2021-2022

Global Computing Index

Assessment Report
Contents

Core Views 02

Introduction 05

Global Computing Index Assessment Results 09

Assessment of Computing Power Development 21

Economic Impact of Computing Power 26

Social Value of Computing Power 28

Recommendations for Actions 33

Appendix 36

01
Core Views

02
 2021-2022 Global Computing Index
Assessment Report

The COVID-19 pandemic has accelerated global digitalization, traditional industries are adopting digital
technology at an increasingly faster rate. Based on the overall trend and forecast from 2016 to 2025, the
digital economy accounts for a rising proportion of for each country. It is expected that this proportion
will reach 41.4% by 2025.

The regression analysis of the country’s Computing Index and economic indicators in this report show
that there is a significantly positive correlation between the Computing Index and the GDP/digital
economy. On average, for every 1-point increase in the Computing Index of the 15 key countries, the
national digital economy and GDP will increase by 3.5‰ and 1.8‰, respectively, and that trend is
expected to continue through 2025.

Countries with a Computing Index score above 60 are classified as leading countries, evaluation results
show that the US and China rank first and second, taking leading positions globally. Between 40 and 60
are rising countries which include Japan, Germany, the UK, France, Canada, South Korea, and Australia.
Below 40 are startup countries which include India, Italy, Brazil, Russia, South Africa, and Malaysia.

The competition among countries in terms of computing power has become increasingly fierce. All the
countries, except South Africa, have achieved improvement in their computing power scores. The division
of ranks for each country remains unchanged compared to the previous year. This partly reflects that the
competitive landscape around computing power has taken shape among countries. The leading position
of the United States and China in the global computing field is only strengthening.

The position of leading countries in the Computing Index has been extended due to the growth in
computing power supply, market space for application, and infrastructure support. The advantage of
rising countries in computing power lies in efficient computing resources, extensive application of
emerging technologies, and sound infrastructure support. Startup countries are dominated by
developing countries and they enjoy a broad market space and remarkable late-mover advantages in
computing power. Complemented by an ever-improving infrastructure, they are narrowing the gap with
rising countries.

03
 2021-2022 Global Computing Index
Assessment Report

AI computing reflects a country's cutting-edge computing capacity. An analysis of the 15 countries shows
that spending on AI computing increased to 12% of total computing, up from 9% in 2016, and is
estimated to reach 25% by 2025.

This year's study found improvements in the development of edge computing among the 15 countries. In
addition to the prominent performances of the US and China, countries like Germany and the UK have
stepped up their efforts in edge computing at a much faster pace than developing countries. IDC
estimates that investment in edge computing power will grow much faster than core computing power
over the next five years, and worldwide spending on edge computing servers will account for 24.9% of
total servers by 2025.

The Internet industry is still ahead of traditional industries. The top 5 industries also include finance,
manufacturing, telecom, and public affair. Finance has overtaken manufacturing for second place. The
financial industry has outperformed the manufacturing industry in terms of total investment and growth
rate in computing capacity and applications represented by AI and big data, resulting in its rise to second
place. In the post-COVID-19 era, most financial institutions are faced with the challenges of lower yields
and increased bad debt rates. Therefore, such institutions are moving faster to build intelligent platforms,
improve risk control capabilities, and further innovate service models.

The social value of computing power is increasingly prominent. Computing power can facilitate vaccine
and drug development, dramatically improving drug discovery, lowering average R&D costs, and reducing
the risk of clinical failure, AI-assisted drug discovery has become the focus of the global market in recent
years. Green computing power ensures sustainable development of society, driven by the demand to
address climate change with net-zero emissions, a green and centralized transformation of computing
power is continuing to accelerate. Computing power can also help companies cope with the pressure to
reduce emissions and gain a commercial head start in digital transformation.

04
Introduction

05
 2021-2022 Global Computing Index
Assessment Report

The COVID-19 pandemic has accelerated global

digitalization; traditional industries are adopting digital
technology at an increasingly faster rate.

COVID-19 has forced the global economy to deal with added uncertainty and dramatic upheaval. This caused
many countries to experience a decline in economic growth. In response to the economic impact of the
pandemic, and in order to boost economic recovery, governments around the world adjusted their policies to
promote the convergence of traditional industries with digital technology, advance the digital transformation of
industries, and develop the "digital economy".

During the pandemic, digital technology provided an important safeguard for much of the offline economy,
preserving people's livelihoods. Additionally, emerging technologies such as artificial intelligence (AI), 5G, the
Internet of Things (IoT), and cloud computing quickly converged with traditional industries to boost productivity,
cushioning the impact of any economic downturns. Under the impetus of the digital economy supported by
immense computing power, new industries have been developing rapidly. New convergent industries and
entirely new business concepts and models such as the Industrial Internet, intelligent manufacturing, and the
Internet of Vehicles are emerging and making a quantifiable impact. This has resulted in computing power
emerging as a key factor for reviving the economy.

06
 2021-2022 Global Computing Index
Assessment Report

The digital economy continues to grow steadily,

with developing countries growing faster.

As digitalization accelerates, the digital economy is set to become a major driver for global economic recovery
and will be a source of continued economic growth in the future. Based on the overall trend and forecast from
2016 to 2025, the digital economy accounts for a rising proportion of GDP (hereafter refers to proportion) for
each country. It is expected that this proportion will reach 41.4% by 2025. Buffeted by COVID-19, the overall
global economy declined, but the digital economy actually grew. According to the latest data, the digital
economy now accounts for 44.6% of GDP in developed countries, significantly higher than the 24.8% in
developing countries, but the proportion in developing countries grows at a rate of 4.7%, slightly higher than the
2.4% in developed countries. Some newly added countries in this assessment report (Canada, Italy, Malaysia
and India) have a lower digital economy to GDP ratio, making the overall a lower proportion compared to the
previous year. However, that proportion continunes to grow.

In the future, the digital economy will be the mainstay of economic development and recovery. Digital
transformation has become the focus of global attention during the pandemic, with both developed and
developing countries continuing to stress the importance of the digital economy.

Trends of the proportion of digital economy in sample countries from 2016 to 2025

US$ Billion
48.1% 50%

90000 44.6% 45%

41.4%
80000
39.1% 40%
36.9%
70000 35%
31.2% 30.3%
60000 30%
24.8% 25%
50000
20.1%
40000 20%

30000 15%

20000 10%

10000 5%

0 0%
2016 2017 2018 2019 2020(BY) 2021E 2022E 2023E 2024E 2025E

Digital GDP Proportion in Proportion in Proportion in

economy all countries developed countries developing countries Data source: IDC, 2021

07
 2021-2022 Global Computing Index
Assessment Report

The explosion of industry applications provides

a driving force for the digital economy.

With the technological advancement and quickly evolving product innovation typified by AI, digital technology
has accelerated its deep convergence with vertical industries, and its applications have been undergoing
constant improvement and innovation.

AI has accelerated its convergence with other digital technologies, which helps optimize operating procedures
and improve productivity. Edge computing enhances the effectiveness of solutions, providing programs with
faster response times and stronger autonomous operational capabilities. Robotics also plays an extensive
supporting role in production, including working in dangerous environments and COVID-19 control. In the digital
economy, data is a key factor for production, the Internet is a basic carrier, and computing power is an important
tool to improve productivity.

The world is accelerating the deployment of the digital economy in key areas. The United States (US) is focusing
on cutting-edge technologies and high-end manufacturing with the theme of "AI". China is striving to accelerate
the pace of digital technology industrialization in an effort to spur the development of digital technology along
with the digital transformation of traditional industries. Additionally, China is working to promote the deep
convergence of digital technology and the real economy to form an internationally competitive digital industrial
cluster. The United Kingdom (UK) aims to build itself into a global AI and data-driven innovation powerhouse.
South Korea is focusing on smart factories. Japan is targeting the industrial Internet. This technology-driven
focus across so many countries indicates an increasing emphasis on digital technology as a econonomic
cornerstone, making the digital economy a priority for policymakers.

When it comes to the practical application of digital technology, the decisive factor is computing power. Huge
computing resources are behind the practical application of every new technology. As the digital economy
continues to grow, so too will the computing power required to run it.

08
Global Computing
Index Assessment
Results

09
 2021-2022 Global Computing Index
Assessment Report

Updates on the Global Computing Index Assessment Methodology

The Global Computing Index is an index that evaluates how computing power, GDP, and the digital economy
promote each other and grow together.

This report explores the connections between computing power and

economic indicators by setting up a Global Computing Index framework and
evaluating the correlation between computing power, GDP, and the digital
economy. It aims to emphasize the importance of computing power in
economic development where "computing power is productivity."

The Global Computing Index research covers 15 countries on six continents. Developed countries include: the
United States, Canada, Japan, South Korea, Australia, the United Kingdom, France, Germany, and Italy. Developing
countries include: China, India, Malaysia, Brazil, Russia, and South Africa. Canada, Italy, Malaysia, South Korea, and
India are new additions to this year's evaluation. In the indicator system, the level-2 indicator of edge computing is
added to the computing power category, which reveals the computing power form and development level in
various countries more comprehensively. The new technology (such as SSD/SCM) utilization level-2 indicator is
added to the computing efficiency category. Combining this with the utilization of computing, memory, storage
and other resources can enable higher efficiency in evaluating the utilization of computing power. Furthermore, the
data center software and service level-2 indicators are added to the infrastructure support category, increasing the
diversity of the computing power support elements.

Computing Capacity Application Level

General computing capacity: Proportion of server expenditure to Big data: Proportion of overall expenditure to GDP on software,
GDP and server shipments hardware and services
Scientific computing capacity: Number of global top 500 AI: Proportion of overall expenditure to GDP on software, hardware,
supercomputers and their rankings and services
AI computing capacity: Proportion of AI computing hardware (server) IoT: Proportion of overall expenditure to GDP on software, hardware
expenditure to GDP and shipments and services
Terminal computing capacity: Proportion of smartphone and PC Blockchain: Proportion of overall expenditure to GDP on software,
expenditure to GDP and shipments hardware and services
Edge computing capacity: Proportion of edge computing Robot: Proportion of overall expenditure to GDP on software,
expenditure to GDP hardware and services

Computing Efficiency Infrastructure Support

Cloud computing penetration: Proportion of servers with cloud Data center software and services: Proportion of data center
management software installed (survey data) and cloud software and service support
computing-related expenditure of total server expenditure Data center scale: Number of data centers (statistical data)
CPU utilization: Average utilization of server CPU (survey data) Data center efficiency: Average data center PUE (research data)
Memory utilization: Average utilization of server Memory (survey data) Network infrastructure: Annual shipments of enterprise network
Storage utilization: Average utilization of server storage devices devices and annual proportion of operator network expenditure
(survey data) Storage infrastructure: Annual storage shipment capacity (TB) and
New technology utilization: Average utilization of technologies such annual proportion of storage expenditure
as SSD/SCM (survey data)

10
 2021-2022 Global Computing Index
Assessment Report

Analysis of Global Computing Index Assessment Results

Based on the scores of the Computing Index for each country, cluster analysis of each indicator, and the impact
of the Computing Index on the digital economy and GDP when the Index's value is increased, this report divides
these countries into three ranks: leading countries, rising countries and startup countries.

By observing the distribution of the Computing Index in various

countries and the economic growth resulting from an increase in the
Computing Index, this report finds that the boundaries of the Computing
Index for the countries of the three rank types are at 60 and 40 points
respectively. Countries with a Computing Index score above 60 are
classified as leading countries; between 40 and 60 are rising countries;
below 40 are startup countries.

Evaluation results show that the US and China rank first and second, taking leading positions globally. Rising
countries include: Japan, Germany, the UK, France, Canada, South Korea, and Australia. Startup countries
include: India, Italy, Brazil, Russia, South Africa, and Malaysia.

The competition among countries in terms of computing power has

become increasingly fierce. All the countries, except South Africa, have
achieved improvement in their computing power scores. The division of
ranks for each country remains unchanged compared to the previous
year. This partly reflects that the competitive landscape around
computing power has taken shape among countries. However,
compared to the previous year's data, the gap between rising and startup
countries is narrowing. In contrast, the gap between leading and both
rising and startup countries is widening. This reflects that the leading
position of the United States and China in the global computing field is
only strengthening.

11
 2021-2022 Global Computing Index
Assessment Report

Leading countries: Their leading position in the Computing Index has been extended due to the growth
in computing power supply, market space for application, and infrastructure support.

specifically, the country's edge computing servers

increased by 13.6%, and cloud computing
development increased by 4.3%. In addition, the
country's application of emerging technologies such
as AI, big data, Internet of Things and blockchain has
matured, and its growth in these indicators is ahead
of the global average. What's more, the hyperscale
data centers in the US have gained further advantage
due to their immense size and maintained their top
The level of computing power in the US increased by position worldwide. Driven by world-leading
5% year-on-year to 77 points. With the negative hyperscale cloud service providers, the overall
impact of COVID-19 on the overall macro economy energy efficiency of the country's data centers has
and the prominent market size as well as improved, but there is still room for improvement in
development level, the US still achieved significantly the energy efficiency of enterprise-level data centers
better scores on some level-2 indicators. More compared to hyperscalers.

Rising countries: Their advantage in computing power lies not in computing capacity, but in efficient
computing resources, extensive application of emerging technologies, and sound infrastructure
support. There are 7 rising countries, all of which are developed countries. Japan, Germany and the UK are
ranked 3-5 in the Computing Index, with South Korea, Canada and Japan showing the fastest growth. The total
number of data centers in rising countries accounts for about 33% of the total among the complete set of 15
countries, and the PUE indicators of these countries are generally leading the world. In particular, European
countries prioritize the transition toward a circular economy. It is estimated that by 2023, Europe will invest more
than 60 billion USD in enabling digital technology to support sustainable development. At the same time,
European countries are experiencing steady growth in computing power as European companies focus
increasingly on digital-first development strategies. A summary of the evaluated performance of several rising
countries is available:

Germany enjoys a strong industrial foundation. The

country put forward the concept of "Industry 4.0" in
2013, allowing it to seize the opportunity opened up
by the new wave of industrial revolution and play a
leading role in the digital transformation of global
industry. The country's long-term digital investment
has made it more resilient during the pandemic, and
is key to the rapid recovery of its economy. The

12
 2021-2022 Global Computing Index
Assessment Report

evaluation results indicate that, driven by the outbreak of COVID-19, the Japanese government
demand of digital transformation of the industry, introduced and implemented a large-scale economic
Germany's edge computing expenditure increased stimulus plan. Digital transformation based on
by 29.3%. Its expenditure in the Internet of Things computing power has become an important basis for
and blockchain also increased rapidly, driving a Japan to promote industrial upgrades and stimulate
16.1% increase in Germany's application level of economic recovery in the post-pandemic era. As
computing power. shown in the assessment results, the overall level of
Japan's computing power has increased by 12.7%.
Japanese manufacturers in medical devices,
semiconductors and other industry sectors have
used AI and other digital technologies to fully
empower business innovation. Significant progress
has been made in AI computing power.

Japan is also a global manufacturing powerhouse

despite the lack of large domestic IT, internet, and
other tech companies. The country is highly ranked
in high-end manufacturing fields such as basic
electronics, integrated circuits, materials,
automobiles, and equipment. In recent years, the
Japanese manufacturing industry has leveraged The overall development level of computing power
digital technology to consolidate its global edge in in France has improved by 4.6%, mainly contributed
high-end manufacturing, making it more difficult for by the application level, with AI application
other corporate competition to keep up. After the expenditure achieving a significant increase of 32.7%.

13
 2021-2022 Global Computing Index
Assessment Report

Startup countries: These countries are dominated by developing countries and they enjoy a broad
market space and remarkable late-mover advantages in computing power. Complemented by an
ever-improving infrastructure, they are narrowing the gap with rising countries. The results of the
computing power evaluation demonstrate that startup countries have achieved outstanding performance in
different indicators relying on their respective resource endowments:

constitute a global structural growth point, the

country leads the world in its end user base,
expenditure, and growth rate. In addition, India's
expenditure on storage and network infrastructure
has been growing rapidly, with spending growing
faster than any other country in the world. Although
the outbreak of COVID-19 has affected the growth
momentum of India's computing power, the gap
between India's computing power level and that of
Italy is the only developed country ranked as a the rising countries is only 2 points. It’s overall
startup. Although its GDP ranks high, the digital computing power level still has great growth
economy accounts for a lower proportion of Italy's potential over the next few years, and is expected to
economy compared to other Western European transition to a rising country very soon.
countries. Hence, there is still much room for
improvement in terms of investment in computing
technology and new technologies such as AI, and
infrastructure support such as storage and networks.

India's Computing Index has a score of 38. In terms of

computing power, India has achieved a large
increase in AI computing and edge computing. For
mobile and desktop computing markets that

14
 2021-2022 Global Computing Index
Assessment Report

Analysis of Each Component of the Global

Computing Index Level-2 Indicators

Computing capacity
As a core component of the Global Computing Index, computing capacity indicates overall levels of computing
power investment in various countries along with each country's focus by looking at the number and investment
proportion of servers, mobile devices and PCs.

Evaluation of Computing Capacity

79 80 80
78 80

56

44
38 38
32 41

31
24 27
25

General AI High-performance Terminal Edge

Computing Computing Computing Computing Computing

Leading countries Rising countries Startup countries Data source: IDC, 2021

Due to factors such as the COVID-19 outbreak and structural imbalances in economic development, countries
registered a decline in overall spending on servers, but most maintained rapid growth in edge computing. By
innovating and expanding the functions and scope of core data centers, edge computing has been an
important driving force behind global growth in enterprise infrastructure. This year's study found
improvements in the development of edge computing among the 15 countries. In addition to the prominent
performances of the US and China, countries like Germany and the UK have stepped up their efforts in edge
computing at a much faster pace than developing countries. Edge computing provides an important platform for
emerging technologies such as 5G, the Internet of Things, robotics, and AI. The International Data Corporation
(IDC) estimates that investment in edge computing power will grow much faster than core computing power
over the next five years, and worldwide spending on edge computing servers will account for 24.9% of total
servers by 2025.

15
 2021-2022 Global Computing Index
Assessment Report

European countries boost their regional competitiveness in edge computing mainly by rebuilding the
industrial value chain, exploring use cases with strong demand, and providing strategic guidance at
the macro level. Regarding edge computing as a key driver for digital transformation in Europe, European
countries and organizations have picked up the pace in deploying edge computing infrastructure in recent years.
Nearly 30% of European organizations plan to start using edge computing technologies in the next two years. In
industries that are at the forefront of edge computing, such as transportation/logistics, manufacturing, and
energy, some organizations have moved beyond the pilot phase to deploy edge computing solutions on a
broader scale. Edge computing is essentially a distributed computing paradigm oriented to specific use-case
scenarios. Therefore, the synergy between industries and systems is an important factor affecting the
deployment of edge computing. European telecom operators, solution providers, and organizations in verticals
are strengthening industrial collaboration and business capability integration to empower various industries
through ever-improving ecosystems. In Germany, seamless switching across operator networks has been
realized for autonomous driving, ensuring the continuity of key mobile services such as collision warning
systems. In many verticals, European organizations actively explore application scenarios with strong demand
for edge computing, such as real-time analysis of user shopping behavior in retail, management of complex
tasks, automation of back-end processes and operations, and AR-assisted surgeries and remote imaging
diagnostics in the medical industry. In addition, aiming to satisfy the rigid demand of European enterprises and
public administrations for processing sensitive data, the European Union has established the European Alliance
on Industrial Data, Edge and Cloud as an important part of the European data strategy to consolidate Europe's
leading position in industrial data. AI computing reflects a country's cutting-edge computing capacity. In
China, AI computing is developing rapidly. An analysis of the 15 countries shows that spending on AI
computing increased to 12% of total computing, up from 9% in 2016, and is estimated to reach 25% by 2025.

16
 2021-2022 Global Computing Index
Assessment Report

Computing efficiency

This metric reflects the current utilization of computing power in a country. Due to their higher adoption rates of
cloud computing, some countries also have higher computing capacity utilization rates than others.

Evaluation of Computing Efficiency

73

63
64
58
55
54 61
57 56
49
46 46
39
37 37

CPU utilization RAM utilization Storage utilization New technology Cloud computing
penetration penetration

Leading countries Rising countries Startup countries Data source: IDC, 2021

Spending on cloud infrastructure continues to grow, and cloud computing accounts for a smaller share
of investment in startup countries，this report found that investment in cloud infrastructure accounted for an
average of only 20.4% of overall IT infrastructure in startup countries, but up to 60.7% and 31.7% in leading and
rising countries, respectively. After more than a decade of development, cloud computing has entered a new era
characterized by multi-cloud and hybrid cloud architectures. Looking at the segments of infrastructure spending,
traditional data centers remain the largest component. Nonetheless, public and private cloud segments are
growing much faster, and that trend will continue in the coming years.This report found that investment in cloud
infrastructure accounted for an average of only 20.4% of overall IT infrastructure in the startup countries, but up
to 60.7% and 31.7% in the leading and catch-up countries, respectively.This report found that investment in
cloud infrastructure accounted for an average of only 20.4% of overall IT infrastructure in the startup countries,
but up to 60.7% and 31.7% in the leading and catch-up countries, respectively.

Application level

The application of emerging technologies such as big data, AI, the Internet of Things (IoT), blockchain, and
robotics is the core driving force of future IT expenditures, and will also reflect a country's economic
development potential and comprehensive national strength. AI and IoT applications in particular have made
significant contributions to countries and industries.

17
 2021-2022 Global Computing Index
Assessment Report

Evaluation of Application Level

80
73 78 75
68

61 47
40
37
44
37
27 28 28 26

AI Big data IoT Blockchain Robotics

Leading countries Rising countries Startup countries Data source: IDC, 2021

The evaluation shows that the US and China are at the forefront in terms of total spending on AI applications,
with a high growth rate of 27.9% and 34.7% respectively, and that momentum is expected to continue over the
next five years.

Countries around the world are stepping up efforts in AI, a frontier technology for national strategies. For their "Society 5.0" vision,
Japan has introduced a series of policies to promote the development of AI technologies and AI applications in society, such as the
"Cross-Ministerial Strategic Innovation Promotion Program (SIP) Phase II" and "Artificial Intelligence Technology Strategy". France has launched a
national strategy for AI, which aims to improve the country's AI competitiveness, consolidate its leadership in embedded AI and trusted AI, and
accelerate AI applications in the economic sector, with an additional investment of 2.2 billion EUR. This strategy also clarifies AI applications in
the four areas of healthcare, environment, transportation, and national defense.

Based on distributed computing systems, large-scale pre-trained

models have demonstrated unprecedented performance in many AI
tasks, continuously increasing the industry's expectations of the
capability boundaries of deep learning models. Currently, it has
gradually become an industry consensus to serve specific downstream
AI tasks by fine-tuning large-scale pre-trained models instead of training
models from scratch. With an ever-growing wealth of application
scenarios and the rapid growth of data volume, large models may
become the basis for scaling innovation. AI companies around the world
are actively developing and commercializing technologies for
large-scale pre-trained models. In the US, OpenAI, Google, Microsoft,
Meta, and other organizations have established large models with
hundreds of billions or trillions of parameters, such as GPT-3, the Switch
Transformer, and MT-NLG.

18
 2021-2022 Global Computing Index
Assessment Report

In China, organizations like Inspur Information, and Alibaba launched the

large-scale pre-trained models Yuan 1.0 and M6. Inspur Information's
Yuan 1.0 is a massive model with 245.7 billion parameters that has been
trained on a dataset of more than 5TB. These models will empower all
industries and accelerate the intelligent transformation of traditional
industries, and will have an especially noticeable and positive impact in
critical fields such as law, healthcare, and education.

The business value of AI is widely recognized, and global spending on AI applications is growing
rapidly. Most businesses in South Africa are actively experimenting with AI technologies. Agriculture is a pillar
industry in South Africa, and innovative technologies are increasingly important to modernize the sector and
improve the livelihoods of large farming communities. AI technologies help local farms analyze and process
maps, and identify problem areas for crops, thus optimizing yields. In the US, a large number of AI technologies
are applied to the treatment and diagnosis of a variety of diseases such as genetic syndromes, lung cancer,
breast tumors, and post-traumatic stress disorder, as well as research on protein folding, the detection &
identification of antimicrobial resistance genes, and the development of new drugs. Major hospitals in South
Korea are also strategically setting up their big data imaging databases to accurately interpret medical images
with AI technologies to assist physicians in diagnosing breast cancer and tuberculosis from over 100,000 X-ray
images, and recognize and decipher signs of diseases using images of tissue sections.

Infrastructure support
Infrastructure support in the form of data center scale, data center PUE, software & service，and storage &
network infrastructure provide a guarantee for computing capacity, computing efficiency, and application levels
at a macro level.

Evaluation of Infrastructure Support

77 80 80
70
65 61
52
49
55 55 40

39 37
31
25

Data center scale Data center PUE Software and Storage Network
service infrastructure infrastructure

Leading countries Rising countries Startup countries Data source: IDC, 2021

19
 2021-2022 Global Computing Index
Assessment Report

Data centers are the foundation for the digital transformation of a country’s industries. The scale and
energy efficiency of data centers will affect the future development of computing power in the
country. Leading countries have prominent advantages in hyperscale data centers, altogether accounting for
more than 50% of the scale of the 15 countries. The US has the most hyperscale data centers of any country—
four times as many as China, which has the second highest—mainly due to the continued infrastructure
construction by world-leading cloud service providers such as Amazon, Google, and Microsoft. Driven by the
sustained and rapid development of cloud computing and new infrastructure policies, China has experienced
strong growth momentum in the number of hyperscale data centers, providing a strong backbone for the overall
development of the digital economy and the deep integration of digital technologies and the real economy.

In terms of data center PUE, an increasing number of enterprises are turning to modular data centers and liquid
cooling technology to reduce their PUE, while managing energy efficiency using big data mining & analytics, AI,
or virtual simulation technologies. According to the latest research results, computing infrastructure in
developed countries including the US, Germany, the UK, France, Australia, and Canada are more energy-efficient
among the 15 countries, with an average PUE value of 1.4. Consuming 0.01% of the world's electricity, Google
has worked for years to integrate sustainability into its operations by improving the energy efficiency of its
infrastructure and reducing carbon emissions from its data centers. As the world's largest buyer of renewable
energy in the non-utility sector, Google operates 13 renewable energy projects in Europe alone, and the average
PUE of its data centers has now reached 1.1.

20
Assessment of
Computing Power
Development

21 05
21
 2021-2022 Global Computing Index
Assessment Report

Levels in Different Industries

This is the second consecutive year that the computing power of global industries has been evaluated by taking
into account the investment of each industry in general computing, AI computing, and edge computing, along
with the application maturity of emerging technologies. As per the latest data and user surveys, the distribution
of the investment in computing power of global industries is shown below. Overall, the Internet industry is still
ahead of traditional industries. The top 5 industries also include finance, manufacturing, telecom, and public
affair. Finance has overtaken manufacturing for second place.

Evaluation of Computing Power of Global Industries

90

80 75
70 3

60
60 59
7 2 50 49 48
50
2 1 3 40
40 37 36 36 35
3 3 30 29
30 2 2 4
2 4
20

10
0
t

om

re

ia

gy
ne

or
ai

tie
nc

ic

al
rin

tio
ed
ca

er
aff

rv

sp
es
ec
r

na

i
tu

ca
te

til
th

En
Se

an
l

ic

ho
Fi

ac

Te
In

U
l
ea
bl

Ed

Tr
uf

w
Pu

H
il/
an

Data source: IDC, 2021

M

ta
Re

Over the past few years, the finance sector has been increasing its
investment in computing power, especially in heterogeneous computing
servers for AI application scenarios. In the post-COVID-19 era, most
financial institutions are faced with the challenges of lower yields and
increased bad debt rates. Therefore, such institutions are moving faster to
build intelligent platforms, improve risk control capabilities, and further
innovate service models. Judging from the evaluation results, the financial
industry has outperformed the manufacturing industry in terms of total
investment and growth rate in computing capacity and applications
represented by AI and big data, resulting in its rise to second place.

22
02
 2021-2022 Global Computing Index
Assessment Report

Internet: Leadership in global computing power

by actively embracing emerging technologies

As digital-native enterprises, the Internet industry is most open to emerging technologies like cloud computing, big
data, and AI. Statistics from IDC show that more than 90% of the IT infrastructure purchased by Internet enterprises in
2021 was applied to cloud computing deployments,that share was less than 50% for traditional industries. In
addition, the integration of the Internet with AI, big data, and other emerging technologies has also generated
demand for massive computing capacity. According to the data used in this report, the Internet industry is at the
forefront in spending on AI, big data, the Internet of Things, and blockchain. In 2021, the Internet industry accounted
for nearly 30% of global spending on servers, and this trend is expected to continue, according to IDC.

Finance: Intelligent acceleration to effectively support

the innovative development of financial services

The financial industry embraced AI very early on. Over time, validated applications have been rapidly promoted from
large financial institutions to small and medium-sized institutions. AI as been adopted mainly for intelligent customer
services and risk control. AI was used earlier in customer service and is therefore more mature. In this sector,
intelligent customer service is mainly used for optimizing artificial collaboration, Q&A bots, HR services, and
data/service supervision. Chatbots, intelligent customer service, and virtual assistants in the finance sector can
provide users with accurate, personalized consultation and services through text or voice dialogue, thus greatly
saving the costs of operating traditional call centers. By autonomously learning user preferences and habits, they can
accurately identify potential needs and significantly improve user stickiness and conversion. Major global banks,
including the Bank of America (BOA), JPMorgan Chase, Capital One, Mastercard, and American Express, have adopted
chatbots to optimize their digital banking services. Over time, the voice recognition rate of AI-powered customer
service has increased from 55% on average to over 85%. Banks have created AI virtual assistants for corporate clients
to get their account balance information through a simple consultation.

JPMorgan Chase - Innovative applications for AI virtual assistants

JPMorgan Chase has launched an AI virtual assistant for corporate clients to better meet their needs for business
settlements. Since corporate clients have multiple accounts in different currencies, they often have to look up or
export account information from as many as 1,200+ webpages in the portal to access account information, settle
funds, and send remittance. The AI assistant not only learns from customer interactions, but also categorizes and
summarizes the questions it gets in order to quickly find the appropriate answers. Eventually, the assistant will be
able to take proactive actions, such as making advance phone calls to clients who may delay payments. The AI
virtual assistant will be available via computer, mobile apps, and a voice virtual assistant device to provide an
efficient and convenient experience.

23
 2021-2022 Global Computing Index
Assessment Report

Manufacturing: Building digital factories

for intelligent manufacturing

Manufacturing is a key pillar that underpins the development of the real economy. It is one of the traditional
industries utilizing the highest level of computing power. In 2021, investment in manufacturing-related
computing power accounted for 12% of the world's total. Among the major traditional industries, only finance
accounted for a higher share. While promoting digital transformation, the manufacturing sector not only needs
to support the operation of large ERP systems, but also needs to take into account the application of new
technologies such as the Internet of Things and sensors. From a global perspective, manufacturing makes the
largest investment in two emerging technologies: the Internet of Things (IoT) and robotics. In 2021, spending on
IoT and robotics in this sector accounted for 37% and more than 60% of the world's total, respectively. In
addition, manufacturing is also ahead of most traditional industries in the application of AI, big data, blockchain,
and other emerging technologies.

As the industry moves toward smart manufacturing and high-end manufacturing, an increasing number of
excellent business practices have been adopted. For example, in a large manufacturing plant in Europe, a new
laboratory facility utilizing these technologies responds twice as fast as its predecessor, which helps reduce
factory operational costs 11% via the automation of low-value tasks since it was put into production. The
benefits of the five use cases are as follows: the smart factory reduced design refinements from concept to
implementation by 45%; the fully automated inbound logistics reduced indirect labor for individual products by
29%; digital online visual quality control reduced indirect labor for individual products by 17%; process linkage
and traceability cut material usage errors by nearly 100%; and smart personnel matching increased overall
equipment efficiency (OEE) (OEE = Availability * Performance Index * Quality Index) by 3%. These would not have
been possible without substantial computing power.

Telecom: Optimizing internal management and enabling

business innovation with computing power investment

With the application of technologies such as 5G and cloud computing, telecom operators, faced with internal
pressure for business growth, need to better support business support systems (BSS) to increase customer
stickiness and optimize operations support systems (OSS) to improve operation efficiency. Externally, they need
to provide support for value-added businesses such as smart transportation, smart retail, Internet of vehicles,
games & entertainment, and AR/VR applications.

24
 2021-2022 Global Computing Index
Assessment Report

It was found that most telecom operator equipment is aging with high failure rates and subpar performance.
Moreover, traditional data centers occupy large tracts of land and have high energy and maintenance costs,
which result in a high OPEX. With concurrent growth in key businesses, such as a sharp increase in computing
demand from billing and CRM, telecom operators are more willing to purchase new IT infrastructure. In order to
provide efficient computing power, they accelerate the integration with and replacement of old equipment to
modernize IT systems. Meanwhile they significantly optimize infrastructure investment, allowing the industry to
better handle business growth and ensure business efficiency.

Healthcare: Increasing investment in computing power

to promote the construction of information platforms

Amidst an aging global population, the bar is being raised for healthcare standards. The global healthcare
industry will maintain high investment in biotechnology, pharmaceuticals, and medical informatization. It shows
a positive growth trend in the procurement of ICT infrastructure. The industry has seen preliminary results in
digital development, with comprehensive information systems centered on computing platforms, and
hospital-wide systems for data interoperability, regional collaboration, tiered diagnosis & treatment, and health
services.

Medical institutions are dedicated to establish unified health information platforms to provide better diagnosis
and treatment services for patients within and across regions. With the development and application of new
technologies such as 5G, AI, IoT and cloud computing, big data will enable an intelligent upgrade of the medical
industry. As hospitals continue their digital transformation, the number of business systems, such as public
application services, clinical information, collaborative interaction, and electronic medical records are
increasing and resulting in massive amounts of data. Consequently, there is a growing demand for centralized
management, sharing and analysis of data resources. Increased investment in computing power can effectively
guarantee business systems operation, improve operation capacity & continuity of key businesses, ensure
system response during peak hours, and maintain high system availability and security. In addition, by
streamlining system structures and establishing more linked data relationships, hospitals can realize data
connectivity and collaboration efficiently, promoting the medical informatization to support data structures.
These efforts lay a solid foundation for a smoother medical care experience and provide effective support for the
construction of big data platforms and core systems operation in hospitals as well as application scenarios such
as public health and community governance.

25
Economic Impact
of Computing Power

26
 2021-2022 Global Computing Index
Assessment Report

From a quantitative perspective, computing power plays a significant role in a country's macroeconomic
development. The regression analysis of the country’s Computing Index and economic indicators in this report
show that there is a significantly positive correlation between the Computing Index and the GDP/digital
economy. In the linear regression goodness-of-fit test, the model has a better fit when the Computing Index is
the independent variable and the digital economy and GDP are the dependent variables, as opposed to having
the Computing Index as the dependent variable and the latter two as the independent variables. Therefore,
using the Computing Index to predict the digital economy and GDP better fits the reality than using the latter two
to predict the Computing Index.

On average, for every 1-point increase in the Computing Index of the 15 key
countries, the national digital economy and GDP will increase by 3.5‰ and
1.8‰, respectively, and that trend is expected to continue through 2025. After
further regression analysis on the Computing Index and GDP of countries at
different tiers, it was found that the influence of a country’s Computing Index
on GDP is different based on the point range. If m refers to the impact of every
1-point increase when the Computing Index is below 40 points, then the impact
will increase to 1.5m or 3m when the index is more than 40 points or 60 points,
respectively, which means more significant economic growth.

As mentioned above, the increase in the Global Computing Index mainly depends on the computing capacity
and application level of emerging technologies. That is, while computing power investment provides a basic
guarantee for the development of emerging technology applications, it is also further promoted by the demand
from emerging technology applications, thus having a multiplying effect on overall economic growth.

Regression analysis of Computing Index and GDP

GDP(US$ Billion)
25000
A3
x+
m

USA
3.0

20000
Y=

15000 2 China
+A
.5 mx
1
Y=
10000 Leading countries
Japan
Rising countries
+A1 Germany
Y=mx UK Startup countries
Italy France
5000 South Korea Data source: IDC, 2021
Russia
Malaysia
South Canada
India
Africa Brazil Australia Global Computing Index
0
0 20 30 40 50 60 70 80 90

27
Social Value of
Computing Power

28
 2021-2022 Global Computing Index
Assessment Report

Computing power facilitates new drug development

and epidemic prevention and control

Vaccine development

In recent years, emerging infectious diseases such as the novel coronavirus (COVID-19), severe acute respiratory
syndrome (SARS), avian influenza (H5N1, H7N9), influenza A (H1N1), Middle East Respiratory Syndrome (MERS),
and Ebola (EBOV) have had a global impact. They are all highly infectious and wide spreading. Most lack an
effective drug or vaccine treatment and are incredibly unpredictable. They seriously endanger human life and
health, undermine social stability and economic development, and present a major challenge to public health
systems around the world. The development of vaccines and drug therapies to fight against such diseases is of
the utmost importance.

Computing power can facilitate vaccine and drug development, dramatically improving drug
discovery, lowering average R&D costs, and reducing the risk of clinical failure. After four decades of
global development, computational drug discovery and design (CDDD) has been widely used in drug discovery
research. For example, with the help of parallel computing clusters, technologies like high-throughput virtual
screening (HTVS) can obtain candidate compounds in a short period of time. These technologies have
significantly improved the efficiency and success rate of drug molecule design, and provided a solution for the
long development cycles previously associated with developing drug therapies. AI-assisted drug discovery
has become the focus of the global market in recent years. New drug development includes such steps as
target selection and validation, compound screening & optimization, pre-clinical research and clinical trials. The
development of AI technologies has provided new techniques for drug development and assisted the
biopharmaceutical industry in accelerating digital and intelligent transformation. According to Deep Pharma
Intelligence, there were 240 AI-assisted drug development companies worldwide as of 2020. Major
pharmaceutical companies, such as Pfizer, GlaxoSmithKline, and Novartis have launched schemes in AI new
drug R&D, with popular research directions being target selection and validation, and lead compound screening
and optimization.

29
 2021-2022 Global Computing Index
Assessment Report

Epidemic prevention and control

AI computing has been critical in epidemic prevention and control since the COVID-19 outbreak. AI
computing has played an integral role in the prevention and control process.

AI computing accelerates genetic sequencing analysis of suspected cases and significantly reduces
contract tracing time. Genetic sequencing has become a key tool for the prevention and control of epidemic
infectious diseases caused by viruses. By sequencing the genes of pathogenic microorganisms, the genome
sequence information of new unknown viruses can be obtained very quickly. This allows researchers to reveal
pathogen-related characteristics, analyze the evolutionary origin of viruses, and study the pathogenic and
pathological mechanism of viruses, thereby supporting the prevention, control, and research of outbreaks. Each
genetic sequencing generates data over 10TB in size. AI is required to analyze such massive amounts of data.
Analysis is further enhanced by edge computing, which allows analysis to take place in a wide range of
environments and locations, significantly boosting the sequencing capabilities and efficiency for
microorganisms, viruses, and human genes.

Green computing power ensures sustainable

development of society

Computing power can help companies cope with the pressure to reduce emissions and gain a
commercial head start in digital transformation. Oil and gas companies are exploring how to use depleted
underground oil and gas fields and reservoirs for carbon storage. Graphics computing technologies can enable
geographic information systems (GIS), 3D modeling, and imaging, which allows these companies to assess the
target sites efficiently. Non-ferrous metal manufacturers use edge computing and AI to upgrade metallurgical
processes, optimize drilling rig operations, and shave peak loads, thereby ensuring safe production and
improving resource usage efficiency. Equipment manufacturers integrate equipment systems for a higher level of
automation and intelligence to monitor, analyze, and optimize plant energy usage, which generates data to
support asset management and production processes. With the help of AI computing power, car makers optimize
the distribution of charging stations and vehicle operators' fleet mix, routing, and resource allocation to reduce
their carbon footprint. Power grids control power transmission and distribution through edge computing and the
Internet of Things. Transmission efficiency is enhanced without increasing the carbon footprint. Data from all
segments is collected through sensors and smart meters, and recommendations on power consumption
optimization are offered via AI and big data analytics.

30
 2021-2022 Global Computing Index
Assessment Report

Driven by the demand to address climate change with net-zero emissions, a green and centralized
transformation of computing power is continuing to accelerate. As cloud computing develops, modern
data centers take on massive computing and storage tasks. The boom in AI technologies and the popularity of
accelerated computing applications continue to drive the exponential growth in processing power consumption
(270/280W+ for CPUs and 500W+ for GPUs), which has become a serious bottleneck for efficient data center
system design. Challenges such as reducing the carbon footprint and greenhouse gas emissions of data centers,
integrating energy conservation into daily operations, reducing operational expenses, and lowering PUE are
becoming key considerations for solution providers and data center owners.

From a green perspective，to build green data centers, a variety of initiatives are being taken to reduce carbon
emissions from data centers, including adopting liquid cooling for higher cooling efficiency, introducing new
energy sources such as wind and solar energy, using AI/ML technologies to upgrade operation & maintenance
management processes, and adopting new chip architectures to achieve higher computing power output per
unit of power consumption. The world's leading cloud computing players are embracing more new energy. AWS
has built solar powered and wind powered plants to power its data centers in Spain, Scotland, Sweden, Ireland,
and many other countries. Since 2010, Google has been buying renewable energy on a large scale, and it
operates 13 renewable energy projects in Europe alone. In 2019, the average PUE of Google's data centers
reached 1.1. Equinix, a leading third-party data center provider, is driving intelligent upgrades to data center
infrastructure management and adaptive control systems by deploying liquid cooling, AI/ML, and other
operational technologies, with its global annual average PUE lowering to 1.5.

There are two types of liquid cooling: direct liquid cooling (DLC) and
immersion cooling (IC). Cold plate liquid cooling is the primary form of
DLC solution, which is a direct upgrade to existing rack systems for more
efficient cooling, especially for high computing power/high power
processors. IC has a higher cooling efficiency than the cold plate
solution, but it costs more.

Of all liquid cooling solution options, IDC believes that DLC solutions will
be more widely used in the data centers, especially for enterprise users,
thanks to cost-effectiveness and scalability based on existing cooling
systems and data center infrastructure. For IC, given factors such as the
redesign/reconfiguration of data center systems, racks, floors, and other
facilities, as well as costs associated with data center construction and
operation (CAPEX/OPEX), such solutions will be more suitable for
enterprises such as cloud service providers that are already running
large-scale systems and data centers.

31
 2021-2022 Global Computing Index
Assessment Report

Liquid cooling solutions still in their infancy are mainly used in

accelerated computing systems and hyperscale data centers. As the
demand for green and energy-saving operations in data center systems
continues to increase, IDC believes that liquid cooling solutions will be
more widely adopted in the data center market, with at least 40% of
systems with high computing power in data centers to be equipped with
liquid cooling by 2023.

From a centralized perspective，one of the key factors in reducing carbon emissions is to increase the density
of computing resources for higher efficiency. IDC has found that the average PUE of enterprise data centers is
2.8, while cloud service providers are between 1.1 and 1.3. Large cloud service providers and data center
operators are also leaders in clean energy applications such as solar power, wind power, and hydropower. They
have been ranking highly on the Green Power Partnership National Top 100 list of the US Environmental
Protection Agency (EPA) for several years. When computing resources from discrete enterprise data centers are
aggregated into larger data centers, power capacity can then be managed more efficiently, cooling facilities can
be optimized, and server utilization can be improved to maximize the energy efficiency of IT resources and
achieve reductions in emissions. According to IDC's Global cloud computing carbon dioxide emission reduction
forecast, 2021-2024, continued adoption of cloud computing through 2024 could reduce carbon dioxide
emissions by more than 1 billion tons.

32
Recommendations
for Actions

33 33
 2021-2022 Global Computing Index
Assessment Report

As one of the important indicators to measure the development of a country's digital economy, computing
power plays a fundamental supporting role for countries to enhance their competitiveness under the new
technological revolution and industrial changes. Its strategic significance has been recognized worldwide. The
acquisition and enhancement of computing power requires large and continuous investment. At the same time,
there are high risks in technology development and application. These features mean that the global
competition for computing power is inevitably a competition among major countries.

Against a complex and changing international political and economic landscape, countries around the world are
moving faster to localize their computing power industry. All countries, whether they be leading, rising, or a
startup, need to outline a feasible path for building their own computing power ecosystem by taking into
account their own conditions, development stages, and resource endowments. They should also draw on the
successes of other countries in terms of top-level facilities and infrastructure for computing power development
and relevant talent cultivation, and should actively seek cooperation with other countries in digital infrastructure
and other fields.

Encourage countries to invest more in

computing power infrastructure

Developed countries generally attach more importance to investment in computing power, and large
economies' spending on computing power takes up a bigger share of their GDP. Among the 15 countries in this
report, the US spends 1.9% of its GDP on servers, while developing countries, excluding China, spend only 0.5%
on average. Through empirical analysis and theoretical model construction, the report proves that computing
power construction has a non-negligible impact on the country's macroeconomic development. Computing
power construction has become a core economic powerhouse. First, all countries should give full play to the
positive network externality effect and spillover effect of computing power investment, and invest even more in
computing power-related industries, especially data center infrastructure. For rising and startup countries,
increasing computing power comes as an important means to fuel the digital transformation of traditional
industries.

Second, efficient computing capacity, extensive application of emerging technologies, and sound infrastructure
support are why developed countries can lead in computing power. The startup countries should, on one hand,
invest more in computing power, and on the other, actively leverage various new technologies to enhance the
usage efficiency of computing power and promote the synergistic development of computing, storage, and
network infrastructure. In addition, rising and startup countries should be careful to avoid overcapacity and
over-competitiveness in data center construction. They should strike a balance between participation in
international competition and fulfilling their own demand, and avoid duplicate investment and uneven
development.

34
 2021-2022 Global Computing Index
Assessment Report

Strengthen international cooperation and shared

development in computing power

Open cooperation in the digital economy is an important part of each country's global integration. International
political risks are impacting the synergy and cooperation of countries in the computing power industrial chain.
The leading, rising, and startup countries should invariably seek cooperation with other countries in the digital
economy, especially in promoting international cooperation and shared development of computing power.

International cooperation faces many practical challenges in cross-border information sharing, localization of
computing power infrastructure, international regulation and governance of the Internet, etc. That said, when
the leading countries are shaping their own computing power ecosystems, they must undertake the
responsibility of building a new mechanism to drive globalization through the digital economy, promoting the
formation of a diversified and multi-level international cooperation and exchange mechanism in computing
power, creating an international open-source platform, participating in the formulation of international
standards & data governance mechanisms, strengthening international training & research cooperation of
computing talent, guiding the green development of the computing industry, working together to provide digital
infrastructure public goods for the world, and improving the global digital economy governance system.

35
Appendix

36
 2021-2022 Global Computing Index
Assessment Report

Methodology

The Global Computing Index is calculated by weighting four sub-items: computing capacity, computing
efficiency, application level, and infrastructure support. The score of each sub-item is calculated by weighting
the scores of the indicators for the sub-item. The specific score of each indicator is calculated as follows:

• The total score for each indicator is 100 points.

• For each country, the score of an indicator is calculated based on the comparison between the value for the
current year and the target value for 2025. If the actual value for the current year is the same as the target value
for 2025, the score for this indicator is 100 points. If the actual value is different from the target value for 2025, the
indicator data is standardized and converted to the score based on the target value for each country. The target
value for 2025 is calculated based on the data from authoritative organizations, institutions, and enterprises in
various industries.

Score of Sub-item A = Score of a1 * Weight of a1 + Score of a2 * Weight of a2 + Score of a3 * Weight of a3 + ... +

Score of aN * Weight of aN

The data sources of this research include IDC, EIU, IMF, the World Bank, International Telecommunication Union
(ITU), China Academy of Information and Communications Technology (CAICT), Ookla, etc. In this research,
telephone surveys are carried out for a total of 900 samples across 15 countries and regions. This research is
based on the data from various industries, such as telecom, Internet, finance, manufacturing, energy,
transportation, public utilities, public affair, medical care, retail, and professional services, ensuring that the
first-hand research information is obtained.

The global Computing Index is a comprehensive, extensive, and complete indicator system that adopts a sound
model system and credible data sources. The index provides organizations and individuals with a
comprehensive blueprint for analyzing the relationship among computing power, the economy and the digital
economy. The overall ranking of the global Computing Index scores reflects the current status of the global
economy and computing power, and provides suggestions for the development and evolution of ICT in the next
decade.

In this research, regression analysis is performed on the Computing Index and the economic development to
obtain the degree of impact between them. In addition, the Computing Index of different countries and regions
are categorized for presentation, and suggestions for future development are provided.

From the quantitative perspective, IDC performs linear regression to analyze the relationship between the
Computing Indexes and economic indicators. In the linear regression, it is assumed that the relationship
between two variables can be represented by one straight line and reflected in the formula y = a + b(x):

37
- y represents a dependent variable and can be understood as the predicted variable.

- x represents an independent variable and can be understood as a predictor variable.

- b represents the inclination of the straight line. The magnitude of b represents the change in y caused by each
unit change in x. In other words, when x increases by 1 unit, y changes by b units.

- a represents the y-intercept. In other words, a is the point where the straight line intersects the y-axis when x is 0.

The value of y is the year-on-year growth rate of the GDP or digital economy for the country. The value of x is the
year-on-year growth rate of the Computing Index for the country. The value of b represents the change in the
total GDP/digital economy of the country caused by each unit of change in the Computing Index. In this analysis,
the reliability of the results is tested. according to the normal distribution and the 2σ rule. If the 95% confidence
interval is between (-2σ, 2σ), the analysis results are 95% reliable. In other words, the analysis results fully explain
at least 95% of the samples and are consistent with the actual situation reflected by the samples.

Data Specifications

The model for evaluating the Global Computing Index consists of four Level-1 indicators and 20 Level-2
indicators. The model is used to evaluate the development level of the computing power and the potential for
future development of the 15 key countries around the world. The four Level-1 indicators include computing
capacity, computing efficiency, application level, and infrastructure support. The indicators are independent of
each other, but are also closely correlated. The computing capacity reflects the theoretical maximum computing
power of a country. It is the core of the entire evaluation system and occupies the highest weight among the four
indicators. The computing efficiency reflects the current level of computing power utilization of the country.
Some countries have higher utilization in cloud computing and virtualization, and therefore are more efficient in
using their computing power. The application level is complementary to the computing capacity and the
computing efficiency. The computing capacity and efficiency of a country are fundamental to the
implementation of emerging technologies. In addition, the application of emerging technologies greatly
promotes the development of the computing power of a country in the future. Therefore, the application level
can reflect the development potential of the computing power of a country. Infrastructure support provides
guarantees for the computing capacity, computing efficiency, and application level at a more macro level. Only
when a country achieves balanced and synergistic development in these four dimensions can the country
increases the overall Computing Index and gain greater economic benefits.

38
 2021-2022 Global Computing Index
Assessment Report

The computing capacity covers general computing, AI computing, edge computing, scientific computing, and
terminal computing. It indicates the overall level and focus of computing power investment in different
countries by evaluating the number and proportion of servers and terminal equipment. For general computing,
AI computing, and edge computing, statistical specifications include the number and spending ratio of servers,
AI-based servers, and edge computing servers. For scientific computing, the number and ranking of
supercomputers of a country in the list of global top 500 supercomputers are evaluated and the comprehensive
scientific computing power of the country is calculated using weighted algorithms. For terminal computing,
statistical specifications include the number and spending of smartphones and PCs.

The computing efficiency reflects the current utilization of computing power in terms of the utilization of the
CPU, memory, storage, new technologies, and the penetration of cloud computing. Virtualization, the basis of
cloud computing, pools resources to implement on-demand computing, improving the utilization efficiency of
software and hardware resources. Therefore, the penetration of cloud computing plays a crucial role in
improving the utilization of existing computing capabilities. This indicator has a higher weight in the computing
efficiency evaluation.

The application level focuses on the application of several emerging technologies such as AI, big data, IoT,
robotics, and blockchain in different countries. Among these five emerging technologies, AI, IoT, and big data are
most widely applied in industries, and therefore have higher weights in this evaluation.

Infrastructure support measures the sustainability of the future development of the computing power of a
country in various dimensions, including the data center scale, software and services of data centers, power
usage effectiveness (PUE) of data centers, storage infrastructure, and network infrastructure. Server spending
accounts for about 70% of the overall hardware spending and is the core of infrastructure spending. The balance
between storage and network spending is also important for the future development of computing power.
Therefore, storage and network infrastructure are fully evaluated in this indicator.

39
 2021-2022 Global Computing Index
Assessment Report

Level-1 indicator Level-2 indicator Level-3 indicator Statistical specification

General computing Server spending1 Server spending / GDP of the country

capacity for the current year

Scientific computing Number of global top 500 Weighted score of the ranking of all
capacity supercomputers and their supercomputers of the country in the
rankings global top 500 supercomputers

AI computing capacity AI server spending1 AI server spending / GDP of the

Computing
country for the current year
capacity
Terminal computing Smartphone and PC Smartphone and PC spending / GDP
capacity spending1 of the country for the current year

Smartphone and PC user Smartphone and PC user base

base2

Edge computing capacity Edge computing hardware Edge computing hardware spending
spending1

New technology utilization New technology utilization1 Average utilization of new

technologies (SSD, SCM, and
heterogeneous storage)

Cloud computing Cloud computing Cloud computing spending / GDP

penetration penetration of the country for the current year
Computing
efficiency CPU utilization Average utilization of Average utilization of server CPU
server CPU

Memory utilization Average utilization of Average utilization of server memory

server memory

Storage utilization Average utilization of Average utilization of server storage

server storage devices devices

Big data Overall spending of Spending / GDP

software, hardware
AI and services
IoT
Computing
application level
Blockchain

Robotics

Data center software Data center software Data center software and service
and services and service spending spending
Data center scale Number of hyperscale Number of hyperscale data centers
data centers (with over 10,000 server racks)

Data center efficiency Average data center PUE Average data center PUE

Network infrastructure Network spending1 Network infrastructure spending / GDP

Infrastructure of the country for the current year
support
Telecommunication Telecommunication spending / GDP
spending2 of the country for the current year

Storage infrastructure Storage device spending1 Storage device spending / GDP

Shipments of storage Shipments of storage devices (TB)

devices (TB)2

• GDP: the GDP calculated by International Monetary Fund (IMF).

• If level-2 indicators include multiple level-3 indicators, 1 means the primary scoring items, and 2 means reference item.

40