services
wishing to use alternative business models, and to gamers who would benefit
from the availability of such models.

o Does the remedy cover non-Windows operating systems?

11.103 The Microsoft Cloud Remedy applies only to the PC and console
versions of a defined list of games (the Eligible Games). The PC versions are
those which are developed to run on a Windows OS.

11.104 This means that any cloud gaming service wishing to stream these
games would have to use, or be compatible with, the Windows OS version of
those games. This would exclude or restrict providers that may wish to
provide cloud gaming services using other OSs, either now or in the future.
This could exclude, for example, Apple, which has its own proprietary OS, as
well as Linux OS-based cloud gaming services and potential new entrants

1486 Microsoft, Annex 1 to the response to the Remedies Working Paper.

1487 Microsoft, Annex 1 to the response to the Remedies Working Paper.

364
 using a different OS. We have found that Microsoft would have the ability and
incentive to foreclose rival cloud gaming service providers post-Merger. This
means that cloud gaming services using other OSs would either need to
switch to Windows (and incur Windows licensing fees, which we found to be a
high proportion of overall costs for cloud gaming providers) or seek to adapt
the Windows version of the game to run on their alternative OS (eg through
the use of a compatibility layer such as Proton, which we have already found
is not sufficiently effective to overcome Microsoft’s advantage through
Windows). 1488 This has the potential to make it significantly more difficult for
such providers to enter, grow and compete against Microsoft.

11.105 In response to these concerns, Microsoft told us that it would grant

streaming rights for MacOS and other PC OS versions of the Eligible Games
‘as may be released during the term of the Microsoft Cloud Remedy’. It noted
that Activision had previously released MacOS versions of a small number of
games. 1489

11.106 While this commitment brings non-Windows PC versions within the

scope of the Microsoft Cloud Remedy, it only covers games for which
Microsoft chooses to make alternative versions. In the absence of the Merger,
Activision would seek to maximise the value that it was able to derive from
these games, which would have involved consideration of making non-
Windows PC versions of its games (as it already has done in some cases). In
our view, it is likely that Activision would continue to develop non-Windows PC
versions of at least some of its games absent the Merger, and the incentives
to do so would increase if demand for cloud gaming services that use PC
operating systems other than Windows were to grow. However, after the
Merger, Microsoft’s incentives to make these games compatible with rival OS
would be significantly lower, as this would both increase the attractiveness of
rival cloud gaming services and divert demand away from Windows OS.
Given this, the proposed remedy would put non-Windows based cloud gaming
services at a disadvantage, and potentially distort the choice of OSs for new
entrants.

11.107 In the Remedies Working Paper, we stated that an effective remedy

should ensure that future versions of Eligible Games would be able to run on
non-Windows OSs through compatibility layers. 1490 In its response, Microsoft
told us that this was not a reasonable requirement because it could not
ensure that a game could operate through a compatibility layer, that only the
third-party developer of the compatibility layer could update, when Microsoft

1488 Chapter 8, paragraph 8.221.

1489 Microsoft response to the Remedies Working Paper.
1490 CMA, Remedies Working Paper.

365
 updated Windows, and that some titles could not run on open-source OS
because of security concerns regarding cheating. Instead, it proposed a
commitment within the Cloud Remedy that it ‘shall not implement any features
or functions in the Eligible Games after the Commencement Date with the
deliberate purpose of preventing or negatively affecting the operation or
performance of any compatibility layer or of the Eligible Games running on a
compatibility layer’. 1491

11.108 We acknowledge that it may not be practical for Microsoft to ensure

that the Eligible Games would be able to run through compatibility layers.
However, while this commitment aims to provide some reassurance that
Microsoft will not deliberately seek to circumvent the intent of the remedy by
preventing or negatively affecting the Eligible Games from being played on
non-Windows OSs via compatibility layers, we consider the effect of any such
reassurance to be limited. It relies on an assessment of the motivation behind
any action or inaction by Microsoft, which is likely to be difficult for any party
other than Microsoft to assess in a sufficiently robust manner for the purposes
of implementing the remedy. Moreover the remedy includes circumstances
where the commitment would not apply – ‘where such measures are required
for an objective reason (eg to prevent hacking) and Microsoft has obtained the
prior consent of the Monitoring Trustee’. 1492 In our view, the reference to
‘objective reasons’ makes this exception both extremely broad, ambiguous
and very difficult for a Monitoring Trustee to oversee and adjudicate on.

11.109 The Cloud Remedy also covers PlayStation versions of the Eligible
Games. We considered whether Microsoft could foreclose SIE’s cloud gaming
service by making the PlayStation version of the game less attractive than the
Xbox or PC streamed versions.

11.110 Based on the above, we have concluded that the Microsoft Cloud
Remedy does not comprehensively address the SLC and would afford only
limited protection from the adverse effects of the SLC to providers of cloud
gaming services wishing to use a non-Windows OS, and to gamers who
would benefit from the availability of such services.

o Conclusion on scope of the Microsoft Cloud Remedy

11.111 In summary, the Microsoft Cloud Remedy seeks to standardise the

terms and conditions on which the Eligible Games are available, as opposed
to them being determined by the dynamism and creativity of competition in the
market, as would be expected in the absence of the Merger. The design of the

1491 Microsoft, Annex 1 to the response to the Remedies Working Paper.

1492 Microsoft, Annex 1 to the response to the Remedies Working Paper.

366
 Microsoft Cloud Remedy, and in particular its restriction to copies of the
Eligible Games purchased by consumers, and its restriction to PC and
console versions, present significant limitations to the scope of the remedy
and its ability to comprehensively address the SLC we have found, particularly
in relation to different business models and PC OSs other than Windows.

11.112 These limitations lead us to consider that the Microsoft Cloud Remedy
would not provide a comprehensive solution to the SLC. In addition, we
consider that the limitations are inherent in the design of the remedy. Given
the dynamic and rapidly changing nature of the market, and the likely entry
and development of cloud gaming services using different OSs and business
models, we do not consider that the remedy could be modified to
accommodate these developments to a sufficient extent to replicate the
competitive conditions that would have prevailed absent the Merger.

11.113 We have identified other limitations and effectiveness issues relating to

the scope, timeliness, practicality and risk profile of the remedy, which we set
out in the sections below. These issues reinforce our concerns that the
Microsoft Cloud Remedy is not an effective remedy, and mean that we would
not have a high degree of confidence as to the success of the remedy even if
it could be appropriately scoped. However, even absent the significant
additional risks below, we consider the fundamental limitations in scope set
out in paragraphs 11.91 to 11.112 above would by themselves render the
remedy ineffective in terms of providing a comprehensive solution to the SLC.

• Timing and duration

11.114 The Microsoft Cloud Remedy has a duration of ten years. We note that
some third parties told us that this was a sufficient duration to remedy the
SLC. However, the SLC in cloud gaming services that we have found arises
from a structural change in the market. As a result, the SLC and the adverse
effects arising from it are not time limited and could endure beyond ten years.
Although we recognise the possibility that the changing nature of the market
might result in circumstances where the SLC may no longer apply, we do not
have a high degree of confidence in such an outcome. While the Consumer
and Streaming Provider Licences are perpetual, the other protections and
commitments of the remedy, including monitoring and enforcement, would
expire after ten years, leaving it materially weaker.

11.115 Our view is that the time-limited nature of the Microsoft Cloud Remedy
is a clear and further weakness in terms of its effectiveness as a
comprehensive solution to the SLC. While the duration of the remedy could be
extended, or the end-date removed, this would create additional risks in terms

367
 of specification in the context of obsolescence and/or distortion of the market,
and in terms of effective monitoring.

• Practicality

11.116 The Microsoft Cloud Remedy does not affect a single party, but instead
represents a set of commitments that would be open to any eligible cloud
gaming service.

11.117 However, it is a complex remedy involving Microsoft committing to

develop and maintain commercial and technological relationships with
competing cloud gaming services and game storefronts. Furthermore, it
requires third-party arbitration, monitoring and enforcement, with the functions
of these third parties and the CMA needing to be fully defined and
established. We examine the difficulty of specifying a remedy in this evolving
context below, as well as other implementation risks in the section on
monitoring and enforcement risks starting at paragraph 11.126 below.

Risk profile of the Microsoft Cloud Remedy

• Specification risks

11.118 Cloud gaming is an early-stage and growing dynamic market, and

there is considerable uncertainty as to how it will develop and what competing
business models will emerge. We believe, for the reasons set out in detail in
Chapter 8, that foreclosure of Microsoft’s rivals in cloud gaming services may
be expected to result in substantial harm to competition in this market. We
recognise that we cannot predict with any certainty how exactly the market
might evolve absent the Merger (or if the Merger is allowed to proceed on the
basis of the Microsoft Cloud Remedy). Neither, in our view, can the Parties or
third parties. We consider this represents an inherent specification risk in the
Microsoft Cloud Remedy – even if the remedy could be well-specified to cover
the current status of the market, it may not be suited to future changes. This
means that we cannot have a high degree of confidence that the terms of the
remedy would be sufficiently well-specified to address these unpredictable
market changes.

11.119 We consider that this additional specification risk contributes to the

overall risk profile of the remedy.

• Circumvention risks

11.120 Given Microsoft’s ability and incentive post-Merger to foreclose other

cloud gaming providers, and the complexity of offering a high-specification

368
 gaming experience to customers, there are a variety of ways in which user
experience might worsen on competing platforms (see, for example, the
factors set out in paragraph 11.65). While the Microsoft Cloud Remedy is
intended to restrict Microsoft’s ability to pursue such a strategy, the range of
means by which it could worsen user experience on other platforms, and the
asymmetry of information between Microsoft and other parties, represent a
source of circumvention risk.

11.121 In this context we note that Microsoft’s agreement with NVIDIA [].
Similar provisions are in Microsoft’s agreements with other cloud gaming
services.

11.122 We note that the inclusion of these terms in the NVIDIA agreement
illustrates that even when there is a greater parity of bargaining power and
information between the parties, the []. This indicates that the parties are
not able to anticipate or capture all of their needs in advance. We anticipate
that any other cloud gaming provider is likely to have analogous needs to
NVIDIA to access Activision content.

11.123 Furthermore, the NVIDIA agreement itself recognises that its terms
cannot accurately specify future market conditions. The [] clause states:
[]. 1493 [], giving rise to the specification, circumvention and monitoring
risks set out in the previous paragraph. The presence of this clause
demonstrates clearly that it is inherently difficult to accurately specify the
scope and detailed terms of a remedy ex ante in a rapidly evolving market
such as cloud gaming.

11.124 We note, in addition, that the obligations set out in Microsoft’s

agreement with NVIDIA are [], 1494 []. 1495 The remedy does not include
such provisions, which would in any case be difficult to specify
comprehensively and precisely, and impracticable to monitor and enforce
effectively. This limitation in its specification represents a risk that Microsoft
would be able to deteriorate the service of its cloud gaming rivals.

• Distortion risks

11.125 As discussed above in paragraphs 11.91 to 11.112, the limitations in

scope of the Microsoft Cloud Remedy give rise to substantial distortion risks
relating to the business models and OSs employed by rival cloud gaming
providers. For example, it risks limiting innovation and competition in the B2P

1493 Microsoft, Annex 2 to the response to the Remedies Notice.

1494 Microsoft, Annex 2 to the response to the Remedies Notice.
1495 Microsoft, Annex 2 to the response to the Remedies Notice.

369
 and BYOG segments, as there would be extremely limited scope for content
exclusives, which are a current feature of the console market and one we
would expect, in the absence of the Merger, to be part of the cloud gaming
market.

• Monitoring and enforcement risks

11.126 As part of our consideration of this remedy option and taking into
account the CMA’s experience of the difficulties that can arise, in practice,
when overseeing complex behavioural remedies as captured in the CMA’s
Mergers Remedies Guidance, 1496 we have assessed the risks and challenges
associated with the monitoring and enforcement of the Microsoft Cloud
Remedy.

11.127 In addition to the specification, circumvention and distortion risks set

out above, we have identified five main areas of concern in relation to
monitoring and enforcement:

(a) The fact that this is a new, dynamic, and growing market means that the
scope of the core definitions (such as Eligible Game, Authorised PC
Game Store and Eligible Streaming Service) is likely to evolve
substantially over the lifetime of the Microsoft Cloud Remedy. This will
make it harder for the CMA and Monitoring Trustee to assess whether or
not Microsoft is meeting its obligations at any point in time.

(b) The remedy essentially proposes that the CMA would oversee various
arrangements that seek to regulate the behaviour of global firms in a
complex technological sector that is subject to rapid growth, evolving
business models and changing commercial practices. Notwithstanding the
appointment of a Monitoring Trustee, the Microsoft Cloud Remedy is likely
to place significant demands on CMA resources for the duration of its
proposed term, principally through the CMA’s extensive monitoring and
enforcement responsibilities across the broad scope of the remedy, its
oversight and governance of the Monitoring Trustee and any of its
advisers, and its participation in any dispute resolution process.

(c) The Microsoft Cloud Remedy requires the interaction of a number of third
parties and processes (eg Microsoft’s Compliance Director, reporting
requirements, Monitoring Trustee, third party dispute resolution), each
contributing to the monitoring and enforcement of the remedy. While there
is a need for the involvement of these different parties, the resulting

1496 For example, see CMA87, paragraphs 3.42, 3.52, 7.4 and 7.18.

370
 organisational complexity creates an additional challenge in ensuring that
the Microsoft Cloud Remedy is monitored and enforced effectively in the
longer-term. 1497

(d) The proposed compliance reporting processes are insufficient, with no

responsibility for Microsoft to proactively report non-compliance, and
insufficient provision for the Monitoring Trustee to audit Microsoft’s
compliance. In addition, the Monitoring Trustee’s analytical duties are
broad and complex, giving rise to risks that it would not have the
capability to monitor and enforce the remedy effectively.

(e) The dispute resolution process, while described as ‘fast-track’, may take
several months, risking potential harm that could be caused to Eligible
Streaming Services through a protracted process. Notwithstanding the
possible extent of the demands that implementing the remedy is likely to
place on the CMA (as noted in sub-paragraph (b) above), the CMA’s role
in the process, which is primarily one of oversight, gives rise to a risk that
it will not have adequate control over the practical application of the
remedy (which would instead largely be determined by the Monitoring
Trustee).

Conclusions on the effectiveness of the Microsoft Cloud Remedy

11.128 The Microsoft Cloud Remedy seeks to regulate a complex set of

arrangements between Microsoft, competitor cloud gaming services,
competitor PC game storefronts, and consumers.

11.129 The task of specifying these arrangements and Microsoft’s behaviour in

such a way that we can be confident that the SLC is comprehensively
remedied is made more difficult by the developing and dynamic nature of the
cloud gaming market, and our expectation that the future will see novel and
innovative business models and competitive strategies from both current
cloud gaming providers and new entrants.

11.130 In our view, the Microsoft Cloud Remedy suffers from material
limitations in scope. It is confined to a model based around customers buying
the Eligible Games before streaming them. It therefore risks excluding
alternative business models, such as inclusion of games within subscriptions,
that competing cloud gaming services may wish to pursue. It also risks limiting
innovation and competition in the B2P and BYOG segments, as there would
be extremely limited scope for content exclusives, which are a current feature

1497 See, the Open Banking Lessons Learned Review, 27 May 2022, paragraphs 112-115.

371
 of the console market and one we would expect, in the absence of the
Merger, to be part of the cloud gaming market.

11.131 In addition, although the remedy is in theory scoped to include versions

of the Eligible Games for non-Windows PC OSs, in practice, we consider that
Microsoft would have limited incentives to produce such versions, and lower
incentives than Activision has in the pre-Merger conditions of competition.
This could be expected to lead to distortions in the cloud gaming market and a
reinforcement of Microsoft’s advantages in OSs.

11.132 In addition to these fundamental limitations in scope, we have found

that the Microsoft Cloud Remedy gives rise to several significant risks to its
effective design and implementation, such that we do not have a high degree
of confidence as to the success of the remedy – ie that it will achieve its
intended effect – regardless of its limitations in scope. Specification of such a
complex remedy in an early-stage, growing, and dynamic market brings
inherent risks that future market developments cannot be effectively captured.
These specification risks in turn give rise to risks of circumvention by the
Parties and of market distortions. Furthermore, a complex remedy involving
third parties such as the Monitoring Trustee and those involved in dispute
resolution, and operating for a long duration, presents risks to the ability of
those third parties and the CMA to effectively monitor and enforce its
provisions.

Conclusions on remedy effectiveness

11.133 Based on the evidence provided to us and assessed above, we have

concluded that only a prohibition remedy would be effective in remedying the
SLC and adverse effects that we have found.

11.134 We have reached the following conclusions about two other remedy
options:

(a) Partial divestiture might in principle be an effective remedy, but []; and

(b) The Microsoft Cloud Remedy would not be effective in remedying the SLC
and adverse effects that we have found.

Proportionality

11.135 In order to be reasonable and proportionate, the CMA will seek to

select the least costly remedy, or package of remedies, of those remedy
options that it considers will be effective. In addition, the CMA will seek to

372
 ensure that no remedy is disproportionate in relation to the SLC and its
adverse effects. 1498

11.136 In conducting this proportionality assessment, we first consider whether

there are any RCBs which would be preserved or foregone under our
preferred remedy, before considering the other factors relevant to
proportionality.

Relevant customer benefits

11.137 When deciding on remedies, the CMA may have regard to the effects
of remedial action on any RCBs. 1499 In this section, we consider whether there
are any RCBs (within the meaning of the Act 1500) that should be taken into
account in our remedy assessment.

11.138 RCBs that will be foregone due to the implementation of a particular

remedy may be considered as costs of that remedy. The CMA may modify a
remedy to ensure retention of an RCB or it may change its remedy selection.
For instance, it may decide to implement an alternative remedy, or in rare
cases, it may decide that no remedy is appropriate. 1501

Framework for assessment of RCBs

11.139 The Act defines RCBs as a benefit to relevant customers in the form of
lower prices, higher quality, or greater choice of goods or services in any
market in the UK, or greater innovation in relation to those goods or
services. 1502 For these purposes, relevant customers are direct and indirect
customers (including future customers) of the merger parties at any point in
the chain of production and distribution – they are not limited to final
consumers. 1503

11.140 In addition, in the case of anticipated mergers, to be properly

considered as an RCB under the statutory definition, the CMA must believe
that: 1504

(a) the benefit may be expected to accrue within a reasonable period as a

result of the creation of the relevant merger situation concerned; and

1498 CMA87, paragraph 3.6.

1499 Section 36(4) of the Act.
1500 Section 30 of the Act.
1501 CMA87, paragraph 3.16.
1502 Section 30(1)(a) of the Act.
1503 Section 30(4) of the Act; and CMA87, paragraph 3.18.
1504 Section 30(3) of the Act.

373
 (b) the benefit is unlikely to accrue without the creation of that situation or a
similar lessening of competition.

11.141 Our Merger Remedies Guidance states that the merger parties will be
expected to provide ‘convincing evidence’ regarding the nature and scale of
RCBs that they claim to result from the merger and to demonstrate that these
fall within the Act’s definition of such benefits. 1505 The merging parties’
incentives to implement and pass on to customers the benefits post-merger
will also be relevant to the likelihood of RCBs being realised in practice.

RCBs claimed by Microsoft

11.142 In our Remedies Notice, we invited views on the nature of any RCBs
and on the scale and likelihood of such benefits and the extent (if any) to
which these were affected by different remedy options. 1506

11.143 Microsoft claimed that five RCBs would arise as a result of the Merger:

(a) Benefits from CoD and other Activision games being available on cloud
gaming services (the claimed Cloud Gaming RCB);

(b) Benefits from future versions of CoD being available on Nintendo

consoles (the claimed Nintendo RCB);

(c) Benefits from Microsoft’s expansion into mobile gaming (the claimed
Mobile Gaming RCB); and

(d) Benefits from placing Activision content on Game Pass (Xbox and PC),
including day and date releases of future CoD titles (the claimed Game
Pass RCB).

11.144 The claimed RCBs were presented to the CMA staff team orally by
Microsoft and its advisers at a meeting on 20 February 2023. Further detail
was provided in Microsoft’s response to the Remedies Notice (received on 22
February 2023), and at Microsoft’s response hearing on 27 February 2023. An
updated estimate of the value of RCBs was included in Microsoft’s response
to the Provisional Findings (received on 2 March 2023), and a new version of
the Microsoft Cloud Remedy was received on 6 March 2023. RCBs were also
discussed at a meeting with CMA staff on 14 March 2023. Microsoft also
made further submissions on RCBs in its response to our Remedies Working

1505 CMA87, paragraph 3.20. For example, in a previous phase 2 case in which RCBs were accepted, the type of

evidence provided included implementation plans which were detailed and advanced. See the CMA’s
investigation into the anticipated merger between Central Manchester University Hospitals NHS Foundation Trust
and University Hospital of South Manchester NHS Foundation Trust (2017).
1506 CMA, notice of possible remedies (Remedies Notice), 8 February 2023, paragraph 52(b).

374
 Paper. This was followed up by a call between the CMA staff team and
Microsoft on 4 April 2023, which included a discussion on certain aspects of
Microsoft’s submission on RCBs in its response to our Remedies Working
Paper, and a further submission from Microsoft on 6 April 2023 containing
Microsoft’s response to the CMA staff team’s questions.

11.145 In the remainder of this section, we first summarise the views of third
parties. We then assess each claimed RCB in turn. Finally, we conclude on
the scale of any of the claimed RCBs that we consider meet the statutory
threshold.

Views of third parties

11.146 Third parties provided the following evidence on RCBs:

(a) Three third parties ([], [] and []) told us that there are no potential
customer benefits arising from the Transaction. 1507

(b) Two other third parties ([] and []) told us that the remedies involving
open distribution across cloud gaming and the offers to console players
would be a good outcome for the franchises. 1508

(c) Another third party ([]) told us that Microsoft’s commitment to make
Activision-Blizzard games widely available via the cloud would provide
gamers with a fair choice as to where they play games, that other industry
players will explore multiple ways they can compete with each other (eg
gameplay quality, [], support of new platforms, etc), and could
potentially result in lower prices. 1509

(d) One third party ([]) told us that any behavioural commitment from
Microsoft to grant rivals access to CoD could pose a risk for consumers,
as there are numerous ways Microsoft could withhold or degrade access
which would be ‘extremely difficult to monitor and police’. 1510 The same
third party also told us that adding CoD to Game Pass would be a ‘good
thing’ for existing Game Pass subscribers who play CoD, but that a larger
population of [] gamers would suffer due to the foreclosure strategies
Microsoft could engage in. This third party told us that it would have to
consider adding its own first-party content day and date on its subscription
platform if Microsoft were to offer CoD day and date on Game Pass, but
that doing so would diminish its incentives to invest in its first-party

1507 [] and third party responses to the remedies questionnaire: []; []; and [].
1508 [] call note. [] response to the remedies questionnaire.
1509 [] response to the remedies questionnaire.
1510 [].

375
 content and would not be good for its gamers. 1511 Overall, the same third
party submitted that bringing CoD to Game Pass would not enhance
customer welfare and result in Merger-specific RCBs: 1512

(i) First, the third party submitted that Microsoft has explained that in the
future it expects the vast majority of gaming revenues to come from
gamers purchasing individual games rather than MGS. It noted that in
this scenario any RCBs resulting from the Transaction would be
limited and insufficient to outweigh the harm to competition.

(i) Second, it told us that, in the alternative scenario in which MGSs do

take off, Activision would make CoD available on PlayStation Plus
and Game Pass on similar terms and therefore, against this
counterfactual, the Transaction would reduce consumer choice and
foreclose rivals. 1513

11.147 We received one third-party submission on Microsoft’s published

response to the Remedies Notice. This third party ([]) told us that there was
no evidence that there were substantial customer benefits that would be
preserved by a behavioural remedy but not a structural one – in this regard, it
told us that: 1514

(a) First, in relation to Microsoft’s claims that day and date release of
Activision content on Game Pass would bring RCBs, even if such benefits
did exist, they would be ‘at best, short-term and limited’, and moreover,
any such benefits would be insufficient to outweigh the harm the SLC will
cause to cloud gaming – an entirely new form of game distribution that
may be ‘an important disruptive force in the gaming industry’, where ‘harm
to competition is particularly significant’ because the market ‘has the
potential to introduce more competition into a context where it has
previously only been possible for a small number of suppliers to compete’.

(b) Second, this third party told us that while Microsoft claimed that its
partnership with NVIDIA and Nintendo would bring CoD to ‘an additional
150 million gamers worldwide’, it told us that to derive the 150 million
figure, Microsoft had summed Nintendo’s entire installed base (around
123 million) and all existing GeForce Now subscribers (around 20 million).
However, this third party told us that according to Ampere Analysis, only 2
million of GeForce Now’s 20 million subscribers were paid members, and
Microsoft was silent on whether CoD would become available to the 18

1511 [].
1512 [] submission to the CMA.
1513 [] submission to the CMA.
1514 [] submission to the CMA.

376
 million free members. This third party also told us that it would not be
‘economically sound’ for Microsoft to make CoD available to trial
subscribers with free accounts when tens of millions of avid fans were
willing to pay $70 to buy the latest game.

(c) Third, even if the entire GeForce Now subscriber base were to have
access to CoD, the customer benefits that may be derived from
Microsoft’s agreement with NVIDIA were ‘hypothetical at best and
unlikely’ to materialise – in this regard, a user who could access CoD on
both B2P and streaming by subscribing to Game Pass Ultimate would not
buy an additional subscription to GeForce Now, ie Microsoft’s partnership
with NVIDIA would neither change the anticompetitive effects of the
Transaction on cloud gaming services market, nor disprove Microsoft’s
incentives to establish Game Pass dominance.

Our assessment

11.148 For each claimed RCB, we first assess whether it qualifies as an RCB
under the definition in the Act. 1515 This requires an assessment of whether it
meets the criteria in section 30(1) of the Act – which relate to the nature of the
claimed benefit – and section 30(3), which require that:

(a) the benefit may be expected to accrue within a reasonable period as a

result of the creation of the RMS; and

(b) the benefit is unlikely to accrue without the creation of the RMS or a
similar lessening of competition.

11.149 For any RCBs that are found to meet these statutory criteria, we then
assess the likely scale of that RCB.

11.150 Before going into our detailed assessment of the five claimed RCBs,
we first consider some broader points raised by Microsoft about the correct
application of the relevant statutory framework.

• The application of the statutory framework governing RCBs

11.151 In its response to our Remedies Working Paper, Microsoft argued that
the CMA had erroneously applied the relevant sections of the Act in its
assessment of the claimed Cloud Gaming RCB and claimed Nintendo RCB,
and that this ultimately resulted in an ‘irrational choice of remedy

1515 Section 30(3) of the Act.

377
 selection’. 1516 Microsoft submitted that, under the Act, a benefit may qualify as
an RCB where two cumulative tests are met: (a) the benefit may be expected
to accrue within a reasonable period as a result of the creation of the relevant
merger situation concerned; and (b) the benefit is unlikely to accrue without
the creation of that situation or a similar lessening of competition. Microsoft
submitted that the CMA had confused the analysis of section 30(3)(a) of the
Act by effectively asking two questions relating to the same causation issue
and imposing additional requirements on the RCB qualification that did not
exist in the Act or CMA guidance. For example, Microsoft argued that in
relation to the claimed Cloud Gaming RCB, while the Remedies Working
Paper had provisionally concluded that it would occur within a reasonable
period (ie the ‘timing element’), it (wrongly) sought further and separately to
consider under section 30(3)(a) whether the benefits might be expected to
accrue as a result of the creation of the RMS. Thus, Microsoft told us that the
Remedies Working Paper wrongly treated section 30(3)(a) of the Act as
having two separate tests, ie one on the timing of the RCB and the other on
the causation of the RCB. 1517

11.152 Microsoft argued that when considering the correct legal approach to
section 30(3)(a), it was simpler first to focus on section 30(3)(b) as the
relevant causation test (which asks whether the benefits are unlikely to accrue
without the RMS), and that if they are unlikely to accrue without an RMS, then
section 30(3)(b) is met, ie the relevant ‘causal threshold’ is passed. Microsoft
submitted that it was not appropriate for the CMA then to impose a further
causation threshold by reference to the language of section 30(3)(a): the use
of the words ‘as a result of the creation’ of an RMS in section 30(3)(a) cannot
impose a higher threshold for causation – if they did, section 30(3)(b) would
serve no practical purpose. Microsoft submitted that it was a basic principle of
statutory interpretation that an interpretation which gives meaning to all of the
relevant provisions is to be preferred. 1518 Microsoft also submitted that ‘[the]
CMA’s approach would perversely remove any incentive for companies to
take proactive steps to address regulatory concerns during a merger review
process, since under the [CMA’s] logic, such steps can simply be ignored
[…]’. 1519

11.153 We disagree with Microsoft’s interpretation of the statutory framework

for the assessment of RCBs. Microsoft’s proposed approach would fail to
comply with a key element of section 30(3)(a): the requirement for the benefit
to be causally connected to the RMS. Whilst we agree with Microsoft that

1516 Microsoft response to the Remedies Working Paper.

1517 Microsoft response to the Remedies Working Paper.
1518 Microsoft response to the Remedies Working Paper.
1519 Microsoft response to the Remedies Working Paper.

378
 there is a temporal element under section 30(3)(a): “within a reasonable
period”, there is also a causal element: “as a result of”. This is separate from
the requirement under section 30(3)(b) for the CMA to believe that the benefit
is unlikely to accrue without the creation of the RMS or a similar lessening of
competition. Under this limb, the CMA must consider what was likely to
happen in the absence of the RMS.

11.154 “Relevant merger situation” is defined in section 23 of the Act and

incorporates the concept of two or more enterprises ceasing to be distinct
(along with conditions as to turnover or share of supply). To satisfy the causal
requirement under section 30(3)(a), the CMA must therefore believe that the
benefit may be expected to accrue as a result of the enterprises ceasing to be
distinct.

11.155 The Merger Remedies Guidance gives examples of when this condition
may be satisfied: it notes that the merger may lead to economies of scale or
efficiencies that lead to lower prices, improved quality or greater
innovation. 1520 These are examples of how the merger might change the
merging parties’ marginal costs or commercial incentives, which in turn may
result in a benefit for customers.

11.156 Claimed benefits that are linked to reduced marginal costs or changes
in commercial incentives can be distinguished from a situation in which a
claimed benefit is conditional on completion of a merger but does not arise as
a result of the enterprises ceasing to be distinct. For example, an acquirer
could unilaterally pledge or contractually commit to make a charitable
donation to an organisation (of which relevant customers are beneficiaries) on
completion of the merger. Whilst any benefit arising from the donation will only
materialise in the event that the merger completes, it does not arise as a
result of any changes (eg in commercial incentives) flowing from the creation
of the RMS. Since the claimed benefit does not flow from the creation of the
RMS, the condition in section 30(3)(a) would not be satisfied.

11.157 As for section 30(3)(b), by asking whether the claimed benefit is

‘unlikely to accrue’ without the creation of the RMS, this section focuses on
what might occur absent the creation of the RMS. It asks whether the benefit
might be achieved by some means other than the creation of the RMS or a
similar lessening of competition. This condition therefore requires separate
consideration from the conditions in section 30(3)(a).

1520 CMA87, paragraphs 3.22 to 3.24.

379
11.158 We note that it is possible for a benefit to accrue within a reasonable
period as a result of the creation of an RMS (and thus satisfy the
requirements of section 30(3)(a)) but for it to have been likely to accrue in any
event even without the creation of that RMS or a similar lessening of
competition (and thus fail the requirements of section 30(3)(b)). Similarly, it is
possible fora benefit to be unlikely to accrue without the creation of an RMS or
a similar lessening of competition (and thus satisfy the requirements of
section 30(3)(b)) but not have a causal link with the RMS (and thus fail the
requirements of section 30(3)(a)).

11.159 The natural interpretation of these statutory provisions is consistent

with the explanatory notes accompanying the statutory provision, which
provide that RCBs are “narrowly defined [and]are not expected to arise often”
and that the definition is “further narrowed” by the requirement that “the
authority has to have an expectation that the benefits will be realised within a
reasonable time-frame as a result of the merger” and that “the benefits are
unlikely to arise without the merger (unless the only other ways of realising
the customer benefit would have a similarly detrimental effect on
competition)”.

11.160 We also disagree with Microsoft’s view that the CMA’s approach would
perversely remove any incentive for companies to take proactive steps to
address regulatory concerns during a merger review process. In addition to
the possibility of offering undertakings in lieu of a reference to phase 2, 1521
merging parties also have the option to engage in discussions with the CMA
about potential remedies at any stage of the phase 2 process, including
before the provisional findings (‘without prejudice’ to the question of whether
the merger gives rise to an SLC). 1522 There are further opportunities to
engage with possible remedies where the CMA provisionally finds an SLC at
phase 2, as the Merging Parties have done so in this case.

11.161 We therefore assess the requirements of each of sections 30(1),

30(3)(a) and 30(3)(b) separately in our assessment below.

• The claimed Cloud Gaming RCB

11.162 On [] February 2023, Microsoft signed an agreement with NVIDIA to

include Microsoft first-party content on its GFN cloud gaming service,
including Activision post-Merger. 1523 Under the terms of this agreement,

1521 Section 73 of the Act.

1522 CMA2, paragraph 12.14.
1523 GeForce NOW Listing Agreement, [] February 2023.

380
 Microsoft and NVIDIA will [] make Activision games available on GFN [].
The duration of the agreement is 10 years, []. [].

11.163 On [] March 2023, Microsoft signed an agreement with Boosteroid to

make Microsoft games (and Activision games from completion of the Merger)
available on Boosteroid’s clouding gaming service []. 1524 The duration of the
agreement is 10 years, [].

11.164 On [] March 2023, Microsoft signed an agreement with Ubitus to

make Microsoft games (and Activision games from completion of the Merger)
available on Ubitus’ GameNow clouding gaming service []. 1525 The duration
of the agreement is 10 years, [].

o Views of the Parties

11.165 In its response to the Provisional Findings, Microsoft told us that ‘the
availability of Activision content, including CoD, on NVIDIA GFN will enhance
competition in cloud gaming and result in a greater choice of goods and
services.’ 1526 Microsoft did not seek to quantify the benefits. Microsoft
submitted that customers will benefit from CoD and other Activision games
being made available on cloud gaming services. Microsoft referred to its
understanding of the CMA’s provisional view that [] and noted that the
Provisional Findings also found that strong indirect network effects mean that
publishers will not license their content to small providers given the costs
involved. Microsoft submitted that Activision has been clear []. Microsoft
submitted that even if []. 1527

11.166 Microsoft’s agreements with Boosteroid and Ubitus were signed after
Microsoft’s response to the Provisional Findings (and subsequent remedy
submissions as summarised in paragraph 11.144 above), and were not
therefore included in these submissions.

11.167 In its response to our Remedies Working Paper, Microsoft submitted

that the analysis put forward in the Remedies Working Paper as to whether
the benefits resulted from the Merger was flawed, and that: 1528

(a) The fact that the agreements were only entered into post-Transaction
announcement and in the course of regulatory investigation was not

1524 Cloud Gaming License Agreement, 9 March 2023.

1525 Cloud Gaming License Agreement, 11 March 2023.
1526 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 4.12.
1527 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 2.9.
1528 Microsoft response to the Remedies Working Paper.

381
 relevant – Microsoft submitted that there was no specific timing
requirement for when an RCB must arise.

(b) It was wrong to say that the benefits do not arise from the Merger on the
basis that there might (on the CMA’s analysis) be countervailing
incentives not to enter into the Agreements on completion of the Merger.
It told us that the simple fact remained that those benefits would be
unlikely to accrue absent the Merger and therefore do arise as a result of
the Merger.

11.168 Based on the above, Microsoft told us that the position taken in the
Remedies Working Paper on the claimed Cloud Gaming RCB was unsound
and imposed stricter standards for the assessment of RCBs than required by
the Act and the CMA’s own guidance, and added that the CMA’s approach
would ‘perversely remove any incentive for companies to take proactive steps
to address regulatory concerns during a merger review process, since under
the Remedies Working Paper’s logic, such steps could simply be ignored on
this basis alone’. 1529 We have already addressed Microsoft’s submissions on
the statutory criteria in sections 30(3)(a) and (b) of the Act above.

11.169 In its response to our Remedies Working Paper, Microsoft told us that
(a) any cloud streaming [] would be certain and much more immediate and
therefore benefit more consumers over a longer period; and (b) more cloud
streaming providers would be able to access the content than would have
been the case absent the Merger. 1530

11.170 On the merger-specificity of this claimed RCB and the potential for
Activision content to be put on a cloud gaming streaming service [],
Activision told us that it has a limited amount of resources and tries to allocate
those resources to the opportunities that are going to provide the greatest
return to its shareholders, or are aligned with its skills and capabilities. [].
On this basis, it considered Microsoft’s plans to put Activision content on
cloud as a result of the Merger to be both a benefit and merger-specific. 1531

11.171 Microsoft told us that the ‘criticism’ in the Remedies Working Paper that
the Parties had not quantified these benefits was ‘unfair’, and added that the
primary benefit of the claimed Cloud Gaming RCB was the expansion of
choice and wider accessibility of Activision content, which by its nature was
hard to quantify. However, it told us that it was undeniable that such features
would be clear benefits to gamers and that more gamers who did not have a

1529 Microsoft response to the Remedies Working Paper.

1530 Microsoft response to the Remedies Working Paper.
1531 Activision response hearing transcript.

382
 device suited to playing Activision content natively would have the option to
access it via cloud streaming on devices they already owned. 1532

11.172 Finally, Microsoft told us that once the claimed Cloud Gaming RCB
was accepted, the only possible conclusion was that prohibition would be a
disproportionate response to the SLC. 1533

o Our assessment of the claimed Cloud Gaming RCB

11.173 We have considered whether the claimed Cloud Gaming RCB qualifies
as an RCB under the Act.

11.174 First, we have considered whether the claimed Cloud Gaming RCB
satisfies the condition in section 30(1)(a) of the Act. As described above,
Microsoft’s agreements with NVIDIA, Boosteroid and Ubitus provide for
cooperation between the parties to allow customers of these cloud gaming
providers, including in the UK, to access Activision’s content. As such, to the
extent that any benefit arises (which is considered below), this would be in the
form of greater choice of goods or services in the market for cloud gaming
services in the UK, and would therefore satisfy the requirement under section
30(1)(a) of the Act.

11.175 Second, we have considered whether the claimed Cloud Gaming RCB
may be expected to accrue within a reasonable period as a result of the
creation of the RMS, as required under section 30(3)(a) of the Act.

11.176 We note that the agreements between Microsoft and NVIDIA,

Boosteroid and Ubitus all take effect with respect to Activision content
following completion of the Merger []. However, as explained above, the
fact that the relevant provisions in these agreements are conditional on
completion of the Merger does not necessarily mean that the claimed benefits
may be expected to arise from the creation of the RMS.

11.177 We note Microsoft’s submission in response to the Remedies Working

Paper that the agreements arise from changes to commercial incentives as a
result of the Merger. However, we have not seen any evidence to support this.
On the contrary, the evidence considered in the competitive assessment in
Chapter 8 supports our finding that Microsoft would have the incentive to
make Activision’s content exclusive to Microsoft’s cloud gaming service post-
Merger. The fact that Microsoft has entered into these agreements does not
undermine our findings on its post-Merger incentives. Microsoft may have

1532 Microsoft response to the Remedies Working Paper.

1533 Microsoft response to the Remedies Working Paper.

383
 short-term incentives to enter into these agreements to seek to address the
competition concerns arising from the Merger, but this is not informative of its
longer-term commercial incentives.

11.178 Accordingly, we do not believe that the claimed Cloud Gaming RCB
would arise from the creation of the RMS for the purposes of section 30(3)(a)
of the Act. However, even if the claimed Cloud Gaming RCB did not fail to
satisfy the condition in section 30(3)(a) on this basis, we are not satisfied that
any material benefits to relevant customers would accrue in any case for the
reasons given below.

11.179 As part of our SLC assessment, we have found that Microsoft will have
the incentive to foreclose its cloud gaming rivals. It is within this context that
we need to consider whether the claimed benefit may be expected to accrue.
As noted above, we have found a tension between the terms of the
agreements, which seek to provide rival cloud gaming services with Activision
games, and Microsoft’s post-Merger commercial incentives to make Activision
content exclusive.

11.180 As set out in our assessment in Chapter 8, 1534 we consider that

Microsoft’s overall strength in cloud gaming is relevant in this regard, as we
would expect it to hold considerable leverage in relation to any subsequent
negotiation or contractual dispute. In addition, these agreements contain
specific terms which introduce further uncertainty – for example:

(a) The NVIDIA agreement only requires []. 1535 This allows Microsoft to
[]. However, as noted above, we have found that Microsoft’s
commercial incentives are not to provide Activision’s content. In these
circumstances, we consider that [] may not lead to the Activision’s
content becoming available on NVIDIA GFN.

(b) The Boosteroid and Ubitus agreements each contain a provision allowing
Microsoft to []. 1536 It is not clear the extent to which such [] may arise,
although the fact that this provision has been included in these
agreements indicates that Microsoft sees this as a material risk. The
claimed benefit will not arise from these agreements to the extent that
[].

1534 See paragraphs 8.415-8.432.

1535 Preamble to the GeForce NOW Listing Agreement, [] February 2023.
1536 Clause 4.4 of the Cloud Gaming License Agreement, [] March 2023; Clause 4.4 of the Cloud Gaming

License Agreement, 11 March 2023.

384
 (c) The agreements with NVIDIA, Boosteroid and Ubitus all contain an []
clause (as discussed in paragraph 11.123 above). Under this clause, the
parties expressly recognise that [].

(d) For each of the agreements with NVIDIA, Boosteroid and Ubitus, [].

11.181 In light of the tension between the agreements and Microsoft’s post-
Merger commercial incentives, together with the material limitations on the
contractual obligations and protections referred to above, we consider that
there is significant uncertainty as to whether any material benefits, in
particular relating to the expansion of choice and wider accessibility of
Activision content on cloud gaming services, would accrue in practice.

11.182 We therefore do not consider that the condition in section 30(3)(a) is

satisfied in relation to the claimed Cloud Gaming RCB (at least in any material
respect). However, even to the extent that any benefit may arise, we consider
that this benefit is likely to have arisen anyway in the absence of the Merger,
and would not in any case be material, for the reasons given below.

11.183 Third, we have considered whether the claimed benefit was unlikely to
accrue without the RMS or a similar lessening of competition, so as to meet
the condition in section 30(3)(b) of the Act.

11.184 We have found in Chapter 8 that we expect the cloud gaming market to
continue to grow and that Activision would likely have made its games –
including day and date releases – available for cloud gaming in the next five
years. We found it likely, in particular, that Activision would have [] in the
near future absent the Merger.

11.185 We therefore consider that it is likely that the claimed benefit, in

particular the expansion of choice and wider accessibility of Activision content
on cloud gaming services, would accrue without the RMS. Given that
Activision’s incentives to make its content available on cloud gaming services
absent the Merger are greater than Microsoft’s incentives post-Merger (for the
reasons given above), any such benefit would likely be greater in the absence
of the Merger. We note that this benefit may have accrued over a slightly
longer period than any benefit arising under the claimed Cloud Gaming RCB.
However, any short-term additional benefit resulting from earlier access to the
Activision content on these cloud gaming services has not been quantified by
the Parties and in our view would not be material.

11.186 We therefore do not consider that the condition in section 30(3)(b) of

the Act is satisfied for the claimed Cloud Gaming RCB.

385
11.187 Finally, we note that even to the extent that any of the claimed Cloud
Gaming RCB were to satisfy the conditions in section 30(3)(a) and section
30(3)(b) of the Act to qualify as an RCB, we do not consider that the size of
the RCB would have a material impact on the costs of the prohibition remedy.

11.188 We note in this respect that Microsoft has not attempted to quantify the
benefit arising from the claimed Cloud Gaming RCB. Given the likelihood of
Activision content becoming available on cloud gaming services absent the
Merger, and the significant uncertainty in relation to the claimed benefits
arising from Microsoft’s agreements with NVIDIA, Boosteroid and Ubitus, we
find that any residual benefit that may arise as a result of the Merger (and
would not otherwise have arisen) would not be material.

11.189 In summary, we find that the claimed Cloud Gaming RCB does not
qualify as an RCB because it does not satisfy the conditions in sections
30(3)(a) and (b) of the Act. Even to the extent that any of the claimed benefit
does meet these conditions, we consider that any such benefit would be small
and transitory and would not have a material impact on the costs of the
prohibition remedy.

• The claimed Nintendo RCB

11.190 On [] February 2023, Microsoft and Nintendo entered into an

agreement in relation to the development and publishing of CoD titles on
Nintendo platforms. 1537 This agreement took the form of [] (which was
signed on []). 1538 Under the terms of the [], Microsoft will [] develop
and publish future native console versions of the CoD titles for Nintendo
platforms for at least 10 years []. 1539 The [] specifies that []. Except as
provided for in the [], Microsoft agrees to publish the CoD titles pursuant to
[]. 1540

o Views of the Parties

11.191 Microsoft submitted that the agreement will be for [] ten years,
[]. 1541
Microsoft further submitted that both Microsoft and Activision are
confident that the relevant games can be optimised to run on the Nintendo
Switch using standard techniques within a time period of around []

1537 [].
1538 [].
1539 [].
1540 [].
1541 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 2.3.

386
 months. 1542 Microsoft also noted []. 1543 Microsoft estimated the net present
value to Nintendo customers of having access to CoD over ten years to be at
least (£[] million). Further, Microsoft submitted that widening access to CoD
via Nintendo will increase the pool of gamers able to play CoD, thereby
improving the cross-play functionality of the game. It submitted that this
efficiency will arise from []. 1544

11.192 In its response to our Remedies Working Paper, Microsoft told us that
the claimed Nintendo RCB would accrue within a reasonable period, and
that: 1545

(a) The Remedies Working Paper had applied an ‘inconsistent time horizon’
in its assessment of the claimed Nintendo RCB compared to other
aspects of the CMA’s analysis of the Merger 1546 – it told us that if those
key aspects of the CMA’s case were to be judged against longer time
horizons, it followed that the RCB analysis should be as well, especially
as the claimed Nintendo RCB would accrue in a much shorter time period.

(b) Microsoft entered into a legally binding agreement to bring CoD to

Nintendo, [], and that the starting date of the agreement (ie []) was
intended to be the earliest date after which the next upcoming version of
CoD would be released on the Nintendo platform, taking into account the
required time to develop and test the game.

(c) The Remedies Working Paper’s position that the Nintendo agreement
contained an element of uncertainty (as to whether CoD would be
available on Nintendo at all) mischaracterised the contract, and that the
[] would fully apply to Microsoft []. Microsoft told us that there was no
uncertainty that CoD would be made available on Nintendo, since: (i) [];
and (ii) [].

11.193 In relation to whether the claimed Nintendo RCB is specific to the

Merger, Microsoft submitted in its response to our Remedies Working Paper
that: 1547

1542 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 2.3(b).
1543 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 2.3(b).
1544 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 2.3(c).
1545 Microsoft response to the Remedies Working Paper.
1546 In this regard, Microsoft cited that the counterfactual on which the Remedies Working Paper relied only

considered that Activision would put its content on cloud streaming services at some unspecified point in the
future and, likewise, the CMA placed limited weight on the fact that there was no evidence to suggest that cloud
game streaming would be profitable in the next five years. Source: Microsoft response to the Remedies Working
Paper.
1547 Microsoft response to the Remedies Working Paper.

387
 (a) From the very outset, shortly after the deal was announced in January
2022, Microsoft had expressed a clear intention and commitment to bring
CoD to Nintendo in future. 1548

(b) Microsoft has submitted that the reason it has a different strategy to
Activision in relation to making CoD available on Nintendo is because
large publishers with popular franchises have []. By contrast, as a
platform []. Accordingly, Microsoft submitted that []. 1549

11.194 Finally, Microsoft told us that Nintendo was a strong competitor in

consoles, with a []% share of UK console hardware in 2021. It also told us
that the available evidence did not support a view that, due to the
demographics of its customers, Nintendo would attract less demand for
games like CoD. As such, Microsoft submitted that there was no plausible
basis to conclude that the impact and size of the claimed Nintendo RCB
would be small. 1550

o Our assessment of the claimed Nintendo RCB

11.195 We have considered whether the claimed Nintendo RCB qualifies as

an RCB under the Act.

11.196 First, we have considered whether the claimed Nintendo RCB satisfies
the condition in section 30(1)(a) of the Act. Under the terms of the Nintendo
agreement, Microsoft will [] develop and publish future native console
versions of the CoD titles for Nintendo platforms. As such, to the extent that
any benefit arises (which is considered below), this would be in the form of
greater choice of goods or services in the market for console gaming services
in the UK, and would therefore satisfy the requirement under section 30(1)(a)
of the Act.

11.197 Second, we have considered whether the claimed Nintendo RCB may
be expected to accrue within a reasonable period as a result of the creation of
the RMS, as required under section 30(3)(a) of the Act.

11.198 We note that the relevant obligations under the Nintendo agreement
only take effect from []. In the event that the Merger does not complete,
these obligations will therefore not arise. However, as explained above, the
fact that the relevant obligations are conditional on completion of the Merger

1548 Microsoft response to the Remedies Working Paper.

1549 Microsoft submission to the CMA.
1550 Microsoft response to the Remedies Working Paper.

388
 does not necessarily mean that the claimed benefits may be expected to arise
from the creation of the RMS.

11.199 We have concluded in Chapter 7 that Microsoft would not have the
incentive to foreclose rival console gaming services in the UK. However, this
is not determinative of whether Microsoft has a commercial incentive to
develop and publish native CoD titles for Nintendo. This would involve
substantial costs that would need to be recouped through additional sales of
native versions of CoD on Nintendo. In addition, placing valuable CoD content
on Nintendo would in principle increase diversion away from Xbox and
towards Nintendo. While this effect may be relatively limited given our findings
in Chapter 7 that Nintendo competes less closely with Xbox, making CoD on
Nintendo could make it a closer competitor to Xbox, which we consider would
not be in Microsoft’s interest. The Parties have not provided us with
convincing evidence of the expected costs, revenues and profitability that
might inform an assessment of Microsoft’s commercial incentives in this
respect.

11.200 Microsoft also submitted that it has greater incentives to make CoD
available on Nintendo than Activision, for the reasons described in paragraph
11.193 above. However, Microsoft has provided no evidence in support of this
submission (for example, in terms of []). []). Microsoft has also not
provided any additional evidence demonstrating why the Merged Entity would
have stronger incentives to place CoD on Nintendo, as compared to Activision
alone pre-Merger. We note that Microsoft may have short-term incentives to
enter into this agreement to seek to address the competition concerns arising
from the Merger, but this is not informative of its longer-term commercial
incentives.

11.201 Accordingly, we do not believe that the claimed Nintendo RCB would
arise from the creation of the RMS for the purposes of section 30(3)(a) of the
Act. However, even if the claimed Nintendo RCB did not fail to satisfy the
condition in section 30(3)(a) on this basis, we are not satisfied that any
material benefits to relevant customers would accrue in any case for the
reasons given below.

11.202 As noted above, Microsoft’s obligations under the Nintendo agreement

with Nintendo only arise from []. As such, Microsoft would not be obliged to
start work on development of CoD titles for Nintendo platforms until [].
Microsoft has submitted that Microsoft and Activision consider that CoD
games can be optimised for Nintendo Switch in around [] months. However,
we have not been provided with any detailed plan or supporting evidence for
this estimate. Even on an []-month timeframe, any benefits for consumers

389
 may not accrue until [] at the earliest, and [] if Microsoft starts work in line
with its obligations.

11.203 We also note that there is significant uncertainty as to whether this

claimed benefit will arise at all (or, to the extent it does, that it will have a
material impact). While we note Microsoft’s submissions [], 1551 the fact that
Activision estimates optimisation for Nintendo Switch will take [] months
demonstrates the challenges that may be involved in this. The Parties have
not provided us with any detailed plans for how this optimisation is intended to
be achieved. There is therefore limited evidence for us to determine the risks
that these efforts may not be successful.

11.204 Further, Microsoft has provided no submissions or evidence on the

possible timeframe or risks involved in developing a CoD title [].
Furthermore, to the extent that []. As such, we are not satisfied that any
benefit would arise within a reasonable timeframe [].

11.205 We also note that whilst []. Given that []. The Nintendo agreement
itself acknowledges that []. There is therefore considerable uncertainty as to
the terms on which Microsoft may develop CoD titles for Nintendo platforms.

11.206 Accordingly, we are not satisfied that any such benefits would accrue
within a reasonable period (or would endure so as to provide a material
benefit for consumers in the market for console gaming services in the UK).
We therefore do not consider that the condition in section 30(3)(a) of the Act
is satisfied for the claimed Nintendo RCB.

11.207 Third – although not strictly necessary given our conclusion that the
claimed Nintendo RCB does not meet the condition in section 30(3)(a) – we
have considered whether the claimed benefit was unlikely to accrue without
the RMS or a similar lessening of competition, so as to meet the condition in
section 30(3)(b) of the Act.

11.208 In particular, we have considered whether the benefit might be

achieved through a commercial agreement between Activision and Nintendo.
The Nintendo Switch has a different technical specification to the Xbox X/S
and PS5. We also note that Activision does not consider []. However, [].
As noted above, []. We also note Microsoft’s submission that its financial
case for producing Nintendo versions of CoD may be different to Activision’s,
although as noted above we have not seen significant evidence to support
this claim.

1551 Microsoft response to the CMA’s RFI.

390
11.209 As such, we consider that the evidence is mixed regarding whether the
claimed Nintendo RCB is unlikely to accrue without the RMS or a similar
lessening of competition within the meaning of section 30(3)(b) of the Act.
However, it is not necessary for us to conclude on whether this criterion is
met, since we have already found that the claimed Nintendo RCB does not
meet the condition in section 30(3)(a) of the Act.

11.210 Finally, we note that even if the claimed Nintendo RCB were to satisfy
the condition in section 30(3)(a) of the Act to qualify as an RCB (which for
reasons given above, we do not consider to be the case), we do not consider
that the size of the RCB would have a material impact on the costs of the
prohibition remedy.

11.211 Microsoft has submitted that the claimed Nintendo RCB results in an
annual benefit of £[]m (average benefit of £[] per customer), which
corresponds to a 10-year NPV of £[]m. 1552 However, this calculation is
based on a number of assumptions which we do not consider to be justified.

(a) It is not clear how Microsoft has assessed the average annual ‘consumer
value enhancement’ to be £[] per customer. We have not seen any
evidence to support this assumption. Customers would still need to
purchase CoD on Nintendo, so there is no obvious pricing benefit. To the
extent that Nintendo console owners would benefit from increased choice
due to the availability of CoD, it is not clear why this would have a £[]
average annual value.

(b) Microsoft has also assumed a []% penetration rate of CoD on Nintendo.
We have seen limited evidence to support this assumption. The fact that
Activision does not consider that [] suggests that there is limited
demand for CoD titles on the current Nintendo Switch. As such, the extent
to which providing CoD on Nintendo Switch would benefit consumers is
likely to be limited.

(c) The agreement has a ten-year term. We consider that account must be
made for the inherent uncertainty over the scale of the benefits over time.
The likelihood of the assumptions remaining constant will decrease over
time, and the changing nature of the console market makes it harder to
make any predictions about how a particular retail product (for example,
CoD []) might evolve, or how customer behaviour might develop. As
such, it is not a reasonable assumption that any annual benefit would
remain constant over the term of the agreement. Instead, in our view the

1552 Microsoft Annex to the response to the Remedies Notice.

391
 benefits are likely to decline significantly (and certainly at a rate greater
than the ‘risk-free’ rate used in Microsoft’s NPV calculation) over time,
reducing our confidence that they can be expected to be realised.

11.212 In summary, we find that the claimed Nintendo RCB does not qualify as
an RCB because it does not satisfy the condition in section 30(3)(a) of the
Act. Even if the claimed Nintendo RCB were to qualify as an RCB, we
consider that any expected benefit arising would likely be limited and would
not have a material impact on the costs of the prohibition remedy.

• The claimed Mobile Gaming RCB

11.213 Microsoft submitted that, as a result of the Merger, it will expand into
mobile gaming and ‘challenge the existing duopoly over mobile app
distribution’. It submitted that the Merger will result in benefits to customers in
mobile game distribution in the form of lower prices, higher quality, greater
choice and greater innovation. It submitted that benefits apply to end
consumers and developers of native mobile games, each of which are
relevant customers of mobile app stores. 1553 Microsoft estimated this claimed
RCB to have a value of £[] in the UK. 1554

o Views of the Parties

11.214 Microsoft submitted that while the Remedies Working Paper dismissed
this claimed RCB as being too speculative to occur within a reasonable
period: 1555

(a) Phil Spencer (Xbox’s CEO), in a recent interview with the Financial Times,
again confirmed Microsoft’s intention to launch a new mobile game
distribution platform ‘as soon as next year’. Microsoft told us that this was
consistent with Microsoft’s deal model launching an Xbox model platform
[] and provided further evidence of Microsoft’s intention to pursue this
within a ‘reasonable period’.

(b) The Remedies Working Paper failed to acknowledge that the regulatory
environment would further facilitate Xbox’s entry, and added that
implementation of the EU’s Digital Market Act in March 2024 (as Phil
Spencer noted in his interview) was expected to require Apple and
Google to open up their mobile app stores.

1553 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 2.10.
1554 Microsoft Annex [] to the response to the Remedies Notice.
1555 Microsoft response to the Remedies Working Paper.

392
 (c) As with the claimed Nintendo RCB, the CMA should be consistent in its
application of the relevant time periods when assessing the potential
benefits of the claimed Mobile Gaming RCB. Microsoft submitted that if
the CMA was willing to apply longer time periods as part of its analysis in
relation to the counterfactual and the growth of cloud gaming, these
should apply equally to the claimed Mobile Gaming RCB.

11.215 Microsoft also submitted in its response to our Remedies Working

Paper that the Remedies Working Paper, by focusing on the possibility of
alternative acquisitions, failed to acknowledge the ‘significant strength’ that
Activision currently had in the mobile gaming sector. 1556

11.216 Activision noted that it could not imagine a situation where []. 1557

o Our assessment of the claimed Mobile Gaming RCB

11.217 We have considered whether the claimed Mobile Gaming RCB

qualifies as an RCB under the Act.

11.218 First, we have considered whether the claimed Mobile Gaming RCB
satisfies the condition in section 30(1)(a) of the Act. To the extent that any
benefit arises (which is considered below), this would be in the form of greater
choice of goods or services in the market for mobile gaming services in the
UK, and would therefore satisfy the requirement under section 30(1)(a) of the
Act.

11.219 Second, we have considered whether the claimed Mobile Gaming RCB
may be expected to accrue within a reasonable period as a result of the
creation of the RMS, as required under section 30(3)(a) of the Act.

11.220 Microsoft’s deal model suggests that it would launch an Xbox mobile
platform []. 1558 However, Microsoft also acknowledges there is no
guarantee of this platform succeeding, 1559 and the timing for Xbox to launch
its mobile platform is [] to the timing for any associated RCB to accrue.
While we have not seen evidence of technical barriers to creating such a
platform within this timescale, we understand that Apple and Google currently
control access to their platforms from third-party app stores, and they either
currently prohibit rival mobile gaming app stores or impose strict limits on their
ability to monetise content. We also note that Microsoft said that its plans

1556 Microsoft response to the Remedies Working Paper.

1557 Activision response hearing transcript.
1558 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 2.10.
1559 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 2.10.

393
 []. 1560 As a result, we cannot have confidence that benefits from the claimed
Mobile Gaming RCB can be expected to accrue within a reasonable period or
even at all.

11.221 In its response to the Remedies Notice, Microsoft submitted that the
process of dynamic competition had economic value in the present, and that
its entry attempt constitutes an RCB. Microsoft cited CMA guidance, saying
that in response to its entry attempt, existing providers such as Google and
Apple may ‘invest in order to protect future sales from dynamic competitors,
and the removal of the threat of entry may lead to a significant reduction in
innovation or efforts from other firms’. 1561

11.222 We agree with Microsoft’s view that its entry plans, if credible, might
trigger a pro-competitive response from incumbent firms which, in principle,
might give rise to a consumer benefit. However, the Parties have not provided
any evidence to the CMA to substantiate these claims. Our Merger Remedies
Guidance is clear that merger parties will be expected to provide convincing
evidence regarding the nature and scale of RCBs that they claim to result
from the merger and to demonstrate that these fall within the Act’s definition of
such benefits. 1562 In our view, the claimed Mobile Gaming RCB is too
speculative to qualify as an RCB.

11.223 Third – although not strictly necessary given our conclusion that the
claimed Mobile Gaming RCB does not meet the condition in section 30(3)(a) –
we have considered whether the claimed benefit was unlikely to accrue
without the RMS or a similar lessening of competition, so as to meet the
condition in section 30(3)(b) of the Act.

11.224 We note Microsoft’s submission that Activision has significant strength

in mobile gaming, and consider that the presence of Activision’s games on
any mobile gaming store would enhance its competitiveness. However, we
also consider that this could be achieved by less anti-competitive means than
the Merger, and Microsoft could acquire ‘attractive content and experience
with player engagement and acquisition’ 1563 by buying a different mobile
games publisher. This appears to have been Microsoft’s strategy – it
attempted to buy [] in [], and said []. 1564

1560 Microsoft response to the Issues Statement, 31 October 2022, paragraph 2.12
1561 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 2.10, citing CMA129, paragraph
5.20.
1562 CMA87, paragraph 3.20.
1563 Microsoft response to the Provisional Findings, 2 March 2023, paragraph 4.20(a)
1564 Microsoft Main Party Hearing transcript.

394
11.225 We also consider that to launch a competitive mobile platform,
Microsoft would need a significant quantity and variety of games. This would
be likely to involve making agreements with third party publishers in a similar
way as it does currently with console and PC games; Microsoft has not
provided evidence that its entry attempt would be sufficient relying on
Activision’s games alone. For this reason, we disagree with Activision’s
contention that a competing mobile games store could only be achieved if the
content was from a single organisation. We therefore do not consider that any
such benefit would be unlikely to accrue absent the RMS or a similar situation.
We therefore do not consider that the condition in section 30(3)(b) of the Act
is satisfied for the claimed Mobile Gaming RCB.

11.226 In summary, we find that the claimed Mobile Gaming RCB does not
qualify as an RCB because it does not satisfy the conditions in sections
30(3)(a) and (b) of the Act.

• The claimed Game Pass RCB

11.227 Microsoft submitted that it plans to make Activision content available

day and date in Game Pass, which will increase customer choice and lower
the cost of access.

o Views of the Parties

11.228 Microsoft submitted that Activision would not place new releases in
subscription services day and date (referring to the Provisional Findings
counterfactual) []. Microsoft therefore told us that it considers any benefit
arising from Activision content being available on Game Pass to be Merger-
specific. Microsoft told us that it will make future Activision releases available
on Game Pass on the day of release, and that this will benefit consumers. 1565

11.229 Microsoft submitted that the addition of Activision content to Game

Pass will, in effect, reduce prices, as customers will receive a broader
catalogue of games for the same price – ie the quality adjusted price of Game
Pass would fall post-Merger. Microsoft estimated the benefit of bringing day
and date Activision content to Game Pass to have a net present value over
ten years of around £[] million for UK Game Pass subscribers on Xbox and
around £[] million for UK Game Pass subscribers on PC. 1566 [], the
benefits are estimated to be £[] billion for UK Xbox customers and £[]

1565 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 2.4(a)-(d).
1566 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 2.4(e).

395
 million for UK PC customers. Globally, the claimed benefits are £[]billion
and £[] billion respectively. 1567

11.230 Microsoft submitted that ‘[], and certainly will not increase to a point
that offsets the substantial benefits of Activision titles coming to Game Pass
on a day and date basis’. 1568 Further, it submitted that inclusion of Activision
content on Game Pass will spur SIE to invest in its subscription offering. 1569

11.231 Microsoft submitted that the availability of existing CoD exclusive

content on Xbox and PC will result in a higher quality of goods and services.
Microsoft submitted that CoD is not currently available on equal terms on
PlayStation, Xbox and PC due to the agreements that SIE has in place with
Activision. It noted that SIE has CoD content and timed exclusives and that
following the Merger these will be available to Xbox and PC gamers. 1570

11.232 In its response to our Remedies Working Paper, Microsoft told us that it
had refined its estimate of the claimed Game Pass RCB following the
Remedies Notice and subsequent discussions with the CMA, and estimated
that the RCB per year was around £[] worldwide and £[] in the UK. It
added that limiting this RCB to a yearly figure did not accurately reflect the
‘true value to customers’ that might accrue over a ten-year period, and that
these were ‘significant and expected to generate benefits of approximately
£[] billion worldwide and £[] million in the UK’. 1571 It added that
Microsoft’s refined estimated continued to show a ‘substantial benefit’ from
the Merger that would be eliminated with a prohibition remedy, but which
would preserved under the Microsoft Cloud Remedy. 1572

o Our assessment of the claimed Game Pass RCB

11.233 We have considered whether the claimed Game Pass RCB qualifies as
an RCB under the Act.

11.234 First, we have considered whether the claimed Game Pass RCB
satisfies the condition in section 30(1)(a) of the Act. Microsoft is proposing to
make Activision content available day and date in Game Pass. As such, to the
extent that any benefit arises (which is considered below), this would be in the
form of higher quality or greater choice of goods or services in the markets for

1567 Microsoft, Annex 1 to the response to the Remedies Notice.

1568 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 2.4(f).
1569 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 2.4(g).
1570 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 2.6.
1571 Microsoft response to the Remedies Working Paper.
1572 Microsoft response to the Remedies Working Paper.

396
 console gaming services and cloud gaming services in the UK, and would
therefore satisfy the requirement under section 30(1)(a) of the Act.

11.235 Second, we have considered whether the claimed Game Pass RCB
may be expected to accrue within a reasonable period as a result of the
creation of the RMS, as required under section 30(3)(a) of the Act.

11.236 We considered whether the claimed benefits would accrue from the
creation of the RMS. The Merger brings Microsoft’s Game Pass product under
common ownership with Activision’s content, creating an incentive to exploit
synergies between the two. This is a form of elimination of double
marginalisation, that we consider would arise as a result of the Microsoft and
Activision businesses ceasing to be distinct. We therefore consider that any
benefits arising from the claimed Game Pass RCB would accrue from the
creation of the RMS. It is plausible that to the extent Activision content is
added to Game Pass, the price of Game Pass subscriptions may increase
commensurately and therefore the purported benefit from the Merger may not
ultimately accrue to UK consumers to the extent claimed, or even at all. We
consider this potential outcome below in our assessment of the likely scale of
the claimed Game Pass RCB.

11.237 Microsoft submitted that []. 1573 The benefits from other Activision
games could accrue earlier than this, []. We also note that the benefit from
other games has not been included in Microsoft’s model.

11.238 On this basis, we would expect that the benefits would start to accrue
within a reasonable period (although there would, in practice, likely be some
delay between completion of the Merger and CoD arriving on Game Pass in
2025). We therefore find that at least some benefit from the claimed Game
Pass RCB may be expected to accrue within a reasonable period, such that
the condition in section 30(3)(a) of the Act is satisfied.

11.239 Microsoft has assumed that the benefits from the claimed Game Pass
RCB would continue forever. Our view is that this is not likely in a dynamic
and evolving market, which would decrease the likelihood over time that these
benefits would accrue. This has been taken into account in the assessment of
the scale of the RCB below.

11.240 We also considered Microsoft’s contention at paragraph 11.230 that

the inclusion of Activision content in Game Pass would spur SIE to invest in

1573 Microsoft, Annex 1 to the response to the Remedies Notice.

397
 its subscription offering to qualify as an RCB. SIE told us that it would []. 1574
Based on the available evidence, we cannot predict what SIE’s reaction would
be with sufficient confidence or accuracy to be able to take any benefit into
account as an RCB.

11.241 Third, we have considered whether the claimed benefit was unlikely to
accrue without the RMS or a similar lessening of competition, so as to meet
the condition in section 30(3)(b) of the Act. In particular, we considered the
likelihood of Microsoft reaching a commercial agreement with an independent
Activision to put its content on Game Pass, unlocking the wider benefits to
Microsoft such as greater uptake of Game Pass.

11.242 Activision’s stated and past policy was not to put its titles on
subscription services on a day and date basis. In Chapter 7, we explain that
‘we consider it unlikely that Activision would make its most valuable games –
such as CoD – available on MGS services on the date of release in the
foreseeable future absent the Merger. We believe the evidence indicates,
however, that Activision would be likely to place increasingly valuable parts of
its gaming catalogue on MGS services as these services continue to grow.
This would be likely to include back-catalogue games, as well as Activision’s
latest releases, although some time after they are released (ie, not on the
date of release)’. 1575

11.243 In line with this assessment, we consider that in order for Activision to
change its policy, it would need a sufficient share of the value that Microsoft
would attach to acquiring rights to add Activision content to its MGS to
compensate it for the lost revenue from []. In principle, we consider this
agreement could be possible to construct – it is a similar negotiation to the
one Microsoft had with Activision’s shareholders when it made its offer to buy
Activision. 1576 However, we note that structuring such a deal as a commercial
contract with Activision, rather than the outright purchase of Activision,
introduces additional risks relating to each counterparty, specification of the
terms of the contract and potential circumvention by one or other of the
parties. Given the likely commercial value of the agreement and the premium
that would need to be paid to Activision, we consider these risks are likely to
be material. We therefore consider, on balance, that such an agreement
would be unlikely to occur absent the Merger. We have not seen any

1575Chapter 7, paragraph 7.132.

1576Microsoft’s valuation of Activision, including the benefits it could achieve after the Merger, was $[] billion.
This represented a premium of $[] billion over Activision’s standalone value of $[] billion. Microsoft shared
these benefits with Activision’s shareholders by paying $[] billion of the $[] billion benefits over to Activision’s
shareholders, resulting in a purchase price of $[] billion.

398
 evidence to suggest that this claimed RCB would be likely to accrue as a
result of any other less anti-competitive arrangement, such as an alternative
acquisition.

11.244 We therefore find that at least some benefit from the claimed Game
Pass RCB – insofar as it relates to Activision content being made available
day and date on Game Pass – satisfies the condition in section 30(3)(b) of the
Act.

11.245 In summary, we have found that at least some of the benefit of the
claimed Game Pass RCB qualifies as an RCB under the Act. The claimed
Game Pass RCB may be expected to result in higher quality and greater
choice of goods or services in the markets for console gaming services in the
UK. To the extent that Game Pass customers utilise cloud streaming, there
would also be a benefit arising in the market for cloud gaming services in the
UK. We have found that at least some of the benefit of the claimed Game
Pass RCB may be expected to accrue within a reasonable period as a result
of the creation of the RMS. We also believe, on balance, that this benefit is
unlikely to accrue without the creation of the RMS or a similar lessening of
competition.

o Likely scale of the claimed Game Pass RCB

11.246 We have considered the likely scale of the claimed Game Pass RCB.
Microsoft provided an analytical model (the Game Pass model) setting out
the likely benefits. In this section we summarise the model and our view of it.

11.247 The Game Pass model splits Microsoft’s customers into different
cohorts based on current purchasing behaviour, with separate calculations for
Xbox and PC customers. The cohorts are:

(a) [];

(b) [];

(c) [];

(d) [];

(e) [].

11.248 The Game Pass model assumes that, []:

(a) [];

(b) [];

399
 (c) [];

(d) [];

(e) [].

11.249 Microsoft submitted that each of these cohorts would benefit in different
ways. Multiplying the number of customers in each cohort that would change
their behaviour by the estimated benefit gives an annual benefit. For all
cohorts this was estimated to be £[] million per year on Xbox and £[]
million per year on PC.

11.250 We considered the extent to which all of these benefits were likely to
arise. We found that there are uncertainties regarding the likelihood of these
benefits being realised due to three factors:

(a) assumptions in the model;

(b) potential post-Merger actions by Microsoft; and

(c) general uncertainty that the future turns out differently to that predicted in
the model.

11.251 We considered the assumptions in the model. Predicting and modelling

future behavioural changes in response to a hypothetical action is inherently
difficult, and any model would require a number of simplifying assumptions.
The Game Pass model estimates the proportion of gamers in each cohort that
would change their behaviour. However, these proportions are only a ‘best
estimate’, without detailed supporting evidence and may be overstating the
benefit. 1577 In addition, []. Furthermore, the model []. Again, we have not
seen detailed evidence to support this figure, which appears to be inherently
uncertain.

11.252 We also considered whether actions by Microsoft after the Merger

completes would reduce the scale of the benefits. In particular, the Game
Pass model assumes [].

11.253 We note that part of Microsoft’s stated rationale for the Merger was to
bring Activision games to Game Pass on a day and date basis, and also that
an increase in the price of a product is generally likely to lead to a reduction in
demand. While a reduction in demand serves as a disincentive to increase
prices, the higher earnings on customers that do not switch away serves as

1577For example, the Game Pass model assumes that []% of [] would start playing CoD (and therefore gain
a benefit from it) as a result of if appearing on Game Pass. While this may be a plausible figure, other lower
figures may also be plausible.

400
 an incentive acting in the opposite direction. Pre-Merger, we would expect
those disincentives and incentives to be balanced. However, post-Merger,
adding content of the importance of Call of Duty to Game Pass introduces a
new economic cost (ie an opportunity cost) associated with the supply of
Game Pass to Microsoft customers, because – in line with the Parties’
submissions – []. 1578 A higher economic cost of a product will therefore, all
else being equal, tend to induce a profit-maximising firm to increase its price.
In addition, alongside this increase in the economic cost of providing Game
Pass, the quality of the Game Pass subscription would be higher which will,
all else being equal, tend to increase the willingness of subscribers to pay a
higher price and reduce the proportion that would switch away in response to
any given price increase. Consequently, these factors will have the tendency
to increase the profit-maximising price of Game Pass in the post-Merger
scenario relative to the pre-Merger scenario, whether for all subscribers or for
a subset of subscribers through differentiated pricing.

11.254 Microsoft itself has indicated that it expects to [] (Microsoft said that
any []). 1579 More broadly, we would expect Microsoft to behave in a way
that maximises its returns on the $69 billion it paid for Activision, and therefore
to act on the incentives described in the previous paragraph. Microsoft’s
[]would reduce the value of the relevant customer benefit relative to the
Parties’ estimates. Any [] price increases accounting for the addition of
Activision content to Game Pass—in line with the incentives described
above—would further reduce that benefit relative to the Parties’ estimates.

11.255 We also note that the post-Merger constraint from B2P would be likely
to change as Microsoft would set both the wholesale B2P and subscription
prices for Activision content, allowing it to have some control over the extent
to which B2P will constrain the price of Game Pass. While Microsoft could
potentially choose to forego otherwise profitable price increases in the short
run in order to establish a stronger position of market power in cloud gaming
services, the incentive to price Game Pass more highly post-Merger in the
medium term would remain.

11.256 Even a small price increase would have a significant effect on the scale
of benefits. This is both because the value to customers of switching to Game
Pass from B2P would decrease, and also because those customers in []
that do not play CoD would suffer a price rise without any corresponding
benefit. For example, taking the benefits figures claimed by Microsoft as a
starting point, a £1 per month or 9% increase in the cost of Game Pass would

1578 Microsoft response to the Remedies Notice, 22 February 2023.

1579 Microsoft response hearing transcript.

401
 reduce the annual benefit for Xbox customers to £[] million, and reduce the
annual benefit for PC customers to £[] million. 1580 These figures assume no
loss of customers, which would further reduce the size of the benefit.

11.257 In its response to the Remedies Notice, Microsoft said ‘[]’. 1581
However, Microsoft also said that []. In the response hearing, it said [] 1582
[]. 1583 It said further that [].’ 1584

11.258 As a result, our view is that, given the economic costs arising from
putting Activision games on Game Pass, [] and the likely future desire to
make a return on the acquisition of Activision, the assumption in the Parties’
analysis that CoD would be added to Game Pass and that prices would not
increase from their current level is unlikely to be long lasting, which leads the
Parties’ analysis of RCBs to significantly overstate the value of those benefits
to consumers.

11.259 We also consider that account must be made for the inherent
uncertainty over the scale of the benefits over time. The likelihood of the
assumptions (for example, as discussed above, that prices will not rise)
remaining valid will decrease over time, and the dynamic and evolving nature
of the market makes it harder to make any predictions about how a particular
retail product (for example, CoD or Game Pass) might evolve, or how
customer behaviour might develop. Other factors include a change in the
relative competitiveness of Microsoft’s current or future consoles, a particular
release of CoD being less well-received, the relative competitiveness of CoD
with other games, or issues beyond Microsoft’s control. These are some
examples of how the scale of benefits may change over time, rather than an
exhaustive list.

11.260 The extent of these uncertainties increases over time. In its original
submission in response to the Remedies Notice, Microsoft claimed benefits
for ten years from completion of the Merger. In its subsequent response to the
Provisional Findings it added a “terminal value” – the present value of the sum
of the benefits after year ten in perpetuity. This terminal value increases the
total present value of the benefits threefold.

1580 A price increase of £12 per year is applied as: (a) Xbox customers – for [], the price increase is applied to
the number of subscribers that will switch, and for [], it is applied to the total number of subscribers; (b) For
PC customers – for [], the price increase is applied to the number of subscribers that will switch, and for
[], it is applied to the total number of subscribers. There is no [] for PC customers.
1581 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 2.4(f).
1582 Microsoft response hearing transcript.
1583 Microsoft response hearing transcript.
1584 Microsoft response hearing transcript.

402
11.261 The benefits of the claimed Game Pass RCB might, in theory, endure
beyond ten years. However, given the uncertainties referred to above, we
cannot have any confidence that in practice these benefits can be expected to
accrue after ten years. As a result, we consider that the terminal value
element of the claimed Game Pass RCB should not be given material weight
in our consideration of the expected scale of this RCB. Furthermore, the
extent of these uncertainties increases over time, reducing our confidence as
to the total scale of the benefits may be expected to be realised.

11.262 We considered how we should reflect the uncertainties summarised

above in our assessment of the scale of benefits. We considered applying a
discount rate and discounting the annual benefits by greater amounts over
time. The Game Pass model adopts this approach, using a ‘risk-free’ discount
rate based on ten-year UK Gilts. We also considered trying to estimate a
discount rate based on a consumer’s weighted average cost of capital.
However, this is a complex exercise which would involve sensitive
assumptions, and would risk suggesting a degree of spurious accuracy on our
assessment. This approach would also not capture the uncertainties arising
from Microsoft’s behaviour, or caused by the assumptions in the Game Pass
model.

11.263 Estimation of the likely scale of the claimed Game Pass RCB will
inevitably involve a degree of judgement. However, we considered it possible
to put some boundaries on the scale of the claimed Game Pass RCB in order
to support our consideration of proportionality below. Bearing this in mind, and
having regard to the uncertainties concerning realisation of the claimed
benefits set out in paragraphs 11.250 to 11.260 above, we consider that the
claimed Game Pass RCB is likely to be materially below Microsoft’s claim of
£[] million per year from 2025, and likely to decline over time, such that we
do not place material weight on any claimed benefits after ten years.

Conclusion on RCBs

11.264 We have found that the following benefits claimed by Microsoft do not
constitute RCBs (and, even if they did meet the necessary conditions, would
be small and would not have a material impact on the costs of the prohibition
remedy):

(a) The claimed Cloud Gaming RCB;

(b) The claimed Nintendo RCB; and

(c) The claimed Mobile Gaming RCB.

403
11.265 We have found that the claimed Game Pass RCB qualifies as an RCB,
and can be expected to have annual benefits materially below the £[]
million per year figures presented by Microsoft. We have also found that the
benefits are likely to decline over time. We have not found it necessary (or
possible) to estimate the precise amount of the benefit that may be expected
to accrue from the claimed Game Pass RCB, given the inherent uncertainties
with this exercise. This is in any case not necessary for us to conduct the
proportionality assessment as set out below. We have also found that even if
the claimed Cloud Gaming RCB and claimed Nintendo RCB did constitute
RCBs, the scale of the benefits would not be material to our assessment.

Proportionality assessment

11.266 In paragraphs 11.133 and 11.134, we summarised our conclusions on

which remedies would be effective in addressing the SLC and the resulting
adverse effects. We concluded that prohibition of the Merger would be
effective, while the Microsoft Cloud Remedy would not be effective. While
partial divestiture might in principle be an effective remedy, the Parties have
stated that they would not proceed on this basis and not provided us with the
information we would need to reach a conclusion on this.

11.267 We set out below our assessment of, and conclusions on, the
proportionality of prohibition as a remedy.

Framework for the assessment of proportionality of merger remedies

11.268 The CMA will seek to select the least costly remedy, or package of
remedies, of those remedy options that it considers will be effective (we call
this the ‘least onerous effective remedy’). In addition, the CMA will seek to
ensure that no remedy is disproportionate in relation to the SLC and its
adverse effects. 1585

11.269 In this case, as explained above, our view is that only prohibition would
be an effective remedy. Accordingly, we are not choosing between multiple
remedies that we consider will be effective, and our proportionality
assessment is therefore focused on considering whether this remedy would
be disproportionate in relation to the SLC and its adverse effects. In doing so,
we compare the extent of harm associated with the SLC with the relevant
costs of the proposed remedy. 1586

1585 CMA87, paragraph 3.6.

1586 CMA87, paragraph 3.6.

404
11.270 We first consider whether there are any relevant costs associated with
the effective remedy option identified. When considering relevant costs, the
CMA’s considerations may include (but are not limited to): 1587

(a) distortions in market outcomes;

(b) compliance and monitoring costs incurred by the Parties, third parties, or
the CMA; and

(c) the loss of any RCBs that may accrue from the Merger which are
foregone as a result of the remedy (see paragraphs 11.137 and 11.138
above).

11.271 However, the CMA will generally attribute less significance to the costs
of a remedy that will be incurred by the merger parties than the costs that will
be imposed by a remedy on third parties, the CMA or other monitoring
agencies. 1588 The merger parties have the choice of whether or not to enter
into a merger agreement, and on what terms. It is for the merger parties to
assess whether there is a risk that the merger may be subject to an SLC
finding and prohibited – any costs for the merger parties resulting from this
outcome are, in essence, avoidable.

Views of the Parties and third parties

Views of the Parties

11.272 Microsoft told us that prohibition would result in the loss of RCBs, risk
raising concerns around extraterritoriality, and would be disproportionate.
Microsoft told us that prohibition would be ‘taking us in the opposite direction’
to the goal of ‘trying to increase customer access, customer choice, on
competition in the market’. 1589 Similarly, Activision told us that it disagreed
with prohibition as a remedy and considered that it would be disproportionate
and unnecessary, highlighting that RCBs would be lost as a result. 1590

11.273 Microsoft told us that to the extent that the CMA has concerns about
the impact of the Merger on cloud gaming services as a nascent technology, it
is not a proportionate response to default to prohibiting the Merger. Microsoft
submitted that not only would this require reliance on an ‘insufficient body of
evidence, it would also forego the benefits which it considers to be certain to
arise from the Merger.’ It told us that prohibiting the Merger on this basis

1587 CMA87, paragraph 3.10.

1588 CMA87, paragraph 3.8.
1589 Microsoft response hearing transcript.
1590 Activision response hearing transcript.

405
 would be tantamount to lowering the evidentiary bar required to block a
transaction because it is in a digital market. Microsoft submitted that the UK
Government has recently concluded that lowering the evidentiary bar in phase
2 is not appropriate – even for digital mergers involving companies with
‘Strategic Market Status’ – due to the potential impacts on innovation and
investment. Microsoft submitted that refusing to consider behavioural
remedies in new digital markets simply because such markets are ‘dynamic’
or ‘evolving’ has the same effect, making it harder to obtain merger
clearances in digital mergers. Microsoft submitted that the UK Government
has been clear that the new merger control rules overseen by the CMA’s
DMU must avoid disproportionate burdens on businesses, and that the CMA
can only hope to achieve this aim if it is prepared to impose behavioural
remedies in dynamic and evolving digital markets. 1591

11.274 Microsoft also told us that prohibition would be disproportionate to the

size of the Cloud Gaming SLC identified. It submitted that the cloud gaming
market is de minimis and will remain so. On this basis, it considers prohibition
to be a disproportionate remedy and considers that a divestment remedy will
deliver minimal benefits. 1592 It told us that the ‘supposed adverse effects’ were
therefore ‘vastly overstated in the Remedies Working Paper’, and that these
adverse effects were ‘not only remote and of limited magnitude' but even
those were uncertain. 1593

11.275 In its response to our Remedies Working Paper, Microsoft told us that
the CMA had ‘substantially understated the costs of prohibition’, and that
other than the RCBs, prohibition would have ‘huge collateral effects on the
way in which Microsoft is able to carry on its business as it wishes to outside
the realm of cloud gaming’. It added that prohibition would limit Microsoft as to
how it would be able to operate in all non-cloud gaming activities where it
would be able to use Activision content to pursue its strategic goals. It told us
that as the Addendum to the Provisional Findings recognised, there was no
basis for competition concerns in relation to these activities and yet Microsoft
would be precluded from pursuing legitimate business opportunities. It
therefore told us that the proportionality concern could not simply be seen
through the prism of provable RCBs: the impact on Microsoft’s business
freedom was ‘very extensive and not justifiable’. 1594

1591 Microsoft response to the Remedies Working Paper. Microsoft also referenced the Furman Report, which

notes that a “presumption against all acquisitions by large digital companies is not a proportionate response to
the challenges posed by the digital economy” (Furman Report, March 2019, paragraph 3.103).
1592 Microsoft response to the Remedies Notice, 22 February 2023, paragraph 4.10(a).
1593 Microsoft response to the Remedies Working Paper.
1594 Microsoft response to the Remedies Working Paper.

406
11.276 In addition, Microsoft submitted that the legal framework and
jurisdiction of the CMA should be considered in assessing the proportionality
of any remedy. Microsoft submitted that the language of the statute does not
require full prevention of any SLC identified and instead recognises that
mitigation may be sufficient. It told us that to exclude taking action which is
reasonable and practicable to ‘mitigate’ the SLC or any adverse effects would
be wrong. Microsoft referred to the Ryanair judgment, as referred to in
footnote 13521352 above, and submitted that the judgment does not suggest
that mitigation was no part of the statutory assessment test. 1595 Microsoft
submitted that the view that a remedy is only effective if it ‘fully remedies or
prevents’ the SLC and its adverse effects (not just mitigating them) is wrong in
law. It told us that the reference to a remedy being effective is a shorthand for
the remedy fulfilling the statutory criteria, which, it submitted, include
mitigation. 1596

11.277 Microsoft also submitted that the CMA need only ‘have regard’ to
achieving as comprehensive solution to the SLC and any adverse effects
arising as is ‘reasonable and practicable’, and that the CMA is therefore not
bound by an obligation to fulfil the requirement. 1597 It told us that this is not an
‘absolute obligation’ but is conditioned by the qualification that the solution
should be as comprehensive as is ‘reasonable and practicable’, which it told
us also means proportionate. Microsoft submitted that the reference to an
effective remedy only being one which is comprehensive – as set out at
paragraph 11.12 above – is wrong in law. 1598 On this basis, Microsoft
submitted that so long as the CMA has regard to achieving as comprehensive
a solution as is reasonable and practicable, it is open to it to discharge its
obligations through a remedy which mitigates the effects of any SLC and
adverse effects. It told us that even if the CMA were to have concerns that a
behavioural remedy might not necessarily deal with every hypothetical
concern, it would still be a legitimate exercise of the CMA’s functions since it
would (at least) be as comprehensive as was reasonable and practicable and
would substantially mitigate any SLC and adverse effects. Microsoft noted
that the assessment of what is reasonable and practicable would need to take
into account the proportionality of any remedy. 1599

11.278 Microsoft submitted that the extra-territorial nature of any remedy

should be taken into account, noting that the greater the effects of any remedy
outside of the UK, the ‘less reasonable proportionate and practicable’ the

1595 Microsoft response to the Remedies Working Paper.

1596 Microsoft response to the Remedies Working Paper.
1597 Microsoft response to the Remedies Notice, 22 February 2023, paragraphs 5.2.
1598 Microsoft response to the Remedies Working Paper.
1599 Microsoft response to the Remedies Working Paper.

407
 remedy will be.’ 1600 Microsoft also told us that worldwide prohibition would be
disproportionate because it would have an extra-territorial impact. 1601

11.279 In its response to our Remedies Working Paper, Microsoft told us that
any prohibition remedy would not be limited to the UK and would undermine
the deal worldwide, and as such ‘the effect of the CMA’s conclusion in relation
to a tiny part of the gaming world (cloud gaming) would have a grossly
disproportionate and extensively extra-territorial effect’, including ‘denying a
non-UK company, Microsoft, the freedom to conduct its business as it wishes
outside the UK (denying substantial benefits to gamers outside the UK,
estimated to amount to at least USD [] billion)’. 1602

11.280 Microsoft told us that it was a well-accepted principle of English law

that every statute is interpreted ‘so far as its language permits, so as not to be
inconsistent with the comity of nations or the established rules of international
law’. 1603 It added that whilst it was ‘recognised that Parliament in the
Enterprise Act 2002 did not preclude the CMA from adopting remedies which
may have impacts outside the UK, equally it did not permit the CMA simply to
ignore the extra-territorial consequences of its actions’. It told us that in the
assessment of the proportionality of any remedy, the CMA must recognise
that having extra-territorial effects is an ‘undesirable outcome’, and that to do
otherwise would be to apply the Act in an ‘aggressively’ extra-territorial
manner which would be unjustified. 1604 Microsoft told us that the Merger had
already been approved unconditionally in several jurisdictions while the
European Commission was engaging extensively with a behavioural remedy.
It added that by imposing a prohibition remedy, the CMA would in practice be
cutting across the regulatory assessments and remedial solutions of other
regulators in other jurisdictions. 1605

11.281 Activision told us that this is a case where a behavioural remedy would
be workable and there is no practicable divestiture solution. It submitted that it
would be disproportionate to discard a behavioural remedy in favour of
prohibition or divestiture. 1606

1600 Microsoft response to the Remedies Notice, 22 February 2023, paragraphs 5.4-5.5.
1601 Microsoft response hearing transcript, and Microsoft response to the Remedies Working Paper.
1602 Microsoft response to the Remedies Working Paper.
1603 R (Al-Skeini) v. Secretary of State for Defence [2007] UKHL 26 per Lord Rodger at [45]. Source: Microsoft

response to the Remedies Working Paper.

1604 Microsoft response to the Remedies Working Paper.
1605 Microsoft response to the Remedies Working Paper.
1606 Activision response hearing transcript.

408
 Views of third parties

11.282 As already discussed at paragraphs 11.33 to 11.35 above, some of the

third parties we heard from ([], [], [], [], [] and []) told us that
they viewed prohibition as excessive. 1607 Some noted that prohibition would
be disproportionate if there were alternative, effective remedies available,
while others noted that they did not consider there to be an SLC in the first
place. Other third parties ([] and SIE) told us that prohibition of the Merger
would be a proportionate remedy. 1608

Our assessment of proportionality

11.283 We have considered whether prohibition of the Merger would be

disproportionate to the SLC and its adverse effects.

The harm arising from the SLC

11.284 We have found that the scale of harm to consumers likely to arise from
the Merger and the resulting SLC would be substantial. The SLC arises from
a major, permanent structural change in the market, and as such cannot be
expected to fall away over time. Harm arising from the SLC is therefore likely
to endure. Moreover, as demand for cloud gaming is growing rapidly in the
UK, reflecting increasing global take-up, we would expect the harm from SLC
to similarly continue to grow over time.

11.285 We have found that Activision’s titles are likely to be an important input
for the success of cloud gaming services. Microsoft already holds a strong
position in this growing market by virtue of its high market share of [60-70]%,
unparalleled advantages through its ownership of Windows OS, its cloud
infrastructure, and its catalogue of first party titles. 1609 There are a few
emerging rivals with their own respective strengths, such as Amazon, SIE,
and NVIDIA, but none are currently as well positioned as Microsoft. In a
concentrated market, harm to smaller rivals is likely to constitute harm to
competition, unless there is some way to avoid that harm. 1610 We have also
found significant barriers to entry and expansion in cloud gaming, increasing
the adverse effect on competition.

1607 Third party responses to the remedies questionnaire: ([]; []; []; []; []; and []. See also [] call
note.
1608 [] response to the remedies questionnaire, and SIE response to the Remedies Notice, 22 February 2023,

paragraph 9.
1609 Microsoft’s strengths in cloud gaming are discussed further in paragraphs 8.96 to 8.224.
1610 CMA129, paragraph 7.21.

409
11.286 By distorting and stifling competition in the growing and dynamic
market for cloud gaming services, the Merger could alter the future of gaming
for the worse. UK cloud gaming monthly active users more than tripled from
the start of 2021 to the end of 2022. The market for cloud gaming seems,
based on the evidence, poised to grow rapidly and significantly. Market
reports suggest that the UK market will be worth $0.6 to $1.1 billion (£0.5 to
£0.9 billion) by 2025, growing to $1.2 billion to $1.3 billion (£1 to £1.1 billion)
by 2026. 1611 These forecasts also suggest continued growth beyond the short-
to medium-term. Correspondingly, this means that the harm arising from the
SLC will continue to increase over time.

11.287 Furthermore, cloud streaming is an increasingly important conduit for

playing games, both for new users who are unable or unwilling to buy an
expensive console solely for the purposes of gaming, and for existing gamers
looking for an alternative to consoles. Absent the Merger, strong competition
in this market could make cloud gaming better and more affordable for
consumers. By contrast, we are of the view that the Merger would make
Microsoft – which is an already strong incumbent in this market – even
stronger, retaining a big share of the market and facing limited competition
from current and potential rivals.

11.288 This reduction in competition is likely to substantially harm consumers

in the long run. The evidence we found suggests that cloud gaming could be
transformative for the gaming industry in the next few years, helping to reach
new customers and improve choice for existing customers (potentially
replacing consoles altogether for some of them). 1612 The relative immaturity of
the market, its potential for growth, and exogenous factors such as technology
advances all lead us to consider that rivalry in this market is likely to be
intense, dynamic and characterised by innovative competitive strategies
absent the Merger. This dynamic competition would be harmed by the
Merger, resulting in substantial adverse effects for consumers.

11.289 In summary, we consider that there would be substantial adverse

effects arising from the Merger. These adverse effects are likely to endure
and grow over time as the market for cloud gaming services develops.

Costs of the remedy

11.290 We also considered the costs of the remedy, taking account of the
considerations set out in paragraph 11.270. Prohibition leaves the market

1611 A summary of market forecasts and monthly active users can be found at paragraphs 8.54 to 8.56. Currency
has been converted at $1.25 = £1.
1612 CMA, Provisional Findings, 8 February 2023, paragraph 59.

410
 structure unchanged and therefore does not cause distortions in outcomes.
The implementation of the remedy (see section below) would lead to no
compliance and monitoring costs.

11.291 We acknowledge that the Parties may incur costs as a result of

prohibition. However, in line with the CMA guidance set out above, we do not
attribute material weight to these costs. Further, we have found no other
effective remedy which would give rise to lower costs for the Parties in this
regard.

11.292 As set out above, we have found that the claimed Game Pass RCB
satisfies the relevant conditions under section 30 of the Act to be taken into
account in our assessment. We recognise that this RCB would be foregone as
a result of prohibition, and this therefore represents a relevant cost of
prohibition as a remedy. As explained above, we consider the claimed Game
Pass RCB would amount to a materially lower benefit than Microsoft’s claimed
estimate of benefits to UK customers of £[] million per year. We also expect
the extent of the benefit to fall over time. However, it has not been necessary
(or possible) to estimate the precise benefit that may be expected to accrue
from this RCB.

Balancing exercise

11.293 We have weighed the relevant costs of prohibition against the SLC and
the adverse effects that prohibition would effectively remedy.

11.294 We consider that in the circumstances of this case a precise

mathematical weighing exercise is neither necessary nor possible. We
recognise that whist the relevant costs (including the claimed Game Pass
RCB) are more readily quantifiable, the adverse effects arising from the SLC
cannot be accurately quantified, particularly given the dynamic and fast-
moving nature of the affected market. In carrying out this assessment, we
have therefore had regard to all the available evidence in the round – both
quantitative and qualitative – that informed our findings on both the SLC and
the relevant costs, in order to make an informed judgement on the
proportionality of prohibition.

11.295 We consider the main relevant cost of prohibition to be the loss of

benefits from the claimed Game Pass RCB. In this regard, while the foregone
RCB is likely to represent a material sum (even if this is materially lower than
£[] million per year), we consider this is likely to decline over time. On the
other hand, the harm arising from the SLC is substantial and may be expected
to increase over time as the cloud gaming services market in the UK grows. In
addition, we note that the claimed Game Pass RCB would not start to accrue

411
 until 2025, whereas the harm from the Merger may be expected to start to
arise from completion.

11.296 Considering our views on the likely growth and importance of the cloud
gaming market and the negative impact that the Merger may be expected to
have on competition within this market, we consider that the costs of the
remedy by foregoing the claimed Game Pass RCB are significantly
outweighed by the scale of the harm expected to arise from the SLC.
Moreover, we would expect the balance between the scale of the harm and
the lost RCB benefits to tip further towards the harm over time.

International context

11.297 As part of our proportionality assessment, we have also had regard to

the international context of the Merger. We note in this respect the CAT’s view
that ‘the demands of comity do require the CMA to be at least conscious of
the international dimension’ and that ‘in international cases, regard needs to
be had (even if it is not determinative or even immaterial) to the wider
context’. 1613

11.298 The Merger is a transaction between two global businesses that will
have effects across multiple jurisdictions. As noted in Chapter 2, the Merger is
the subject of review by competition authorities across the world. Some of
these reviews are still pending – including those in the EU and US – whilst
others have already concluded.

11.299 Following an in-depth investigation, we have concluded that we have

jurisdiction to review the Merger, and that the Merger gives rise to an SLC in
the supply of cloud gaming services in the UK. As explained above, this is a
growing and dynamic market that is expected to be worth more than £1 billion
in the UK by 2026. We expect the SLC we have found to result in substantial
adverse effects for UK consumers, which may take the form of higher prices,
worse quality or lower innovation in cloud gaming.

11.300 When considering remedies, we have similarly sought to identify as

comprehensive a solution to the SLC and resulting adverse effects in the UK
as is reasonable and practicable. In circumstances where we have found that
the only effective remedy to address the SLC and its adverse effects in the
UK is prohibition, the fact that this will necessarily have an impact outside the
UK does not conflict with the principles of international comity.

1613 Meta Platforms Inc. v CMA [2022] CAT 26, 14 June 2022, at [127(1)] and [129].

412
11.301 Under section 86(1) of the Act, the CMA is permitted to impose
remedies that extend to a person’s conduct outside the UK if that person is
carrying out business in the UK (as is the case for both Microsoft and
Activision). 1614 The Parties themselves accept that the Act allows the CMA to
adopt remedies which may have impacts outside the UK. 1615 In this case, we
have not identified any effective remedy that would avoid extra-territorial
effects. We note in this respect that the Microsoft Cloud Remedy (which we
have found to be ineffective) would also have had extra-territorial effects, as it
would have involved the imposition of regulatory restrictions on the Parties’
conduct on a global basis.

11.302 Our Merger Remedies Guidance notes that: “Where competition

authorities in other jurisdictions are considering a merger which the CMA is
also investigating, the CMA will consult with some or all of these authorities to
seek consistency and effectiveness in the approach to remedies where
relevant.” 1616 In this case, we have consulted with other competition
authorities, including in the EU and US. However, nothing arising from this
consultation changes our view on the appropriate remedy in accordance with
our obligations under the Act.

11.303 In light of the above, we are satisfied in this case that prohibition is a
proportionate remedy that respects the principles of international comity,
notwithstanding its extra-territorial effects.

Microsoft’s other submissions

11.304 We note Microsoft’s submission that to exclude taking action which is

reasonable and practicable to ‘mitigate’ the SLC or any adverse effects would
be wrong. We agree with Microsoft that it may be open to the CMA to take
action which only mitigates and does not fully remedy or prevent an SLC and
resulting adverse effects in certain circumstances, and that the Ryanair
judgment does not suggest that mitigation is not part of the statutory test.
However, we consider this might be a relevant and appropriate course of
action to consider in circumstances where any effective remedy which
prevents or remedies the SLC and its resulting adverse effects is either not
practicable or disproportionate. This is consistent with the “high duty” the Act
imposes on the CMA 1617 to try to achieve as comprehensive a solution to the
SLC and resulting adverse effects as is reasonable and practicable. 1618 For

1614 CMA87, paragraph 3.55.

1615 Microsoft response to the Remedies Working Paper.
1616 CMA87, paragraph 3.56.
1617 Ecolab Inc. v CMA [2020] CAT 12, at [74-75], citing the Court of Appeal in Ryanair Holdings PLC v CMA

[2015] EWCA Civ 83, at [57].

1618 Section 36(3) of the Act.

413
 the reasons explained in this Chapter, we do not believe those circumstances
arise here.

11.305 As set out above, the CMA has assessed the Microsoft Cloud Remedy
and found that it would not be effective in remedying the SLC and its adverse
effects in material respects. In these circumstances, the existence of this
ineffective alternative remedy does not render prohibition disproportionate.
Even if the behavioural remedy would mitigate the adverse effects to some
extent, it would not be consistent with the CMA's high duty under the Act to
accept this remedy in place of prohibition in circumstances where the CMA
has found that prohibition is both effective and proportionate.

11.306 Finally, we note our disagreement with Microsoft’s submissions

regarding the evidentiary bar required to prohibit transactions in digital
markets, and the relevance of potential legislative changes concerning
companies with ‘Strategic Market Status’. We have carried out this inquiry in
accordance with existing legislation and guidance and have not applied a
different standard or approach. We do not consider that any potential
legislative changes are relevant to our determination of the statutory
questions set out in the Act in relation to the Merger. We have explained in
this Chapter the framework we have applied for the assessment of remedies,
along with the reasons for our conclusions.

Conclusion on proportionality

11.307 On the basis of the above assessment, and our conclusion that the
Merger is likely to lead to significant and sustained adverse effects, which
outweigh by a significant and growing margin the RCBs that would be lost as
a result of prohibition, we conclude that prohibition of the Merger is the only
effective remedy and is not disproportionate in relation to the SLC and its
adverse effects.

Remedy Implementation

11.308 Having identified the appropriate remedy, we now consider how it

should be implemented.

11.309 The CMA has the choice of implementing any final remedy decision
either by accepting final undertakings if the Parties wish to offer them, or by
making a final order. 1619 Either the final undertakings or the final order must
be implemented within 12 weeks of publication of our final report (or extended

1619 Section 82 (final undertakings) and section 84 (final order) of the Act.

414
 once by up to 6 weeks), 1620 including the period for any formal public
consultation on the draft undertakings or order as specified in Schedule 10 of
the Act.

11.310 In line with our Merger Remedies Guidance, Microsoft would be

prohibited from subsequently acquiring the assets or shares of Activision or
acquiring any material influence over them. Our Merger Remedies Guidance
states that the CMA will normally limit this prohibition to a period of ten
years. 1621 We find no compelling reason to depart from the Merger Remedies
Guidance in this case by seeking a shorter or longer prohibition period.

Final decision on remedies

11.311 We have decided that prohibition of the Merger would be an effective

and proportionate remedy to address the SLC in the market for cloud gaming
services in the UK and its resulting adverse effects.

1620 CMA87, paragraph 4.68. An extension may be made if the CMA considers there are ‘special reasons’ for
doing so (section 41A(2) of the Act).
1621 CMA87, paragraph 5.10.

415