The GenAI Leap

HOW NETWORKS ARE REINVENTING THEMSELVES

75
leading players in the data
and connectivity space
explore how GenAI is transforming
telecom and enterprise networks,
from managing network performance
to enhancing customer experience,
mitigating threats and fraud, and
improving efficiency.

Special edition eBook

 163 FIXED NETWORKS
INDEX
175 QUANTUM COMPUTING

181 INTERNET OF THINGS

03 FEATURED 190 PUBLIC SAFETY

19 AI STRATEGIES 200 WIFI FEATURED

51 AUTONOMOUS NETWORKS 206 VIDEO
1. Five Back-to-the-Future Predictions for AI in 2025�����������������������������������������������������������������������������������������.4

75 CUSTOMER EXPERIENCE 215 SUSTAINABILITY 2. The Duality of GenAI: Protecting Consumers from Phone Fraud While
Raising Brand Awareness���������������������������������������������������������������������������������������������������������������������������������������������������������.9

93
3. From Networks to Support Systems: What Changes Will GenAI Bring
DATA CENTERS to Operators?��������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������.12

4. Three GenAI Predictions That Will Impact Telecom Operations in 2025���������������������������������.15

112 SECURITY

132 EDGE & CLOUD

142 MOBILE NETWORKS

3
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

FIVE BACK-TO-THE-FUTURE PREDICTIONS FOR AI #1: LARGE SCALE ADOPTION OF SMALL SCALE
MODELS
memory requirements, and consume
far less energy.

IN 2025 Large language models (LLMs) like

GPT-3/-4, Gemini, Llama, Claude,
Mixtral, etc. have enabled astonishing
Accordingly, we are likely to see much
wider adoption of SLMs in 2025.
And, in some of the methods used to
breakthroughs in generative AI. While improve results with both SLMs and
Written by Laura Wilber each has unique strengths with certain LLMs – namely fine-tuning and RAG
Senior Industry Analyst, Enea types of output (e.g., text, images, audio, – we can see the contours of another
video, or code), they share a common 2025 trend: a shift from a model-centric
weakness: a ravenous appetite for approach to a more systems-centered

I n the 1980s sci-fi/comedy movie

hit “Back to the Future,” a teenager
accidentally travels back in time and
address fundamental AI issues like se-
curity, governance, resource consump-
tion and social impacts. Fortunately,
These five trends are:
1. Increased adoption of small
language models,
compute power, memory, training data,
and energy.
approach in the development of lan-
guage model-based AI applications.

prevents his parents from falling in there are some emerging trends that 2. Pivot from model thinking to Another challenge LLMs share is their #2: A SHIFT FROM MODEL THINKING TO
love, thus dooming his own existence tap into past wisdom and best prac- systems thinking, generalized nature. While this a key SYSTEMS THINKING
unless he can set things right and get tices in ways that can help us better 3. Sharper focus on AI agent security, LLM strength, it also makes adapting
back to the present. In the world of AI, navigate the current challenges and 4. Greater recognition of humans as them to specific needs challenging. In early 2024, some were already
many are concerned that while our future opportunities of AI in enterprise ‘differentiators,’ and Developers often use extensive manual noting this shift, like the Berkeley
existence may not be threatened, we and telecommunications connectivity 5. Renewed emphasis on data quality. adjustments to prompts to adapt an Artificial Intelligence Research (BAIR)
are charging into the future at a speed and security. LLM to their requirements, which can group. They observed that while it was
that is outpacing our ability to properly lead to deep, even total, model de- understandable that the enthusiasm
pendence for AI applications. Other unleashed by GPT-3 and ChatGPT in-
methods used include fine-tuning spired developers to place LLMs at the
results using human feedback loops, center of AI application development,
and improving relevance and accuracy the focus had (rightly) begun to shift to
through run-time queries to authori- the state-of-the-art results “increasing-
tative external datasets (i.e., Retriev- ly obtained by compound systems with
al-Augmented Generation, or RAG). multi-components, not just monolithic
models,” [2] which one BAIR member
Another strategy altogether is to start later stated is an approach “that can
with a small language model (SLM) produce better results in just about
instead of an LLM, even if one still every way they can be measured.” [3]
applies fine-tuning and RAG. The SLMs
currently being released by both start- Others seemed to remain more

“
ups and the LLM titans are producing firmly ensconced in a model-centered
comparable results to LLMs while be- mindset in 2024, but sensed change While each
ing trained using far fewer parameters was coming. For instance, at OpenAI
and datasets that are much smaller but DevDay in November 2024, when
LLM has
of high quality. [1] OpenAI CEO Sam Altman was asked unique strengths with
about considerations for choosing one certain types of output,
This makes it easier to adapt SLMs LLM over another for an AI application,
they share a common
to specific objectives and constraints, he responded “Developers use multiple
and it reduces (although does not models most of the time and I’m not weakness: a ravenous
eliminate) risks like model dependence sure how that’s all going to evolve as appetite for compute
and hallucinations. Importantly, it also we head towards this more agentified power, memory, training
means they have much more modest world, but...something about the way
resource needs: they can function that we currently talk about it or think
data, and energy.
on standard CPUs, have much lower about [AI] feels wrong. Maybe, maybe

4 5
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

if I had to describe it, we will shift from secure. [9] Unfortunately, the research- be, we can expect to hear more in So in 2025, don’t be surprised if could dramatically improve a model’s cats?” [15]
talking about models to talking about ers found that the potential side effects 2025 about initiatives to aid workers in some current conversations about output, without any algorithmic chang-
systems.” [4] and dangers of the LLM agents studied adapting to the agentic AI era. And, if how AI can ‘augment’ humans is es. [14] This general quality-over-quantity
were not addressed by the frameworks current trends we are seeing at Enea flipped to how humans can improve direction also explains the increase in
A systems approach to LLM-based used to build them, nor by research are an indication, we may even see and extend AI to create a true com- LLMs, however, loosened the GIGO demand we are seeing for our deep
application development employs aimed at improving them. human workers increasingly positioned petitive advantage. rule slightly. Early LLMs were able to packet inspection software library.
the same basic elements systems as a key business differentiator. achieve leaps in qualitative perfor- While the types and number of sources
development has always used, like This finding is echoed in “AI Agents #5: THE (RE) CORONATION OF DATA QUALITY mance largely due to brute computing producing network-traffic related data
modular design, generic architectures, Under Threat,” which provides a com- For instance, customers place a high AS KING power and staggering volumes of train- have exploded, much of this data is of
and data-driven optimization. And it prehensive view of LLM agent threats value on Enea’s use of AI to address ing data – namely the entire corpus of medium to low quality and vulnera-
provides a valuable layer of abstraction across 100+ research papers. [10] The challenges like optimizing message Another factor likely to differentiate the Internet. Some worried progress ble to tampering. [16] Hence the gold
by wrapping language model pro- study concludes the security challeng- delivery costs through categorization products and services in 2025 is data would taper off as the available stock standard of deep packet inspection is
gramming frameworks around prompt es associated with AI agents remain (e.g., 2FA versus promotional messag- quality. While the adage “garbage in, of untapped high-quality data (i.e., proving to be a pivotal tool for develop-
engineering. [5] largely under-explored and unresolved. es), blocking malicious links embedded garbage out” has served as a golden human-generated content) on the ing successful AI applications, and we
in messages in real-time, and enabling rule in computer science since the Internet leveled off. expect this year’s double-digit growth
With the use of AI agents expected It is therefore logical to expect that as regulatory compliance by performing 1950s, it is more relevant than ever in to continue in 2025.
to surge in 2025 [6], this more formal agent use proliferates, so will agent-re- AI-based restricted image detection in our contemporary AI environment. However, returning to Trend 1, the train-
systems approach offers a timely lated security incidents, and activity in A2P messages. At the same time, there ing of smaller models with (relatively) And thus, tapping into back-to-the-fu-
advantage in terms of security and AI Agent security will intensify. And to- has been an uptick in the number of The GIGO rule posits that program- small but high-quality training data is ture traditions like using high-quality
governance as well, and it has a strong day, a debate is already taking shape as customers who tell us that a key factor ming results are inextricably dependent yielding comparable, and sometimes data, following best practices in sys-
historical foundation for developers to to whether existing cybersecurity tools in choosing Enea is the global team of on the quality of the data used to build better, results than LLM-based appli- tems design and information security,
draw on. (e.g., identity and access management deeply experienced human experts in and operate them.In pre-LLM era ma- cations. And so it seems Amir Khos- and nurturing your team’s human ad-
(IAM), zero trust, anomaly detection, our signaling and messaging security chine learning, it was well understood rowshahi (then Intel AI CTO) is getting vantage, will all have a hand in shaping
#3: A HEIGHTENED FOCUS ON AI AGENT observability, etc.) can be adapted to threat intelligence units (TIUs). that data quality had such an important an answer to the question he posed in what is sure to be a successful year
SECURITY incorporate agent security, or whether effect on the quality of ML results that 2018: “Do you really need to see 100 ahead for AI.
new, specialized tools and techniques Similarly, in our embedded security enhancing input data quality alone million cats to learn how to identify
Broadly speaking, AI agents are will be required. [11] division, we work with the most inno-
“artificial entities that sense their vative enterprise cybersecurity vendors Sources:
environment, make decisions, and Beyond security, another source of se- in the business, and to a one, they are [6] See, for instance, data points here https://tomtunguz.com/small-but-mighty-ai/ (noting the 30x cost differential between the largest and
take actions.” [7] When people today rious concern is whether AI agents will achieving superb results by pairing smallest models). For data on a representative SLM from a large vendor (Phi-4 from Microsoft), see: https://techcommunity.microsoft.com/blog/
talk about AI agents, they are usually augment or replace workers, and we their multi-AI agent systems with exter- aiplatformblog/introducing-phi-4-microsoft%E2%80%99s-newest-small-language-model-specializing-in-comple/4357090
talking about LLM-based AI agents, can expect this issue to receive more nal, human-curated threat intelligence [7] “The Shift from Models to Compound AI Systems.” https://bair.berkeley.edu/blog/2024/02/18/compound-ai-systems/ Berkeley Artificial
but general AI agent design and de- attention too in 2025. services and expert in-house security Intelligence.
velopment stretches back to the 1950s analysts. [8] Christopher Potts, Stanford Webinar “Large Language Models Get the Hype, but Compound Systems Are the Future of AI,” https://www.
in the AI subfield of ‘planning AI’ (or ‘AI #4: HUMANS AS ‘DIFFERENTIATORS’ IN AN youtube.com/watch?v=vRTcE19M-KE
planning’), and the 1980s in the subfield AGENTIC WORLD In any case, we humans have an [9] Sam Altman OpenAI DevDay fireside chat: https://www.youtube.com/watch?v=-cq3O4t0qQc
of Distributed Artificial Intelligence advantage vis-à-vis our AI counter- [10] Short survey of the LLM application framework landscape: https://www.infoworld.com/article/3617664/surveying-the-llm-application-
(DAI) and related subdiscipline of Mul- In 2020, the World Economic Forum parts (at least for now) in being able framework-landscape.html
tiagent Systems (all generally intended (WEF) forecast workforce automation to generalize far beyond our direct [11] For example, Gartner has named Agentic AI as its Top Strategic Technology Trend for 2025 (https://www.gartner.com/en/articles/top-
to support industrial automation). [8] would displace 85 million jobs by 2025, experiences, and benefiting richly from technology-trends-2025), which is an expectation shared by tech titans like Google, Amazon, Meta, OpenAI, Anthropic, Salesforce, Microsoft
counterbalanced by a ‘robot revolution’ physical and emotional inputs in learn- and Nvidia, who closed out 2024 with a flurry of major Agentic AI announcements.
As noted in the research paper “Se- which would create 97 million new ing and reasoning. We can also profit [12] “The Rise and Potential of Large Language Model Based Agents: A Survey,”https://arxiv.org/pdf/2309.07864
curity of AI Agents,” securing LLM jobs. [12] More recent studies are less handsomely from serendipity, and even [13] “Multiagent Systems: A Survey from a Machine Learning Perspective,” https://www.cs.utexas.edu/~pstone/Papers/bib2html-links/MASsurvey.
agents is more complex than securing clear in forecasting a net job loss/gain humor, in approaching certain analyt- pdf
non-LLM agents, which operate on ratio for AI, though they are fairly pre- ical and creative tasks, and can forge [14] “Security of AI Agents,” https://arxiv.org/pdf/2406.08689
pre-defined rules or reinforcement scriptive about which roles will be most the kind of human-to-human bonds of [15] AI Agents Under Threat: A Survey of Key Security Challenges and Future Pathways https://arxiv.org/pdf/2406.02630
learning models, which makes them affected. [13] trust, loyalty and goodwill upon which [16] For an example of adapting existing approach, here API security, to AI agent system security, seehttps://www.databricks.com/blog/
harder to generalize to new tasks and business relationships are built. ai-gateways-secure-ai-agent-systems. It is also interesting to note that the often cited first elaboration of the “zero-trust” concept was a
different tools, but easier to control and Whatever the balance will ultimately dissertation on AI planning agent collaboration: Stephen Paul Marsh, “Formalising Trust as a Computational Concept,” https://dspace.stir.ac.uk/
bitstream/1893/2010/1/Formalising%20trust%20as%20a%20computational%20concept.pdf

6 7
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

[17] https://www.weforum.org/press/2020/10/recession-and-automation-changes-our-future-of-work-but-there-are-jobs-coming-report-says-
52c5162fce/
[18] See for example, https://www.mckinsey.com/mgi/our-research/a-new-future-of-work-the-race-to-deploy-ai-and-raise-skills-in-europe-and-
beyond
[19] https://www.enea.com/insights/rich-network-traffic-data-for-machine-learning-ml-artificial-intelligence-ai/

THE DUALITY OF GENAI: PROTECTING

[20] https://venturebeat.com/ai/intel-ai-boss-its-time-to-move-from-brute-force-to-more-efficient-computing/
[21] See the“2024 State of Network Threat Detection” survey report at https://info.enea.com/network-threat-detection-report-pdf and on demand
webinar at https://www.youtube.com/watch?v=Y-Ds54lihYM

CONSUMERS FROM PHONE FRAUD WHILE

RAISING BRAND AWARENESS
Written by Jonjie Sena
VP of Product Management, TransUnion

G enerative Artificial Intelligence

(GenAI) is helping revolutionize
all types of businesses through its
When it comes to GenAI for outbound
voice communications, it’s no different.
Deepfake incidents in the fintech
sector alone increased 700% in 2023
from the previous year, according to a
ability to streamline tasks, enhance While the phone remains a popular tool recent report by Sumsub. And accord-
productivity and improve the customer with consumers, especially for person- ing to Deloitte’s Center for Financial
BY LAURA WILBER
experience. However, its vast potential al, high value or urgent situations, close Services, overall, GenAI could enable
Laura Wilber is Senior Industry Analyst at Enea. She brings with it security risks. Business- to 90% of calls to customers don’t fraud losses to reach $40 billion in the
supports cross-functional & cross-portfolio teams es recognize the need to balance the get answered. Our survey showed United States by 2027, up from $12.3
with technology and market analysis, product mar- benefits of GenAI with issues of data more than half of consumers prefer billion in 2023 The financial sector is
keting, product strategy and corporate development. privacy, intellectual property theft, the phone to discuss matters that are most at risk.
She is also an ESG Advisor & Committee Member. Her deepfakes, misinformation, model ma- personal or high value decisions, such
expertise includes cybersecurity and networking in nipulation and enhanced cyber threats. as retirement planning, and/or urgent As businesses battle hurdles to reach
enterprise, telecom and industrial markets, and she situations. But 74% have declined to customers and fraudsters double down
loves helping customers meet today’s challenges answer the phone as they assume it on the use of GenAI to commit in-
while musing about what the next ten years will could be a scam. Consumers have creasingly sophisticated phone scams
bring. reason to be cautious. like deepfakes that result in significant
financial losses for both enterprises
and consumers, they’ll be paying close
attention to how to leverage GenAI to
both enhance engagement and combat
fraud.

ABOUT ENEA

Enea is a world-leading specialist in software for telecom and cybersecurity. Our cloud-native solutions connect, optimize, and secure services for mobile subscribers,
enterprises, and the Internet of Things. More than 100 communication service providers and billions of people rely on our technology every day when they connect to mobile
networks or use the Internet.

8 9
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

#1: AI AGENTS WILL SPEED RESOLUTION AND #4: MACHINE LEARNING, BIOMETRICS, diagnose fraud and send the inves- There’s no single solutions to re- There are solutions available for busi-
IMPROVE CX AUTHENTICATION AND ANALYTICS WILL HELP tigations to the appropriate team at ducing the negative impact of bad nesses today that can help businesses
BUSINESSES COMBAT FRAUD the bank. Some banks are already actors leveraging GenAI to commit increase engagement and right-party
Because AI agents will be able to un- • Advanced Call Filtering and incorporating large language models to fraud. Keeping the telecom ecosys- answer rates and reduce the number
derstand and respond to more complex Detection: Enhance call filtering detect signs of fraud, such as one used tem safe for businesses and con- of calls that get mistagged as spam,
queries, provide personalized help, and systems by analyzing patterns by JPMorgan for email compromises. sumers will require collaboration while providing end-to-end authentica-
tap into historical data to predict needs, and behaviors associated with Similarly, Mastercard is working to pre- between regulators, communica- tion so spoofed calls don’t even reach
they’ll be able to deal with a much robocalls and spoofing attempts. vent credit card fraud with its Decision tions service providers (CSPs) and consumers. At TransUnion, we offer
wider range of customer service issues. By leveraging machine learning Intelligence tool, which scans a trillion solution providers to keep pace with a suite of Trusted Call Solutions that
The use of AI agents can significantly algorithms, solutions can identify data points to predict if a transaction is sophisticated fraudsters. help businesses connect securely with
cut down the need to request human and block suspicious calls in genuine.” customers, protecting them from fraud
intervention, leading to faster resolu- real-time, reducing the number while promoting their brand so they
tion times and an enhanced customer of fraudulent calls that reach can transact with confidence.
experience. Because customers often consumers.
require a nuanced approach to address • Voice Biometrics and
many issues, we expect AI agents will Authentication: Improve voice
work in tandem with humans. biometric systems used for caller
authentication. By analyzing
#2: PROACTIVE OUTBOUND unique voice patterns, GenAI can
COMMUNICATIONS WILL BE TAILORED help verify the identity of callers,
making it difficult for fraudsters
Through analysis of customer data, to impersonate legitimate users.
GenAI can also easily help tailor mar- This technology can be integrated
keting calls, reminders and follow-ups into call centers and financial
to individual customer preferences and institutions to prevent unauthorized
behaviors. Key applications for Ge- access.
nAI include healthcare (personalized • Behavioral Analytics: Utilize
treatments), marketing (custom content behavioral analytics to detect
creation), financial services (fraud anomalies in call patterns and user
detection) and entertainment (content interactions. By monitoring how BY JONJIE SENA
creation). users interact with their devices
Jonjie Sena is VP of Product Management at
and services, GenAI can identify
TransUnion, responsible for driving the go-to-market
#3: BUSINESSES WILL LEVERAGE REAL-TIME unusual behaviors that may indicate
strategy for Contact Center and Communications
LANGUAGE TRANSLATION AND SENTIMENT fraudulent activity. This approach
Solutions. Jonjie and his team focus on helping
ANALYSIS helps in detecting and preventing
businesses overcome the impacts of the robocalling
fraud before it can cause significant
epidemic and restoring trust in phone calls. Jonjie
This helps break down language barri- harm.
has over twenty years’ experience leading product
ers and enables seamless communica-

“
strategy, technology and marketing teams to launch
tion with customers globally. Sentiment Many businesses are already taking
and drive market adoption of solutions for communi-
analysis can be integrated into out- action to address the growing fraud Because
cations service providers and enterprises. He holds
bound communications, allowing AI to problem, especially financial institu- customers multiple patents in personalized and authenticated
gauge customer emotions and adjust tions where the risk is highest.
often require a nuanced communications. Prior to joining TransUnion, Jonjie
the conversation accordingly, foster-
approach to address held executive roles as VP Product Marketing at
ing more empathetic and contextually Deloitte notes, “...financial institutions
Neustar, VP Marketing at TEOCO, and VP Products &
appropriate responses. today are commonly deploying artifi- many issues, we expect
Technology at ACE*COMM.
cial intelligence and machine learning AI agents will work in
tools to detect, alert, and respond to
threats. For instance, some banks are
tandem with humans.
using AI to automate processes that

10 11
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

FROM NETWORKS TO SUPPORT SYSTEMS: WHAT and autonomous control. These inno-
vations will enable telecom networks
network production support workflows.
This transformation will unfold across
Whale Cloud, as a leading digital
transformation solutions provider, is

CHANGES WILL GENAI BRING TO OPERATORS?

to deliver high-quality, predictable several dimensions, each contributing revolutionizing telecom operations
connectivity services, transforming to a more streamlined and efficient by embedding AI Agents into tra-
customer experiences. telecom industry: ditional OSS systems. This innova-
• Human-Computer Interaction: tion powers OSS Co-pilots to help
#2: COMPREHENSIVE INTELLIGENTIZATION Enhanced interfaces will allow operators address these challenges
Written by Ye Qindan OF SUPPORTING SYSTEMS support staff to issue natural effectively.
Deputy Manager of Commercial Development Department of Domestic Business Operation Center, Whale Cloud language commands through voice
As artificial intelligence overcomes or text. Systems will autonomously The OSS Co-pilot shifts telecom oper-
conventional technology barriers, tele- perform information extraction, ators from reactive, manual processes

T he inherent intelligence of telecom

network infrastructure is emerg-
ing as a defining characteristic of
innovation across supporting systems,
operations, and production processes.
These advancements will unlock trans-
#1: HIERARCHICAL EVOLUTION OF NATIVE AI
CAPABILITIES IN TELECOM NETWORKS
com networks will progressively move
into a stage of intelligent application.
This evolution spans five critical dimen-
execute queries, and complete
business operations, fostering more
intuitive and efficient interactions.
to proactive, autonomous network
management. By harnessing advanced
AI capabilities, it drives improvements
next-generation networks. This trend formative opportunities, reshaping the Artificial intelligence is catalyzing a sions: planning, construction, mainte- • Work Modes: Intelligent assistants in operational efficiency, network resil-
is set to unfold progressively in 5.5G telecom industry. paradigm shift in network intelligence, nance, optimization, and operations. will serve as co-pilots, providing ience, and service reliability. The solu-
networks, steering the evolution toward evolving from the traditional “Network The key changes are: real-time task decomposition, tion automates routine tasks, minimizes
intelligence. Whale Cloud envisions three pivotal + AI” approach to a more integrated “AI • Wide Adoption of AI Co-pilots: guidance, and predictive insights manual errors, accelerates incident
changes that the development of artifi- + Network” model. This paradigm shift AI co-pilots are becoming integral while facing complicated tasks. response, and empowers operators to
With the advancement of Generative cial intelligence technologies will bring represents the foundation for AI-driven to tasks like customer service and This will enhance the efficiency achieve self-managing, self-healing,
AI (GenAI) technologies, operators to operator networks and supporting networks, where intelligence is embed- maintenance, significantly boosting and accuracy of complex network and adaptive networks—meeting the
are positioned to drive significant paradigms: ded in the fabric of telecom operations. operational efficiency. For instance, operations and maintenance tasks. evolving demands of modern telecom
China Mobile spearheaded the • Work Content: With GenAI landscapes.
development of 13 AI+ scenario automating repetitive tasks, support
applications in 2024. China Telecom personnel can focus on higher- With its comprehensive AI-driven solu-
plans to launch 100 large model value activities such as strategic tions, Whale Cloud empowers telecom
applications by 2025, covering collaboration with intelligent operators to confidently transition
all facets of network production systems, reducing reliance on to “AI + Network” operations. From
support. manual operations. intelligent infrastructure and advanced
• Scenario-Based Intelligent • Process System: The continuous large-model platforms to scenario-spe-
“Agents”: Operators are advancing evolution of GenAI, coupled cific digital applications, Whale Cloud
to increase the proportion of with the integration of intelligent unlocks the full potential of GenAI,
unsupervised intelligence in scenarios, will drive a systematic enabling telcos to deliver exceptional
processes, allowing for autonomous overhaul of production processes. customer experiences. This transforma-
operations in specific scenarios. By This shift will enable operators tion paves the way for a smarter, more
considering multiple dimensions to achieve more intelligent, efficient, and future-ready telecom
such as cost, benefits, and risks, streamlined, and efficient industry.
autonomous business scenarios workflows.
are most likely to be incubated Conclusion
in network operations and #4: WHALE CLOUD’S ROLE IN
optimizations in the short term, TRANSFORMING TELCOS WITH AI + Despite challenges such as reaching
achieving higher efficiency in NETWORK the peak of the Scaling Law effect
The intelligent transformation of oper- intelligence, cross-domain intelligence, network operations. and the cost pressures associated
ator networks will follow the existing and cloud-network integration. Achieving large-scale “AI + Network” with intelligent computing operations,
hierarchical network architecture, un- #3: REINVENTION OF NETWORK SUPPORT transformation requires overcoming the long-term trajectory of artificial
dergoing systematic enhancements in Key capabilities of native AI-powered WORK PARADIGMS critical challenges, including integrating intelligence remains unchanged. While
a structured and independent manner networks include real-time awareness, intelligent capabilities into multi-lay- future progress may experience tempo-
across four distinct levels, including digital twin simulations, intelligent de- The integration of intelligent appli- ered network operations and ensuring rary hurdles, AI will continue to propel
network element level, single-domain cision-making, dynamic orchestration, cations and new technologies will seamless automation of supporting telecom networks toward greater
fundamentally reshape operators’ systems. intelligence.

12 13
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

This evolution will persist in optimizing

and reshaping network support tools
experience, and technological innova-
tion, is set to become an integral com-
but also unlock unprecedented op-
portunities for growth and innovation, THREE GENAI PREDICTIONS
THAT WILL IMPACT TELECOM
and workflows, driving the industry ponent of the telecom industry’s future. shaping a smarter and more connected
forward in its ongoing transformation. Through the adoption of AI-driven world.
Generative AI, with its potential to en- strategies and solutions, operators will

OPERATIONS IN 2025
hance operational efficiency, customer not only overcome existing limitations

Written by John Denemark

Senior Vice President, Carrier Provisioning, TransUnion

F or close to a decade, communica-

tion service providers (CSPs) have
been working towards streamlining
questions about how they can harness
the power of AI to automate tasks and
hyper-personalize services. According
products and experiences, the benefits
of GenAI are tangible and far-reaching.
For instance, GenAI-enabled customer
their back-office operations, reducing to BCG’s 2024 Telco Value Creators service may at some point reduce ser-
costs and enhancing the customer Report, “GenAI presents a massive vice interactions by 50% or more, while
experience as part of their ‘digital opportunity to the telco industry. It’s improving customer satisfaction.”
transformation’ efforts. Now, generative not merely a technological advance;
AI has come onto the scene, caus- it can catalyze a profound transfor- With a focus on the transformative
ing nearly every CSP to rethink their mation along the entire value chain. power of GenAI and automation,
operational roadmap and ask strategic From reducing costs to creating new these are my predictions for 2025.
BY YE QINDAN

Ye Qindan is Deputy Manager of Commercial Devel-

opment Department of Domestic Business Operation
Center at Whale Cloud, and Senior Specialist of Com-
puting Power Network. Ye Qindan, a certified expert
with over a decade in telecom BSS and OSS, plays
a key role in developing computing power network
standards at CAICT. He successfully led the delivery
of significant projects at the group level for various
operators in China, thereby gaining invaluable practi-
cal knowledge and hands-on experience.

ABOUT WHALE CLOUD

Whale Cloud is a leading technology company that specializes in the digital transformation of telecommunication service providers and other verticals. We provide
mission-critical solutions and services in Telecom, FinTech, and E&G sectors. At Whale Cloud, we’re passionate about creating frictionless experiences for customers and
empowering their digital ambitions.

14 15
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

#1: THERE WILL BE WINNERS AND LOSERS IN requires data – and lots of it. director analyst at Gartner points out is the same as it was in 1990, it’s clear process of buying and selling whole- for immediate, on-demand access to
THE GEN-AI RACE that, “with gen-AI capabilities, there something’s broken. For telcos to sale services must be repaired now services, and you have a recipe for
Following the adage of ‘garbage in – is an increased demand for intelligent remain competitive, it’s critical that if we want the telecom ecosystem to frustration and unhappy customers.
Fortunately for those CSPs that have garbage out,’ for CSPs to fully harness automation platforms that use analysis every part of the delivery chain is in benefit from the likes of GenAI and
been buckling down on their digital the capabilities of generative AI, they and automation to deliver improved peak performance. It’s something that automation. The time for action is now
transformation checklist, generative AI must ensure they have the right tools operations, efficiencies, and insights.” I believe will become table-stakes for
can be greeted as a huge transforma- and processes in place for maintaining any successful CSP, because if it’s not, Believe it or not, many CSPs still rely Generative AI holds immense potential
tive power that will allow them to expe- accurate, up-to-date data across the Having these platforms or agents han- customers will just go elsewhere. on emails and phone calls to manage to transform telecom operations, ad-
rience significant productivity gains. organization. This is critically important dle some of these routine needs can critical wholesale ordering functions. dressing longstanding challenges while
as more and more processes become boost productivity across the organi- Key challenges facing wholesale This consumes valuable resources, paving the way for new opportunities.
According to a recent McKinsey survey, automated, with less human oversight. zation. telecom must be addressed creates lengthy delays, and increases As the industry continues to evolve,
over 72 percent of companies are the likelihood of errors that can lead it’s imperative for telecom leaders to
actively deploying AI solutions, with a #2: THE RISE OF AI-AGENTS AND #3: BUSINESS BOTTLENECKS WILL (FINALLY) The wholesale ordering process is just to dissatisfied customers. Much of this embrace this technology and integrate
growing interest in generative AI. Some INTELLIGENT PLATFORMS DISAPPEAR one example of an area that is espe- is due to a lack of industry standards it into their operational frameworks. But
telcos are already seeing cost reduc- cially ripe for digital transformation. and interoperability between carriers success in the digital age requires a
tions from generative AI across multiple The next level of Generative AI will While CSPs have been digitally trans- Over the years it’s become a critical in terms of how services are bought commitment to innovation that extends
domains, including customer service, be ‘AI Agents,’ which will become real forming their networks for the past de- bottleneck that all too often brings and sold. Pair this with rising enterprise beyond the network itself, encom-
network, IT, marketing and sales, and business enablers. AI Agents allow cade, traditional telecom operations— business to a standstill. The current customer expectations and their need passing back-office functionalities and
support functions. Estimates indicate generative AI to work alongside you. the back-office processes that keep
in some cases, returns on incremental For example, a telecom provider look- business flowing—are often the last to
margins will increase 3 to 4 percent ing to improve the process of man- get attention. Many BSS systems have
within two years, and as much as 8 to aging wholesale orders could create long been characterized by outdated
10 percent within five. an AI agent that knows everything tools and systems that are labor-in-
about your company’s product catalog tensive, prone to errors, and slow to
But for CSPs playing catch-up—those and inventory so it can draft detailed respond to customer needs. These typ-
still relying upon outdated business responses to customer questions or ically include tasks such as inventory,
management tools, data silos, and automatically create and fulfill orders order management, customer support,
manual processes—the leap to gen- around the clock, in any language. billing, and service delivery.
AI will be slower. Legacy systems
are often incompatible with modern Generative AI is only capable of Regardless of how fast a network is,
AI technologies, making integration evolving in this way if it’s underpinned these operational systems can create
complex and costly. Compatibility by various business systems that are painful bottlenecks. And let’s be hon-
issues can slow down the deployment API-enabled, automated, and designed est. If it takes six months to provision
of AI solutions and limit their effective- to maintain a strong level of data an enterprise customer because the
ness. Additionally, training AI models governance. Chris Saunderson, senior process for ordering off-net services

“
But for CSPs
playing catch-
up—those still relying
upon outdated business
customer interactions. The time for action is now—telecom future operational success and meeting
management tools, leaders must seize the opportunity to the demands of a dynamic industry.
data silos, and manual By adopting generative AI and au- transform their operations and secure

processes—the leap to tomating key processes, telecom their position in an increasingly com- How TransUnion helps
companies can enhance efficiency, petitive market. Embracing generative
gen-AI will be slower improve customer experiences, and AI, paired with robust automation TransUnion’s portfolio of carrier
drive significant cost savings. systems, will be essential for achieving provisioning solutions and services

16 17
The Fast Mode Trends and Outlook 2025

helps process over 90 million inter- effectively in the years to come. connectivity services. And for those
carrier orders per year, and we have who prefer to outsource this piece of
over 30 years of experience under our TransUnion Universal Order Connect their business, our Managed Services
belts. That means we have the tools (UOC) and UOC Marketplace are offering could be the answer you’ve
and expertise to help improve and API-enabled, cloud-based solutions been looking for.
modernize your order management used by some of the world’s largest
function, including your wholesale CSPs to streamline and automate To learn more, click here.
business —so you can compete more the buying and selling of wholesale

AI STRATEGIES
1. 2025: A Bright New Dawn for The Telco Industry������������������������������������������������������������������������������������������.20

2. 2025: The Year Operators Shift to Fiber Expansion, AI Strategies and

‘As-a-Service’�������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������.23
BY JOHN DENEMARK
3. Telecom Is Now a Tech Product��������������������������������������������������������������������������������������������������������������������������������������.26
John leads TransUnion’s Carrier Provisioning busi-
ness, which includes a complete suite of automated 4. The Rise of the Intent-Net in 2025��������������������������������������������������������������������������������������������������������������������������������.28
products and services addressing carriers’ off-net
provisioning needs. John has extensive experience
5. The GenAI Revolution: How AI Will Shape Telecom in 2025���������������������������������������������������������������.31
in product management, operations, development,
service delivery, customer care, and professional
6. In 2025, CSPs Must Look Within to Get Ahead������������������������������������������������������������������������������������������������.34
services. He holds a Bachelor of Science in Industrial
and Systems Engineering from Georgia Institute of
7. From Digital Telco to AI-Native, What Role Will AI Play in 2025?���������������������������������������������������.37
Technology, and a master’s in economics from The
Johns Hopkins University.
8. Elevating the Telco Monetization Playbook in 2025����������������������������������������������������������������������������������� .40

9. The Future of Telecoms: Charting a Course to Renewed Success�����������������������������������������������.42

10. What’s Next for APIs, Enterprise IoT, and Travel eSIMs?������������������������������������������������������������������������.45

11. Five Ways GenAI Will Transform Telecommunications as We Know It�������������������������������������.48

18 19
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

2025: A BRIGHT NEW DAWN portunity to pivot to new services and

offer better customer experiences. To-
divestments, as well as the own strate-
gic posture to consolidate rather to be
allow them to bring new products and
services to market faster and focus

FOR THE TELCO INDUSTRY

gether with gen AI, we are also seeing consolidated on certain layers. Other investments to enable the connected
that more reliable services are increas- industries like the utility industry have services that are most important to
ing the levels of customer satisfaction. been going through that cycle already. customers.
They have successfully “unbundled”
Actions for success in 2025 their value chain, separating gener- In the B2B market specifically, a
Written by Boris Maurer ation, transmission, distribution, and “connectivity as a platform” model
Senior Managing Director & Clearly, we are on the right track to retail into distinct business units. This around secure and sovereign con-
Communications and Media reinvention. But leaders must commit approach has driven efficiencies and nected compute and cloud services,
Industry Lead, EMEA to some clear actions – and they must allowed them to adjust their models to telcos can create an ecosystem of
Accenture make progress in 2025. the economic nature of each segment. high-value transaction-based services.
The details of the implementation will This approach not only addresses the
#1: DELAYERING THE OPERATING MODEL vary from operator to operator, while for broken value equation but also aligns
almost all of them adopting the delay- with the economic nature of the B2B
Many operators will need to consider a ering paradigm will be key to unleash segment, where businesses are more
new strategy, enabled by a simplified, the trapped value in the current way of willing to pay for services that directly
delayered operating model. This model operation. enhance their operations and bottom
involves restructuring the company into line. The implications of that strategy
distinct layers, such as service, net- #2: PRIORITIZE CUSTOMER-CENTRICITY AND will be huge – especially when it comes
work, infrastructure, and support. Each ENGAGEMENT to the investment strategy for opera-
of these layers have fundamentally tors across their traditional network
different business models, investment Customer satisfaction levels are im- and data center capacity close to their
logic, talent needs and performance proving as networks become more re- customers.
metrics. While in the past, heavy liable, but it won’t create loyalty, or im-
regulation and constant change in the portantly, entice customers to pay more #4: CONNECTIVITY IN THE NEW TECH
underlying technology forced operators for services. Leading players need to LANDSCAPE
to build vertically integrated siloes to free their consumer business and look
rush to market and offer a service in for the right balance between a volume Connectivity must be rethought in the
the next technology generation every play at fixed rates and a premium play. context of the digital core, modern
few years. This is fundamentally chang- At the core will be the establishment networks, and the rise of generative
ing in the future. There is no significant of a new loyalty equation with their AI and data analytics. Telcos have the
growth in the traditional core connec- customers. The three key ingredients potential to excel, but this potential has
The telecommunications industry has cord 56% of tech debt on average, with policy is showing more consideration tivity business of a telco. So, it is all are trust, experience and engagement. mostly been untapped. For example,
weathered the perfect storm over the some of their legacy technology dating for the role telcos need to play to make about reducing complexity, which will Telco operators typically start well on the “Open Gateway” initiative can help
last decade: relentless investment, back two decades or more. This makes Europe more competitive. The second establish clarity of purpose for all parts the trust element, but lack on experi- as a starting point, but will not provide
stagnant revenues and increasing it hugely challenging for them to deliver is CAPEX. Many countries have made of the business to radically improve ence and have no “badge value” with the solution. It is not about “monetiz-
levels of mounting debt. The B2C reliable connectivity, let alone good the necessary investments to modern- performance. their customers. Becoming more rele- ing network assets” but to leverage
segment has been hit the hardest as customer service or new products or ize their networks and have reached vant in the digital daily routines of their the traditional telco strength to cater
consumers - despite asking for more experiences. peak CAPEX for their telco industries. The delayered operating model is customers is a key pre-requisite. with the exploding demand for more
bandwidth-heavy services - have been As such, levels will now be lower than different from the structural separa- localized, more secure and sovereign
unwilling, and potentially more recent- Clearer skies ahead at any point in the past. And network tion of the past. It offers a spectrum of #3: ACCELERATE SECURE AND SOVEREIGN workloads via connected cloud that are
ly, unable to pay for them, given the OPEX should be lower too, with only a options, including individual business CONNECTED CLOUD SERVICES AS THE driven by geopolitical drivers and the
cost-of-living crisis. And while B2B has I believe we are now in a place where fraction of the legacy complexity that units, subsidiaries, joint ventures (JVs), FUTURE OF CONNECTIVITY explosion of gen AI driven applications.
potential, it has yet to materialize. that perfect storm I mentioned is pass- piled up with previous generations of and complete carve-outs. The deci- By using their strengths in connectivity,
ing (we’re not through it just yet) and networks. sion on which option to pursue will If telcos are going to make progress they have the best starting point in the
Meanwhile, previous waves of tech 2025 can offer a much-needed signifi- depend on various factors, such as the moving away from their legacy systems ecosystem to become the preferred
have been introduced, which has cant turning point for the industry. Why Finally, new technologies, such as gen level of ownership and management once and for all, they must have a foun- platform for these services.
resulted in the telcos drowning in tech do I believe that? Because there are AI, will enable telcos to reduce tech control the parent company desires, dational core that comprises of digital
debt. In Europe for example, telcos re- three factors at play. The first is that EU debt, improve agility and create the op- or the need to generate cash through platforms, cloud, data and AI. This will Final thoughts

20 21
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

As we embark on a new year, the growth. 2025 will be critical as we see it’ll be an interesting year and after a
industry is at a pivotal point. Behind which telco leaders make the most tough few years, I’m sure it will breathe
them, the telcos have gone through a progress with simplification, accelera- well-deserved new life into the industry.
dark, tough time, but in front of them, tion of their digital transformation, and
there is potentially a bright horizon up most importantly going back to basics This time next year, the winners will
ahead. The combination of deregula- and putting customer needs at the have simplified and started to radically
tion, technological advancements, and heart of everything they do. Of course, delayer their current core business and
strategic capital investment changes is we won’t see total reinvention in the started to shift investments towards the
paving the way for a return to profitable next 12 months. But whatever happens, new growth paradigm.

BY BORIS MAURER

Boris Maurer, Senior Managing Director and Commu-

nications and Media industry lead for Accenture in
EMEA, oversees strategy, offerings, and business for
the region. With nearly three decades of experience,
he has guided telecommunications, energy, and
high-tech clients in developing transformational
strategies, delivering value at scale, and enabling
next-generation services.

2025: THE YEAR OPERATORS SHIFT TO FIBER

EXPANSION, AI STRATEGIES AND ‘AS-A-SERVICE’
Written by Anurag Sinha
Senior Vice President and Head of Communications, Media & Entertainment and Technology, Cognizant

Pressured by changing customer technology to unlock business value for the coming year:
expectations, the telecom industry is through new products and customers
experiencing a major shift: It’s be- and reimagining their core business #1: AI/ML INTEGRATIONS BECOME BUSINESS
coming increasingly intertwined with operations. CRITICAL
technology. The greatest evidence is
the growing number of AI use cases Against this backdrop, we expect 2025 The wide-ranging discussions of gener-
as telecom service providers move be- will be an eventful year for telecom ser- ative AI and automation at industry
yond driving productivity to using the vice providers. Here are five predictions events signal that AI integration isn’t

22 23
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

just a nice-to-have anymore. It’s a emerging as a particularly promising • Developers can create innovative outages. Knowledge bases built with forms will continue to play a big role and telecom operators need to balance
must for telecom companies, and the area. Q3 earnings from industry applications and services such as gen AI-led technologies like Copilot in impacting customer perception and managing current infrastructure in-
results of a TM Forum survey of AI leaders like Verizon and AT&T indicate fraud management or dynamic can help support teams solve problems remain vital for retaining customers. vestments while pivoting toward more
decision makers at CSPs underscore strong fiber adoption rates. In the US, video quality adjustments by faster. By providing intuitive interfaces flexible, AI-enabled service models that
companies’ progress: CSPs now have a multi-billion-dollar federal initiative accessing advanced network and seamless interactions, digital plat- 2025 is going to be an exciting year can drive new revenue streams.
an average of three to six generative AI is giving a boost to expansion of fiber features.
use cases in production. networks. Operators are looking for • Businesses can accelerate their
first-mover advantage to gain market digital transformation by accessing
There’s an increasing number of leadership and competitive advantage. network capabilities that let them BY ANURAG SINHA
real-world telecom examples that high- Fiber networks enable telecom opera- deliver better, faster and innovative
Anurag is a senior vice president and head of Com-
light the tangible business outcomes, tors to offer advanced services such as services.
munications, Media & Entertainment and Technology
like the improved IT productivity expe- programmable networks as a platform. • Telecom companies can monetize
at Cognizant. His area of expertise lies in the inter-
rienced by a US-based pay TV provider their network capabilities by
section of Communication Industry and technology,
after implementing Microsoft Copilot. #3: PLATFORM-BASED SERVICES TAKE ROOT offering them as services through
and he leads the building Next Gen Telco ecosystem
APIs.
with select hyperscalers, Communication platform
Equally important for the industry are There’s a clear shift from selling siloed • Standardized APIs across telecom
providers, innovative AI startups, and the business
the use cases set to unlock business network services to offering integrated, operators allow developers to
technology services of Cognizant to create innovative
value. platform-based solutions. The transition create applications that work
business models for its communication clients.
is especially crucial in the B2B space, seamlessly on any network.
For example, agentic AI for customer where telecom operators can provide
service is ready 24/7 to understand cloud-based offerings that combine #5: SERVICE PROVIDERS CONTINUE TO FOCUS
telecom customer preferences, predict network services, data management, ON RAISING NPS (NET PROMOTER SCORE)
behaviors–and proactively offer tailored security, and industry-specific solu-
services. We’re also seeing gains in tions. The key advantage is moving to The customer care function is under-
hyper-personalized marketing and “as-a-service” offerings that provide going rapid technological change, yet
sales, where AI-led analysis of custom- flexibility for customers and recurring its omni-channel presence continues
er data—such as interaction history and revenue for providers. to fall short of the success metrics
preferences—can help prioritize leads that service providers had hoped for.
and lead to improved conversion rates. Given the challenging backdrop of Automation initiatives across chat and
low services revenue growth, telecom IVR remain fragmented, resulting in a
AI is particularly valuable for improving companies are adapting their business customer experience that feels disjoint-
operational efficiency and enabling models to offer more value for enter- ed. The result is stubbornly high Aver-
the self-healing networks that address prise clients. They’re expanding their age Handling Times (AHTs) to resolve
service providers’ ongoing OPEX roster of network services with enabled customer queries—and low NPS for
pressure. By analyzing network perfor- services such as data management; CSPs compared to their peers in other
mance data, identifying problem spots, identity and access management; and industries.
and automating appropriate solutions, security services and data pipeline
AI algorithms are poised to optimize services. They’re also offering vertical For 2025, telecom service providers
network performance and core busi- solutions on top of their networks for will be looking for opportunities to
ness operations. industries such as healthcare and fleet raise NPS. Because technology plays
management. a crucial role in the trio of factors that
#2: THE 5G ROI CHALLENGE AND STRATEGIC lead to high scores – service reliabili-
PIVOT TO FIBER ROLLOUT #4: USE OF OPEN API ENHANCES NETWORK ty, high-quality support, and CX – we
CAPABILITIES see CSPs investing heavily in all three
The industry’s massive investments in areas.
5G infrastructure haven’t yet delivered Aided by open APIs, digitized networks
the expected returns. As a result, pro- are revolutionizing the telecom industry For example, AI/ML systems can
viders are actively pursuing new reve- and providing endless opportunities ensure uninterrupted service by taking
nue streams, with rapid fiber expansion such as: proactive steps to prevent network

24 25
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

TELECOM IS NOW A TECH PRODUCT and Kayak shifted their focus from
costly marketing strategies towards en-
hancing user experience (UX) in order
are exposed to SIM swaps and other
telecom scams. In most cases, SIM
cards are still shipped to employees
In the past, integrating mobile with
an HR platform would require a huge
investment and many developer hours.
Written by Rafa Plantier to drive direct and repeat bookings. by mail, leaving organizations vulner- But today, it’s possible to embed
Head of Go-to-Market, Gigs able to identity fraud. Then there’s the multi-carrier phone plans seamlessly
By offering an easy way to stay con- question of travel: without access to into a platform in a matter of weeks at a
nected while abroad, these platforms roaming data, employees are forced to comparatively miniscule upfront cost.
can address a major pain point felt by use insecure WiFi connections, putting
2024 was the year tech brands speed. Here are my predictions for the have amassed millions of users and millions of travellers, vastly improving companies at risk of malware attacks For HR tech platforms looking to differ-
dialed into telecom. From neobanks year ahead. now see revenue growth and profitabil- UX and building platform loyalty at and data theft. entiate themselves and enhance prod-
like Nubank to investment apps like ity as their number one priority. minimal cost. uct stickiness in a crowded market, this
Wealthsimple, the world’s most innova- #1: THE FINTECH X TELECOM LOVE STORY In an era where every step of the em- is an easy win, which is why 2025 will
tive brands began embedding connec- WILL HEAT UP Converting existing customers to For consumers, getting an eSIM from ployee lifecycle is secure and digitized, see the first programmatic phone plans
tivity in their products and tapping into value-added services means driving the same trusted app used to book it’s astonishing that these practices for modern teams.
the $1T+ mobile market. There’s a natural affinity between profit without spending a dollar on flights, reserve hotels, and manage cur- persist.
banking and mobile. Both are essential marketing. And the fact that mobile, like rency is an obvious choice, giving any Telecom for the platform age
This movement was sparked by a per- consumer services, highly regulated banking, is a life-essential makes it a platform with a captive audience an It’s a problem that HR platforms—such
fect alignment of events. and dominated by a handful of players perfect candidate, especially now that edge over standalone eSIM providers. as Deel, Remote and Rippling—are in Telecom is undergoing a transforma-
until advances in technology paved the eSIMs can be integrated into a platform prime position to solve. HR platforms tion similar to the one experienced by
The mobile market has long been a way for innovation. in a matter of weeks at minimal cost. As more tech companies embed travel have already streamlined the majority the finance industry nearly fifteen years
fertile ground for neo-players but high There’s a reason Revolut Founder, Nik eSIMs within their platforms, roaming of workplace processes, from payroll to ago.
barriers to entry have kept newcomers In 2024, these two leviathan industries Storonsky, recently named eSIM as the will change from an expensive ordeal expense management. Telecom is an
at bay—until now. collided, with several fintech heavy- product he is most excited about. into a value-added service we can all obvious next step. Global networks, once closed systems,
weights launching their own eSIMs. access with a tap of our favorite app. are now programmable, interoperable
When the world’s first end-to-end tele- First-movers in this space launched Imagine: A simple interface where and automated. This is telecom for the
com OS transformed mobile connec- As Simon Taylor, the writer behind travel eSIMs in 2024. In 2025, we’ll see #3: WORKPLACE PHONE PLANS WILL IT teams can manage phone plans platform age.
tivity into something embeddable and Fintech Brainfood, argues, cross-selling more fintechs launching local mobile BECOME SECURE AND PROGRAMMATIC across carriers and countries. Instead
programmable, it opened the doors for is a smart strategy for neobanks that services. of juggling multiple networks, they rely As more tech companies integrate
new entrants to innovate. Then eSIMs For most companies, managing on a single provider with one invoice. connectivity with their core services,
took off, and the public perception of There are many creative ways that employee phone plans is a costly, Employees get their eSIMs via Slack mobile is getting smarter and more
phone plans began to change. Connec- fintechs can approach this, from pair- inefficient process that raises serious and connect instantly. If a SIM is com- customer-centric. The days of generic,
tivity went from being a mere com- ing kid-friendly phone plans with their security and compliance risks. promised, it deactivates automatically. one-size-fits-all phone contracts are
modity to a digital product. under-18’s bank accounts to bundling No security risks, no rigid contracts, over. Telecom is now a tech product,
connectivity with complimentary Because IT teams are unable to man- and no arduous support calls. and the possibilities for innovation are
You can draw an interesting parallel products like credit builder cards and age SIMs programmatically, businesses immense.#1:
between the rise of eSIMs and em- smartphone financing.
bedded payments. When embedded
BY RAFA PLANTIER
payments emerged in the 2010s, they With so many options on the table, it’ll
unleashed a wave of innovation, paving be interesting to see where the conver- Rafa Plantier is the Head of Go-to-Market at Gigs
the way for everything from Uber rides gence of these two industries takes us and previously held leadership positions at Stripe,
to Buy Now Pay Later. Today, eSIMs are next. Nubank, and Tink. He has 20 years of experience in
driving a similar shift. Tech companies consumer tech spanning finance and telecom—from
are embedding seamless, user-centric #2: ROAMING WILL BECOME AN EMBEDDED startup and Series A to scale-up and publicly-listed
mobile services in the apps we use SERVICE companies.
daily, and we’re only just scratching the
surface of what this technology can do. In today’s experience-driven economy,
“customer love” matters.
If embedded connectivity took off
in 2024; in 2025, it’s due to hit warp In 2024, platforms like Booking.com

26 27
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

activities suitable for kids.” That’s gives telcos the opportunity for of subscriber growth, they will
the Intent - the interaction with the hyper-personalisation. LLMs can continue to see this huge growth in
mobile phone. The rest gets taken scan a lot of data to find unique the use of data.
by the AI agents and apps, on the patterns about the user. Telcos • The time spent on smartphones,
Network. A more complex use have been behind the curve in primarily watching video, will
case could be prioritising network terms of adopting recommendation continue to increase. It is already
resources for a drone data transfer engines as a concept – most mobile more than five hours a day, with
on a construction site. This brings apps are pretty static, whereas data suggesting this is set to
in network slicing, opening up APIs, home screens on Amazon, Spotify, increase by another 16 minutes.
integrating large language models YouTube etc are personalised to Additionally, there are many
(LLMs) and so on. reflect individual user preferences. developing markets where people
Telcos need to see their networks still do not have a mobile phone.
To deliver on these types of use cases, as an AI entity. They have all the These new subscribers will be
we believe a lot of the technology ele- network elements, the customer registering for the first time on a
ments, especially in some of the more data, the churn data – they need mobile network in 2025, so this data
sophisticated telcos, are already there. to learn to train an AI model growth will continue to play out
So, the question is, will the telcos actu- on top of all that information to over the next few years.
ally enable and execute on AI, so they understand how the whole network
become Intent-Nets – in which case is performing and what can be #4: NEW BUSINESS MODELS WILL BE DRIVEN
they have a great future. If not, they will recommended for better customer BY THE ENTERPRISE SECTOR
remain a utility. experience. A lot of AI Agents to
Agents traffic will start flowing The growth in the enterprise segment
#2: 2025 WILL BE THE YEAR WHEN AI TRULY in the network and Telcos need continues to be a big driver for telcos
TAKES CENTRE STAGE infrastructure including security to globally - there is a wealth of potential
manage this. to be exploited, and this will deliver
• The rapid rise of AI-driven • Privacy concerns will remain much value for telcos in 2025.

THE RISE OF THE INTENT-NET IN 2025

mobile telco services will become an issue. The resistance to the • 5G is a critical factor. As 5G SA
widespread as AI agents – LLMs use of LLMs in the cloud such as networks become more widespread
– become search engines and ChatGPT because of privacy issues these will have a real impact on
personal assistants. Equipped with will decrease, as the emergence the enterprise sector in particular.
an AI chip, your mobile device of open-source AI such as Llama Concrete use cases are emerging
Written by Harsha Angeri will build knowledge around your means that LLMs can be hosted at such as T-Mobile’s network slicing
VP of Corporate Strategy & Head, AI Business, Subex preferences and requirements, the edge of the network. However, for first responders – delivering
remember interactions and the debate on the dangers of priority service, lower latency and
accomplish specific tasks. LLMs uncontrolled AI has also become faster speeds.
will become much more agile and louder. This privacy aspect has not • More and more IoT use cases
In this article, Harsha Angeri of Subex becomes the way users will interact what the Intent of the user is and therefore much easier to run on a been addressed as much given the will emerge with the arrival of 5G
explores the rise of the ‘Intent-Net’, and with this whole ecosystem – both the right instructions to follow mobile device. The penetration has importance of the topic. RedCap – with reduced capability,
how this paradigm shift will enable on the customer experience side to accomplish the task. This will started. AI chips were in around there is far more flexibility in the
networks to understand and proac- and on the network side. Telcos will potentially have a huge impact 4-5% of smartphones in 2023, now #3: DATA AND DEVICE USAGE WILL CONTINUE network to deliver less complex but
tively fulfill user intent, fundamentally understand the Intent of a customer on the telco space. Software was it is around 16%, and it predicted to TO RISE still differentiated services.
changing both customer experiences through hyper-personalisation traditionally driven by humans be around 50% by 2028. • 5G Fixed Wireless Access (FWA)
and telco operations. using GenAI recommendations giving instructions... AI agents will • Telcos will be gearing up to cope • Video, video, video - this content has delivered much success and
which will flow into the network to understand the Intent, create its with AI-generated traffic. Telcos will continue to grow exponentially will continue to do so – this is a
#1: THE INTENT-DRIVEN NETWORK be orchestrated to meet individual own instructions and deliver the will have to change the way they – data is all about short videos, use case that has emerged across
• If we take 2025 as the horizon, customer expectations. results much faster. approach network design, both in TikTok, Instagram, YouTube, and geographies.
mobile will truly become the • AI agents will ‘see’ the Intent and • Telcos will need to really think about terms of how the mobile interaction is where the bulk of the traffic will • The focus on fibre will continue
window to artificial intelligence (AI) work out how to facilitate the becoming more like an Intent-Net. happens and how the customer continue to grow. While many and we will see much M & A
experiences in an AI-driven future, request. The user is not telling the As an example, a consumer use sees the result. A lot of knowledge European and American markets activity around this. Users want a
enabled by the Intent-Net. There is AI agent how to do this - this will case would be something like: “We now resides in the mobile, which are pretty saturated in terms seamless experience, and telcos
an Intent, and a Network needs to be the critical advancement. AI want to go to Hawaii next month -
facilitate that Intent. The Intent-Net agents will autonomously compute find us flights, accommodation and
28 29
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

realise that the bulk of internet

traffic is through these WiFi
gagements with some of the low orbit
satellite companies. Satellite mobile
and the fraud issue is a growing
problem. As a fraud management THE GENAI REVOLUTION: HOW AI WILL SHAPE
TELECOM IN 2025
broadband networks, particularly telephony will happen. company, we see a lot of interest
the heavy workload video and deployment of anti-fraud
applications. Telcos are missing #6: THE FIGHT AGAINST FRAUD WILL measures in this space, and with
out on capitalising on that traffic by CONTINUE: AI MACHINES WILL PLAY AN enterprise rollout, strength in cyber
being simply a wireless provider. INCREASINGLY VITAL ROLE IN COMBATING IT security will become even more
• Fintech will emerge as a critical. Written by Brendan O’Reilly
business in its own right. This • Device fraud. As smartphone • Account takeover and various Chief Executive Officer, UK & Ireland, Boldyn Networks
is a trend that has been ongoing penetration increases and identity frauds, including
- we saw it play out in 2024. becomes an expensive asset, synthetic ID use techniques such
Fintech was a small team within consumers running away with as SIM swapping and cloning or Generative AI (GenAI) is redefining tions for how I believe GenAI will shape GenAI will empower the industry to
the telco, but now increasingly devices has emerged as a major exploiting devices for illegal gain. the telecom and enterprise network- the telecoms industry in 2025. create highly personalised customer
they are becoming a full business use case across geographies and This has emerged as a major use ing landscape. In its early days, GenAI experiences by analysing massive
of their own in emerging markets, will continue to do so. This is the case across geographies and deployments have focused on driving #1: MOVING FROM COST REDUCTION TO datasets and predicting customer pref-
with telcos increasingly procuring number one use case that we are will continue to do so. GenAI will efficiencies such as operating cost REVENUE GENERATION erences. Using these insights, telecom
independent fintech applications. dealing with today, whether in the accelerate account takeovers. reductions. As we look ahead, the operators can see substantial bene-
Global players like Orange, which UK, the US, Africa or in the Middle • AI agents will make inroads industry is beginning to leverage GenAI Early implementations of AI in telecoms fits from developing dynamic pricing
is European based, has expanded East. This as a big issue that will against fraud. AI agents are LLMs for newer, richer opportunities includ- focused on streamlining operations and models, tailored marketing campaigns
their fintech presence in developing only get bigger unless there is a equipped with tools, knowledge ing revenue generation, customer ex- reducing expenses. In 2025, we’re set and value-added services. While the
markets, reflecting a growing trend broad management solution. and memory that can accomplish perience enhancements, and services to see operators and vendors exploring specifics of certain revenue streams
of diversification within telecom, • GenAI fraud will become more specific tasks. This will impact right innovation. how they can monetise AI, and specif- may remain unclear, I do believe that
especially in developing regions. prevalent and higher profile in the from mobile phone automation to ically GenAI, through new and innova- the industry is set to unlock significant
news - deepfake audio and video is telco BSS/ OSS automation. We As the industry embraces this transfor- tive revenue generation techniques. growth opportunities.
#5: SATELLITES WILL ATTRACT INCREASING increasingly getting democratised are seeing opportunities for fraud mative technology, here are six predic-
INTEREST FOR MOBILE CONNECTIVITY and there is a lot of open source investigation using AI agents, with
available so we will see more great interest in a TMF Catalyst
Telcos were fighting against this but account takeover fraud using this. with some top telcos. This will
will increasingly adopt it, despite • Subscription fraud will reshape how we think of software
questions around licensing. In India increase, as money flows through and architectures.
for example, both Airtel and Jio are the subscriptions. More and more
two big players with partnership en- payments are going through apps

BY HARSHA ANGERI

Harsha Angeri is VP of Corporate Strategy & Head, AI Business at Subex. Harsha is responsible for steering the
strategic direction of the company. His responsibilities encompass shaping corporate initiatives, developing and
executing growth strategies, portfolio transformation and fostering innovation within our technology landscape.
He runs the AI business of Subex driving the strategy & roadmaps across Conventional and Generative AI applica-
tions. Harsha comes aboard with a wealth of entrepreneurial leadership and strategic insight.

30 31
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

This ability to rapidly process and act opportunity. By leveraging improved #5: A NEW ERA FOR BSS AND OSS likely to have fully integrated GenAI transformative power of GenAI is clear. Together, we are unlocking the full
on large volumes of customer data will first-contact resolution systems, GenAI into business and network strategies, potential of GenAI to create an inter-
prove to be a key competitive advan- can significantly enhance the customer Business support systems (BSS) and allowing them to stand out in an in- As the telecom industry continues its connected future. The question is no
tage. support experience. AI-powered chat- operational support systems (OSS) will creasingly competitive market. Wheth- AI adoption journey, forward-thinking longer how AI will transform the tele-
bots and virtual assistants will provide undergo a transformation as GenAI be- er it’s enhancing network performance, operators that prioritise collaboration, com industry, but how quickly we can
#2: IMPROVEMENTS IN NETWORK standard, context-aware responses, comes a key part of their functionality. improving customer experience, or ethical practices and continuous inno- harness its power to shape the future
PERFORMANCE AND RELIABILITY resulting in reduced wait times and unlocking new revenue streams, the vation will lead the charge. we’re all building.
improved customer satisfaction. Operators in the Western world, who
GenAI also has the potential to make a have been using the beginnings of AI
profound impact as the cornerstone for This evolution in customer support will technologies and tools for years, such
enhancing network management and not only decrease operational costs as Machine Learning, are particularly
optimisation. With intelligent auto- but also build trust and loyalty among ripe for GenAI adoption. With their net- BY BRENDAN O’REILLY
mation, network operators can make customers. As telcos master the use works already upgraded to accommo- Brendan O’Reilly is Chief Executive Officer for the UK
real-time adjustments to network pa- of AI-driven solutions, they’ll create a date advanced AI capabilities, they will & Ireland at Boldyn Networks. Through his extensive
rameters, resource allocation and traffic seamless, efficient support ecosystem lead the way in integrating GenAI into industry background, strong customer relationships,
routing. This will help to unlock new that sets them apart from competitors. their BSS/OSS stacks. This integration and operational expertise Brendan and his team are
levels of network performance, reduce will streamline operations, optimise leading the company’s business into the next phase
latency and improve reliability. #4: REDEFINING BUSINESS PROCESSES resource allocation, allow predictive of growth through innovative connectivity solutions.
ACROSS INDUSTRIES maintenance and improving billing
The ongoing rollout of 5G standalone processes, ensuring a higher quality of
(5GSA) networks also provides new Telco players are well-positioned to service for customers. It can also help
grounds for GenAI to demonstrate expand GenAI applications into other to simplify the customer acquisition
its potential. Operators that integrate industries. For example, AI-driven process.
GenAI into these advanced network solutions can simplify and streamline
infrastructures will benefit from dynam- processes such as request-for-proposal #6: INCREASED FOCUS ON DATA
ic scalability, enabling them to offer (RFP) management and procurement. GOVERNANCE AND ETHICAL AI ADOPTION
tailored services to enterprise custom- By automating repetitive tasks and
ers. Solutions such as private networks providing predictive insights, GenAI will The success of GenAI in telecom will
coupled with the capabilities of mobile help enterprises save time and re- hinge on robust data governance and
edge computing will help enterprise sources, especially in large purchasing a commitment to ethical AI practices.
customers discover new use cases and operations. As operators process vast amounts of
unlock new business opportunities. customer data, ensuring security, trans-
This trend will enable telcos to leverage parency and trust will be crucial.
For example, private networks can Generative AI to help enterprise cus-
enable secure, high-speed communi- tomers identify the types of solutions Telcos must create strong data man-
cation for remote industrial operations, they need, simplifying the process agement frameworks and collaborate
while mobile edge computing allows of discovering and building the right with technology partners to uphold
for real-time data processing in sectors systems. By using AI-driven insights, ethical AI standards. This includes ad-
like manufacturing and healthcare. telcos can guide businesses through dressing concerns around data privacy,
These innovations ultimately provide the complexities of solution selection algorithmic bias and accountability.
a route to growth and transforma- and customisation. Through collabo-
tion, offering businesses the ability to rations with technology partners and A well-governed AI strategy will not
enhance operational efficiency, improve cross-industry alliances, telcos can only ensure compliance but also build
customer experiences, and create new speed up the innovation and adoption customer confidence in AI-powered
revenue streams. of these tailored solutions, empowering services.
enterprises to transform their opera-
#3: ENHANCED CUSTOMER SUPPORT tions with greater efficiency and agility. #7: LEADING THE CHARGE INTO A GENAI-
POWERED TELECOM FUTURE
AI’s ability to reduce customer sup-
port costs is another transformative By the end of 2025, telco players are

32 33
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

#1: TECH-STACK SPECIALIZATION TAKE #2: SERVICE SPECIALIZATION ACCELERATE enabling internal cost savings, promot-
CENTER STAGE ing agility and accelerating time-to-
Software is not the only area in which market.
Today’s enterprise customers expect specialization will take the place of
CSPs to offer cross-industry, multi- genericism. CSPs will also continue Furthermore, these investments will
site and multi-service partnerships to grapple with an existential ques- support a continuous DevOps envi-
that span similarly complex footprints. tion throughout 2025: Do we want to ronment that enables CSPs to launch
With so many variables in play, generic be a connectivity provider (NetCo), a updates, enhancements and bug fixes
backend solutions no longer cut it for service provider (ServCo) or something in days, rather than months. Perhaps
CSPs. Over the next year, we’ll see in between? more importantly, it will allow CSPs
CSPs wise up to the inadequacies of to make these adjustments without
their existing tech stacks and invest in There’s no single “right answer.” Some downtime — a must when the service
specialized, purpose-built solutions. will choose to lean into the network you provide is as mission critical as
side and cultivate the most agile, flex- connectivity. In short, the pace at which
Telecom-specific configure, price, ible and reliable connectivity options CSPs make the leap to the cloud will
quote (CPQ) solutions will be the they can. Others will see consum- separate this year’s winners from the
starting point for many CSPs, as they er-facing operations with innovative pack.
form the foundation of sales organi- bundles as their forte.
zations. Generic CPQ solutions tools #4: AI WILL SEE PROGRESS, BUT NOT TAKE
connect front- and back-office opera- The truth is, CSPs probably won’t land OVER
tions to enhance accuracy and speed neatly on one side or the other. Even
in the quote-to-cash process. They where the CSPs choose to pursue the CSPs have embraced AI through-
are essential for modern sales, helping ServCo business model, they will need out their operations, using predic-
to automate contract proposals and to determine which verticals to run tive analytics and personalization to
terms, reduce seller workloads, improve after based on their core competencies enhance pricing, deal configuration,

IN 2025, CSPS MUST LOOK

cross- and upselling and increase cus- and differentiated capabilities. They service suites and customer experi-
tomer satisfaction. However, most basic simply cannot be all things to all cus- ences. However, few CSPs have seen
CPQ solutions are unable to accom- tomers, even with specialized support- the technology realize its full value,

WITHIN TO GET AHEAD

modate the complexity and nuance of ive technologies. and they’re unlikely to see it in 2025
enterprise CSP sales—and the same either — unless they commit to major
is true for other tools in the CSP tech What is crucial right now is gaining IT transformations.
stack. the advantage of time. The faster CSPs
make their choice, the more time they AI needs context to model scenar-
Written by Greg Tilton Today’s CSPs need solutions that sup- will have to cultivate their expertise, ios—and CSPs do have the data to
Vice President, Catalog, Quote & Order, CSG port the agility, dexterity and integra- build tailored tech stacks, develop provide it. However, decades of siloed
tion that modern enterprises demand partnerships and strengthen customer operations have prevented the kind of
of their partners. Investing in special- relationships to serve the new model — end-to-end insight that is needed to
Compared to organizations in other The decision paid off, helping them that operational complexity.. ized CPQs and other tools will help and less time will be spent spreading use AI tools to their fullest potential.
industries, communications service remain competitive in a fast-changing improve backend management and operations too thin. While other industries face similar
providers (CSPs) have been slower to market and keep pace with customers’ As such, this will be a pivotal year for interactions with customers, helping difficulties, the complexity and breadth
update and retool their internal tech changing connectivity needs. But it has CSPs. They’ll be focused on bringing to reduce seller workloads, improve #3: CLOUD INVESTMENTS FLOURISH of CSPs’ services make this issue even
stacks. Instead of improving internal also left these these organizations in their technological capabilities up to cross- and upselling and increase cus- more pressing. Network performance
processes and technologies, CSPs somewhat of a bind as they enter the speed with those of their partners and tomer satisfaction — a critical function To date, CSPs have shied away from and customer satisfaction have obvious
chose to focus their efforts over the new year. CSPs’ networks are more refining their strategies for long-term as customer retention is a strategic cloud investments, opting to push correlations, but traditional structures
past few years on enhanced network- complex than ever before, and their success. As they do, I predict we will priority in today’s highly competitive updates and changes on a month-by- often keep that information separated,
ing infrastructure. internal processes and tech stacks are see: market. month cadence. That’s not going to cut with the impact of outages not easily
at risk of buckling under the weight of it anymore. Amid rising competition seen by account managers until a cus-
and heightened consumer expecta- tomer reports it.
tions, cloud investments will be crucial,

34 35
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

To leverage AI effectively, CSPs will first insight, they will be able to unlock the need to enable more agile innovation,
need to invest in widespread record full power of AI in the years to come. drive revenue growth and strengthen
and storage integration. Breaking down customer loyalty.
those walls will take time, and that’s As the industry shapes its future, CSPs
where CSPs should focus in 2025. that adapt swiftly and act strategically
Once that foundation is laid, they will fi- will lead the way into the industry’s
nally be able identify the patterns, con- next era. Embracing specialization and
nections and interdependencies lost integration will be key to this journey,
to siloed operations. Armed with this providing CSPs with the insights they

BY GREG TILTON

Greg Tilton is Vice President, Catalog, Quote & Order

at CSG, responsible for product line delivery, support
and scalability of CSG’s CPQ solution. Greg was the
founder and CEO of DGIT Systems, a company that
provided IT solutions to fixed and mobile carriers
worldwide. Under his leadership, DGIT became a
prominent player in the telecommunications industry,
won multiple awards and contributed to global
industry associations such as TM Forum, MEF Forum
& FTTH Council.

FROM DIGITAL TELCO TO AI-NATIVE, WHAT ROLE

WILL AI PLAY IN 2025?
Written by Guy Lupo
EVP of AI & Data Innovation, TM Forum

#1: THE INDUSTRY WON’T FIND THE ‘KILLER’ that’s not a bad thing. zero-touch network management. Even
AI BUSINESS CASE THIS YEAR AI agent marketplaces are essentially
Many AI applications being marketed rebranded app stores - not revolution-
The search for AI’s “killer use case” has as “game-changers” aren’t new at all. ary breakthroughs.
dominated industry conversations, with They’re long-standing ideas now made
operators, vendors, and analysts all possible thanks to a more powerful That doesn’t mean AI isn’t delivering
looking for the breakthrough that will computing that is now more accessible value. The real opportunity lies in au-
redefine telecoms. But the reality is, than ever. Think weather prediction, tomating many of the processes we’re
that moment isn’t coming in 2025 - and automated customer support, and already doing. Operators that focus

36 37
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

on integrating AI into core operations, would provide a competitive advan- 5G capacity by 2027. The networks we #5: CYBER RESILIENCE IS ONE OF AI’S uptime as a way to recoup some of the idating AI leadership roles to ensure
rather than chasing the next big thing tage. rely on today weren’t designed for this BIGGEST CHALLENGES – BUT IT’S ALSO A network investment and better serve strategy and execution align. CSPs
are already seeing benefits. Telcos shift, and without action, operators risk REVENUE OPPORTUNITY their customers. need to stop treating AI as a side proj-
like AT&T and Verizon are using AI to Leaders in the space have claimed that failing to serve customers’ needs. ect and instead integrate it into their
optimize network performance and AI has significantly boosted efficiency, AI isn’t just transforming telecoms; it’s #6: TOO MANY AI CHIEFS, NOT ENOUGH core digital transformation strategies.
customer experience. cutting down employee workload by CSPs need to move now to rethink also making networks more vulnerable. STRATEGY Without a unified vision, AI investments
nearly 17 million minutes each year and how they allocate network capacity. will remain fragmented and underuti-
Telstra is making networks smarter generating savings worth hundreds of Should upload speeds match down- As CSPs optimize their networks to The past five years have seen the lized.
and more efficient. Instead of relying millions of dollars. loads? Should symmetrical uplink traf- enable AI applications, the larger the creation of roles such as CIOs, CISOs,
on rigid, pre-programmed automation, fic become a premium service? Those attack surface becomes. AI-driven au- CDOs, which have positively impacted Looking ahead
they’ve built a system called “Knowl- Instead of simply replacing jobs, AI is who adjust first will not only avoid tomation means more interconnected organizations throughout the digital
edge Plane” that learns and adapts reshaping them. Take customer contact future bottlenecks but gain a competi- systems, increasing the risk of cyber- transformation era. However, as AI 2025 won’t be the year of revolution-
to network demands in real time. This centers, for example; where AI-pow- tive edge in an AI-driven world. attacks, single points of failure, and proliferates, many are choosing to hire ary AI breakthroughs - but that’s not a
means faster decision-making and ered agents can handle routine queries, AI-generated fraud. new headcounts. we’re seeing blurred bad thing. Instead, this will be a year of
fewer manual processes. The impact allowing human agents to oversee and #4: AI REGULATION IS COMING – OPERATORS lines of responsibility that hinder, not practical AI adoption, network adap-
has been huge - processing times have manage multiple AI assistants instead MUST EXERCISE CAUTION But this isn’t just a challenge - it’s an help, overall AI success. tation, regulatory clarity, and smarter
improved by 30%, and setting up new of answering one call at a time. A single opportunity. As businesses embrace cybersecurity investments. The indus-
services now takes just three months employee could soon be responsible As AI regulation tightens worldwide, AI for their own operations, they can’t Often, individual business units run try is moving past hype and towards
instead of up to 18. for 500 AI agents, scaling customer questions of accountability are surfac- afford downtime or disruption, and op- isolated AI experimentation projects real-world AI impact.
service like never before. ing. Who is responsible if AI erators can monetize this resilience. with their preferred partners and then
Proving that small, incremental gains hand them over to CIOs for production, Operators that embrace AI strategically,
can be just as impactful as bold ideas. This shift isn’t about eliminating work- is used for harmful or illegal activities, Expect to see CSPs rolling out pre- expecting seamless integration. This rethink network priorities, and monetize
ers - it’s about expanding their reach or models are trained with inappropri- mium security and resiliency service siloed approach can hinder innovation new opportunities will emerge stronger.
#2: AI WILL CREATE JOBS, BUT NOT A and impact. ate, biased or inaccurate data? guarantees for business customers, and impede AI adoption at scale. But those still waiting for the “killer” AI
LABOUR SHAKE-UP - YET offering AI-driven threat detection, data business case may find themselves left
#3: CSPS TO RETHINK NETWORKS FOR AI- One debate echoes the fair-share con- protection, and guaranteed network In 2025, companies will start consol- behind.
There’s been plenty of fearmongering ERA DATA FLOW troversy: should telcos be responsible
around AI eliminating jobs, but 2025 for monitoring content flowing through
won’t bring mass layoffs or radical For decades, networks were designed their networks? If so, who pays for the
BY GUY LUPO
workforce changes. for one thing: delivering content to us- additional monitoring and interception?
ers, fast. As a result, download speeds Guy is a visionary and strategic thought leader with
Historically, new technologies don’t just have always been prioritised to serve Some argue that enterprises supplying over three decades of experience spanning various
replace workers, they create new roles. end-users with the content they want. AI tools should be accountable for how industries and leadership roles. He is a Distinguished
Cloud computing didn’t kill IT jobs; it But AI is flipping that equation - and they’re used - not the networks they Fellow of the TM Forum, the world’s largest telecom
shifted demand to cloud architects, operators can’t afford to wait to adapt. run on. But the reality is more complex, forum, and the pioneer behind the adoption of
DevOps engineers, and cybersecurity and governments may push responsi- Network as a Service (NaaS) architecture. This
specialists. AI will follow the same path: People are no longer just downloading bility onto CSPs. innovative approach is now widely accepted and
first, a wave of jobs in AI migration, content, they’re uploading massive files utilised as a network transformation method towards
training, and supervision, then a gradu- (videos, documents, datasets) and re- Regulators will increase audits, network software and network APIs. Currently, Guy
al move toward automation. ceiving small AI-generated responses. and operators must tread carefully. is pursuing a PhD at Swinburne University, focusing
Whether it’s AI-powered transcription Over-regulation could mean addition- on developing a metric for the Trustworthiness of AI
Yes, some roles will disappear, but the or cloud-based editing, or training AI al compliance costs without a clear Systems. His research is driven by the anticipation
industry won’t see a dramatic work- models, the demand for uplink capacity revenue upside. The industry will need of future regulatory needs for scalable continuous
force shake-up this year. Studies show is increasing. to engage in discussions about where control monitoring (CCM). In his current role as the AI
that nearly 90% of telecom companies responsibility truly lies, before regula- & Data Innovation industry mission lead at TM Forum,
use AI, with 48% in the piloting phase A McKinsey survey found that 72% tors make that decision for them. Guy will focus on enabling the global telco industry
and 41% actively deploying AI. Most of organizations are already using to accelerate the innovation and adoption of AI at
telecom service providers (53%) agree AI, and AI-powered applications are scale through sustainable, safe, and responsible use
or strongly agree that adopting AI expected to push uplink traffic beyond of AI & data technologies.

38 39
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

ELEVATING advanced network slicing capabilities,

telcos can better segment complex
processes, and manage partner eco-
systems with ease. Forward-thinking
resolutions over traditional phone
support.

THE TELCO
services and optimize revenues. operators I work with are harnessing
digital enablement layers that integrate Beyond support, AI can optimize every
In high-churn markets, AI helps telcos seamlessly with legacy systems, en- stage of the subscriber journey. For

MONETIZATION
deliver hyper-personalized experiences abling them to introduce new services example, when a customer browses
that are critical to retention. Automa- without disrupting their core infrastruc- an operator’s platform, AI-supported
tion-powered customer management ture, accelerating time-to-market and tools can recommend personalized

PLAYBOOK IN
and monetization platforms also reducing complexity. Nobody wants service bundles based on their inter-
streamline partner integrations and to rip out and replace anchored BSS ests, simplifying decision-making and
improve operational efficiency, allowing infrastructure to launch a new video increasing conversion rates. Powerful

2025
operators to quickly test, launch, and subscription offering or modernize subscriber platforms with globally-scal-
deliver compelling new products that their customer support and billing able, data-driven toolkits can help tel-
keep subscribers engaged. system. A new wave of cloud-based cos cater to diverse audiences through
overlays gives product and engineering extensive global payment support and
#3: UNBUNDLING THE CHALLENGES OF A teams the flexibility to enable business features like real-time translation and
Written by Paolo Cuttorelli BUNDLED APPROACH expansion without a complete system multi-language support, ensuring ser-
Senior Vice President of Global Sales, upgrade. vices are accessible to wider subscrib-
Evergent Digital service bundling will remain er bases.
one of the most impactful strategies #4: ALL-IN ON THE DIGITAL-FIRST
for telcos looking to differentiate their SUBSCRIBER EXPERIENCE Finding the winning play
Telecommunications businesses are osition, extending beyond connectivity explosion of billing systems and pricing proposition in a competitive market.
undergoing unprecedented transfor- into digital consumer services like models — all while navigating evolving However, rigid, traditional systems Providing exceptional customer The telco industry is at a critical inflex-
mation as they evolve their positioning smart homes, gaming, and e-health. global compliance requirements. Chal- often hinder the rapid experimenta- experiences is a business imperative ion point. Elevating the monetization
to offer more services and greater According to Omdia’s Telco Tech lenge accepted. tion telcos need to stay competitive. for telcos that want to become the playbook is essential for telcos aiming
value to customers than ever before. Trends to Watch 2025 report, growth in Modern monetization strategies must ultimate digital storefront and seam- to lead in a fast-evolving subscription
Communication service providers these areas is projected to drive signifi- #2: WHERE DOES AI FIT IN THE NEW focus on agility, enabling operators to less subscription marketplace. Today’s economy. As operators embrace their
(CSPs) are no longer just providers of cant revenue. The report estimates that MONETIZATION PLAYBOOK? seamlessly combine any hybrid mix of subscribers expect intuitive, digital-first evolving roles as digital lifestyle pro-
connectivity – they’re rapidly morphing the global B2B opportunity alone will subscriptions, one-time purchases, and interactions at every touchpoint – from viders, the ability to deliver seamless,
into dynamic technology companies, reach $2.7 trillion by 2025, driven by Monetization in the subscription econ- usage-based pricing models. onboarding and billing to customer scalable, and profitable services will
or ‘techcos’, delivering a wide array of cloud, cybersecurity, and AI solutions. omy is becoming increasingly com- support. AI-powered chatbots and determine their success. By leveraging
digital lifestyle services. Offering a broader range of digital plex as telcos navigate multi-partner By adopting modular, cloud-native virtual assistants streamline customer advanced technologies to optimize op-
services allows telcos to meet diverse ecosystems and cross-industry part- monetization platforms, telcos can care by addressing routine inquiries erations, enhance personalization, and
This necessary shift is reshaping how subscriber needs. Operators are now nerships. AI is emerging as a critical modernize their approach without and freeing up human agents to focus unlock new revenue streams, operators
telcos engage with subscribers and prioritizing multi-service ecosystems enabler in optimizing pricing strategies, the need for major overhauls. These on complex issues. This approach is can position themselves as indispens-
integrate with new service partners. To that cater to individual lifestyles. This driving personalization, and enhancing platforms enable operators to launch particularly appealing to younger gen- able players in the digital subscription
meet the expectations of today’s dig- might include combining entertainment retention. AI-powered churn preven- new services quickly, simplify billing erations, who prefer quick, app-based ecosystem.
ital-native consumer while identifying and productivity services or integrating tion solutions can analyze subscriber
new revenue sweet spots that hit the e-health solutions with family-focused behavior to identify at-risk customers
BY PAOLO CUTTORELLI
mark, telcos must elevate their mone- offerings. and implement targeted retention
tization playbook. New tools, creative strategies. Whether it’s offering loyalty Paolo brings over 20 years of expertise and knowl-
power plays and more agile technol- Despite incredible opportunity, the discounts or introducing wide-ranging edge from the media, sports and telecommunications
ogies are the name of the game–let’s transformation from telco to techco pricing tiers, AI-supported tools allow industries, helping global operators like AT&T, Singtel,
explore how operators can play best in introduces new complexities which telcos to be more proactive in their Astro Malaysia, Warner Bros. Discovery, Sony, Verizon,
2025. major players are still grappling with. engagement, reducing churn rates and Telecom Italia grow revenues and transform their
Becoming the ultimate consumer and driving customer lifetime value. subscription businesses.
#1: FINDING A NEW ROLE AS THE DIGITAL subscription option means delivering Predictive analytics capabilities are
MARKETPLACE PROVIDER digital-native, seamless customer expe- also able to help telcos refine their
riences. It means the ability to manage pricing models. By combining automat-
Telcos are redefining their value prop- intricate partner ecosystems and an ed and unified billing technologies with

40 41
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

THE FUTURE OF TELECOMS: CHARTING porates agentic AI will empower telcos

with:
management. The MVNO business
is also an excellent opportunity for
network monetization and partner
ecosystem expansion to unlock growth

A COURSE TO RENEWED SUCCESS

1. Intelligent process automation MNOs, especially those with underuti- opportunities. API monetization and
2. Hyper-personalized customer lized 5G network capacity. B2B2X partnerships are helping to
experiences power scalable success. McKinsey esti-
3. Faster resolution times The speed and connectivity of 5G mates the network API market could
4. Capability to create competitive, make MNOs and MVNEs well-posi- represent $100-$300 billion for telecom
Written by Michele Campriani customer-centric products tioned to differentiate their offerings operators over the next five to seven
Chief Revenue Officer, Optiva 5. Faster time to market by supporting and enabling enhanced years.
6. Sales and operational efficiency services. These include ultra-low
7. Dynamic and continuous innovation latency and high-bandwidth applica- The real value of network APIs is in de-
tion opportunities, such as the Internet veloping services and applications that
Agentic AI will become essential in of Things (IoT), smart cities, auton- go beyond bandwidth and speed. With
telecom BSS transformations, help- omous vehicles, remote education network APIs, telcos can programmat-
ing to simplify IT infrastructures and and telehealth. Further, the increase ically interact with their network and
modernize legacy BSS systems. of eSIM-based devices and the use data systems. This capability is offered
Agentic AI-powered BSS and virtual AI of MVNOs as marketing channels in as a service to external developers. By
agents of the future will work togeth- retail, banking, media, sports and other making the vast amounts of data that
er seamlessly, creating an intelligent industries highlight new businesses telcos collect accessible through APIs,
ecosystem that provides telcos with and innovations that are being built on including voice, SMS, user behavior,
exponential value, essential capabilities the 5G ecosystem. data usage and location data, they can
to meet market demands, Opex reduc- create platforms for third-party devel-
tion across lifecycle management and MVNO growth has also increased de- opers to build applications that utilize
groundbreaking levels of innovation. mand for MVNEs to leverage real-time these insights. It enhances the utility of
data analytics, AI and GenAI to auto- their networks and extends their reach.
#2: WHOLESALE REIMAGINED: MVNES AND mate processes and improve customer
MNOS experience. As GenAI enables MVNEs These data insights unlock monetiza-
to offer more flexible and custom- tion opportunities beyond traditional
The global mobile virtual network ized services to MVNOs, MVNEs will service options that align with the
operator (MVNO) market has more see additional growth from offerings, real-world needs and expectations
than doubled in the last decade and including increased AI-driven customer of consumers. Developing network
was valued at $83.1 billion in 2023. It is support, network automation, hy- APIs and making them accessible to
expected to grow at a CAGR of 7.4% to per-personalized service offerings and developers and enterprises will create
The telecoms industry enters a pivotal most significant opportunities for 2025 of GenAI, AI agents have “agency,” reach $169.7 billion by 2033. MVNOs niche MVNO market opportunities. a new generation of 5G-driven appli-
period in 2025. It must boldly embrace will arise from agentic AI, MVNEs and which means the ability to make inten- have the advantage of entering the cations. Additionally, through initiatives
innovation and actively pursue new network APIs. tional, independent decisions and take market and launching services quickly A final word on MNOs. As the MVNO such as CAMARA, the global telco API
business opportunities to achieve action with autonomy. They can proac- by partnering with mobile network market has expanded, MNOs are eye- alliance, telcos are exploring how to
growth and secure its future. #1: REAPING THE REWARDS OF AGENTIC AI tively achieve goals, complete complex operators (MNOs) and mobile virtual ing the role of MVNEs. This emerg- redefine and approach network APIs
tasks, adapt and make decisions in network enablers (MVNEs), allowing ing trend involves MNOs integrating as significant business assets that can
According to McKinsey & Company, Goldman Sachs and McKinsey project real-time based on changing envi- them to skip investing in extensive MVNEs into their ecosystems to provide unique competitive advantages
telecoms have invested $1 trillion in that new technologies, such as agentic ronments. This behavior differs from infrastructure development. capture a larger share of the MVNO and foster innovation.
network upgrades since the advent of AI, are maturing rapidly. This growth today’s chatbots, which are limited to market. MNOs such as BT Group,
5G. As the urgency for 5G monetiza- offers the opportunity for significant using queries and prompts to interact. As the MVNO market grows, MVNEs Vodafone, Vodacom, and T-Mobile lead Part of redefining the approach to
tion intensifies, they must capitalize productivity gains and the addition of will continue to play a vital role in this trend. network APIs includes telcos redefining
on emerging technologies and growth trillions in economic value to the global Further, agentic AI learns dynamical- expanding the mobile ecosystem and themselves and how they do business
opportunities. The industry is ripe economy. ly from experiences. This enables AI enabling MVNO innovation by provid- #3: A GOLDEN GATEWAY FOR GROWTH: API to best capitalize on API monetization.
with tech concepts that are becom- agents to constantly improve, evolve ing wholesale network access, core MONETIZATION Telcos must consider themselves not
ing a reality and have the potential to Agentic AI builds upon GenAI technol- their capabilities and offer continuous functionalities, regulatory compliance network service providers alone but
redefine and revolutionize telecom ogy, including large language models innovation. A telecom business support oversight, and technical complexity There is increased demand for 5G embrace the mindset of becoming
markets globally. We predict that the (LLMs). Considered the next evolution system (BSS), for example, that incor-

42 43
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

software innovators. The return on

investment for making this cultural
BSS will enable telcos to adopt diverse
monetization models, such as charging
on user data and network conditions.
WHAT’S NEXT FOR APIS, ENTERPRISE
IOT, AND TRAVEL ESIMS?
shift, putting the right technology and based on API call volume, data usage, Seize 2025, start today
ecosystem in place and leveraging or even the number of users accessing
real-time data to drastically enhance the service. This will allow telcos to These opportunities, agentic AI,
service levels and customer experienc- monetize their API offerings effectively. MVNEs and network APIs, are poised
es is high. And opportunities to capital- to reshape the telecom sector to drive Written by Gabriel Salvate
ize on API exposure extend into various Further, when developing a new API efficiency, enhance customer experi- Head of Customer Solutions & Presales, BICS
industries from healthcare to banking ecosystem for 5G-driven services, ence and further foster groundbreaking
and travel to entertainment. using APIs in conjunction with GenAI innovative new business models. In
presents an even more golden oppor- 2025, telcos will embrace these ad-
In developing API monetization op- tunity for telcos. Incorporating these vancements to secure their positions in It’s a true statement every year - the needs of modern businesses and con- When working with businesses, a key
portunities, using BSS as a backbone technologies affords them the ability to the increasingly competitive, dynamic world is becoming more digitized. This sumers continue to widen. In the world reason for the slowdown of telco reve-
to manage and support transactional create more intelligent, context-aware and evolving market. affects how people live their lives, and of technology, being labelled tradi- nue compared to hyperscalers or SaaS
aspects of API services is vital. A robust and personalized applications based how businesses interact and engage tional often signals falling behind, and models is the complexity of accessing
with consumers. Telecoms, once the stagnating revenues in the Telecoms and implementing telco services. While
backbone of enterprise communica- industry would agree. But it’s not all Telcos have unmatched infrastructure
tions, is continuously being forced to doom and gloom -Telecoms is evolv- (the others wouldn’t exist without it),
adapt. ing and thriving. Let’s explore three their services often lack the agility
BY MICHELE CAMPRIANI key trends set to drive global telecom required in a software-driven, digital
Michele Campriani is the Chief Revenue Officer With the surge of instant messaging growth in 2025. world, making integration of telecom
at Optiva, a leader in AI and cloud-native telecom apps and software-based services (to services cumbersome.
revenue management software. With over 30 years of name two), the gap between traditional #1: BRIDGING THE GAP BETWEEN
experience, he has held executive and C-suite roles telecom providers and the evolving ENTERPRISE AND TELECOMS This is where telecom APIs come in.
at Mobileum, Comptel, Empirix, Accanto Systems
and Hewlett Packard. Michele earned a master of
electronics engineering degree at Politecnico di
Milano, Italy.

44 45
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

These software interfaces allow devel- space has struggled to reach its poten- years, with 50% of consumers now revenue streams and operational effi- chain space, in particular, eSIM tech- a commercial race to create offerings
opers to seamlessly integrate telecom tial – the eSIM. While eSIM technology familiar with the technology. ciencies. nology will help continue to expand and capture market share in this brand-
services like identity management or has existed since 2014, enterprise IoT global IoT. While on the consumer side, new space.
verification with just a few clicks. APIs adoption has lagged behind the con- New use consumer use cases for The year 2025 will also be the year of travel eSIMs will make international
are also fuelling the impressive growth sumer market due to the complexity of eSIMs will drive much of this growth. the eSIM. In the industrial and supply travel far simpler, while also prompting
of CPaaS (Communications Platform as provisioning and managing eSIMs. Chief amongst these is the ‘travel eSIM’.
a Service), which is also driven by APIs. These allow users with eSIM-enabled
But change is on the horizon. In 2025, smartphones to purchase connectivity
BY GABRIEL SALVATE
The potential here is huge. According the introduction of the SGP.32 standard in virtually any country, often at a lower
to McKinsey, the network API market from the GSMA will standardize eSIM cost than traditional roaming. The Gabriel is Head of Customer Solutions & Presales at
could unlock an estimated $100 billion deployments. The consumer market Global Travel eSIM Market is expect- BICS, a leading global wholesale carrier in the Tele-
to $300 billion in revenue for Telco has benefited from its own version ed to surge from $2 billion in 2024 to com industry. With more than 15 years of experience
operators over the next five to seven of the standard (SGP.22) for years – approximately $260 billion by 2034. in the Hardware, Software, and Telecommunications
years. With APIs, businesses gain driving widespread adoption in devices Industries, Gabriel has held different leadership
frictionless access to telecom services like smartphones.SGP.32 will finally This trend is not just significant for con- positions across business strategy and development
such as identity verification, location bring much needed standardization sumers. On a commercial level, what (e.g., Business Development Director, Enterprise Sales
tracking, and on-demand quality cov- to enterprise eSIM solutions. This new makes travel eSIM offerings so unique Manager, and now Head of Customer Solutions and
erage. standard will establish clear definitions is how easy they are to launch. Unlike presales) that have allowed him to understand the
of three key roles in the ecosystem: the traditional telecom services, consumer various aspects of the business and its challenges.
Despite this promise, some scepticism network operator, the reseller, and the brands can launch travel eSIM offer- Gabriel’s previous experience has allowed him to
remains. Critics worry that telecom eSIM orchestrator, setting the stage for ings without prior telecom experience develop a rich knowledge in Industrial IoT, Cloud
APIs could follow a path similar to 5G: streamlined deployments and greater or infrastructure. Computing, Connectivity(MVNO/E) and Cellular
overhyped and underdelivered. Howev- scalability across IoT applications. Technologies (5G, LTE, Private 4G/5G and LPWA).
er, all signs point to a meaningful shift This ‘democratization’ of connectivity
in 2025, with standardization efforts like While companies have unofficially is fuelling intense competition, with
the GSMA’s Open Gateway making API occupied this role, standardizing this major consumer brands, mobile virtual
integration more accessible. Addition- will support the growth of the IoT network operators (MVNOs), and travel
ally, the rise of ‘API integrators’, com- ecosystem and give businesses more eSIM enablers vying for market domi-
panies that serve as marketplaces for confidence to deploy eSIMs. However, nance. While 2024 saw the first wave of
telecom APIs, will provide businesses educating the market will be crucial. travel eSIM launches, 2025 will be the
with simplified access point for global The eSIM is ready to drive global IoT year in which key players begin to pull
connectivity services while helping expansion, the question is now: are ahead.
operators engage directly with enter- businesses ready to embrace the
prises. eSIM? Looking ahead

#2: UNLOCKING ENTERPRISE IOT #3: THE TRAVEL ESIM BOOM In 2025, the line between personal and
enterprise communications will blur
Away from business-to-customer com- That’s not the only eSIM story you can even further. Consumers will increas-
munications, industries are increas- expect to see this year. Aside from ingly demand seamless, engaging
ingly investing in the IoT to connect connecting machines and systems, the interactions with businesses as they
machines to other machines. This has eSIM also makes it far easier for people do in personal interactions. To meet
been steadily increasing but is now to stay connected. eSIM availability these expectations and stay ahead on
ramping up. According to Statista, the in consumer smartphones is steadily the curve, companies will increasingly
number of IoT devices worldwide is increasing. leverage telecom APIs to enhance con-
projected to nearly double from 15.9 bil- nectivity and service integration. This
lion in 2023 to over 32.1 billion by 2030. The GSMA predicts that by 2028, half shift not only enhances customer expe-
of all smartphone connections will riences but could also help address the
However, one technology that prom- use eSIM technology. Awareness has telecom industry’s ongoing profitability
ised to make a big impact in the IoT already doubled over the past three challenges by opening doors to new

46 47
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

#4: GENAI WILL ENSURE FAULT-FREE AND

FIVE WAYS GENAI WILL TRANSFORM

Communications Service Providers switches and routers, reduce downtime
(CSPs), GenAI can unleash the real with automated predictive mainte- RELIABLE NETWORKS
potential of 5G, from simply delivering nance, and accelerate how network

TELECOMMUNICATIONS AS WE KNOW IT
connectivity to fostering intelligent dig- engineers respond to network out- When combined with advanced ana-
ital ecosystems to support the expan- ages. With GenAI, network engineers lytics, GenAI can ensure fault-free and
sion of IoT, smart cities, and enterprise can perform root cause analysis and reliable networks. For example, GenAI
applications with AI-optimized solu- resolve issues more efficiently using can help create synthetic data that em-
tions. For example, according to City AI tools that reference historical data and ulates actual customer usage patterns
Written by Bernardo Galvao Lucas Connect, a global community for cities consult manuals to identify solutions for testing and development. Likewise,
Chief Marketing & Strategy Officer, Mobileum to learn about generative AI, nine out for similar past incidents. Ultimately, when adding unstructured data from
of ten mayors in cities across the world GenAI will help with proactive network customer interactions, such as social
want to engage in GenAI - where they management, identifying and resolving media comments or conversations with
2025 is set to be the year when 5G will betting on the advancement of Gener- One of the key benefits of the advance- see the potential for it to address their technical issues before they impact the customer service representatives, CSPs
achieve maximum momentum. Ac- ative AI (GenAI) in wireless networks. ment of GenAI is that it cannot just traffic and transportation, infrastructure, end customer. On the network provi- can add a new layer of precision to
cording to GSMA Intelligence, 5G has Let’s break down GenAI’s role and how analyze existing data but also create public safety, environment and climate, sioning and management side, GenAI their network investment, planning, and
experienced the fastest mobile genera- CSPs turn 5G’s potential into a reality. new original content such as images, and education challenges. Some even brings zero-touch provisioning and design decisions. In addition, GenAI
tion rollout, with connections expected text, videos, sounds, and other out- go further and want to use GenAI to automation one step closer to reality can support dynamic resource alloca-
to surpass 5.5 billion by 2030. So, what #1: GENAI WILL EVOLVE TELECOM’S FOCUS puts. This marks a rapid change from help them predict where house fires by using agents to monitor and help tion to optimize performance across a
is behind the projected acceleration of FROM CONNECTIVITY TO INTELLIGENT machine learning, which focuses on might occur, how to lower healthcare manage the network. diverse range of use cases. Soon, the
5G adoption from 18% in 2023 to 51% DIGITAL ECOSYSTEMS making predictions or decisions based claims, and even anticipate street days of buffering videos and lag times
by 2029? Increasingly, the market is on historical results or patterns. For violence. To do this, smart cities need #3: GENAI WILL STRENGTHEN NETWORK will be a thing of the past. With GenAI,
more than just connectivity from CSPs; SECURITY AND TRUST networks will be able to identify the
computing power at the edge of the content, predict user demand for it,
network close to where data is gener- AI has been a force for good and evil, proactively pre-cache at the edge, and
ated and consumed will be required. particularly when it comes to telecom improve user-perceived latency. GenAI
The vision of truly smart cities requires security and fraud. Today, deepfake can then enable intelligent systems to
intelligent digital ecosystems that scams commonly target us through adjust the streaming quality based on
are only possible when CSPs expose voice, SMS, and video, with AI increas- the user’s device or network conges-
their network capabilities via APIs to ingly amplifying both the scale and tion to ensure a high-quality, seamless
empower these smart city applications sophistication of these security and user experience. At the enterprise level,
and use cases. fraud attacks. However, with GenAI, GenAI can be used to develop custom-
CSPs can get ahead of bad actors by ized 5G network solutions that best fit
#2: GENAI WILL DRIVE NETWORK simulating potential security threats their needs across IoT, smart factories,
TRANSFORMATION AND FAULT-FREE and attack scenarios, which, in turn, and AR/VR.
NETWORKS trains their security systems and teams.
This not only enables network security #5: GENAI WILL ENABLE DIGITAL
According to a recent report by TM teams to better prepare themselves for TRANSFORMATION ACROSS INDUSTRIES
Forum, 55% of communications service the challenges of an AI-enabled world
providers surveyed said they have but also strengthens their network The promise of GenAI cannot be
made good progress using AI and security and trust mechanisms. For understated. According to McKinsey,
machine learning across their business example, with GenAI, fraud systems will GenAI could add the equivalent of $2.6
from customer experience to network be able to identify synthetic content, trillion to $4.4 trillion annually to indus-
operations. While enhancing customer such as fake images and voice clones, tries across the globe. Banking, high
experience with AI-powered support to help distinguish and block untrust- tech, and life sciences are among the
systems has seen the most traction worthy content. This is especially top industries that will benefit the most
with the rise of AI, CSPs have a variety critical as consumers increasingly fall from GenAI developments across their
of ways where GenAI can benefit them victim to telecom fraud, with regulators sales, marketing, software engineering,
as they transform and manage their placing the responsibility on operators product R&D, and customer operations.
networks. For example, GenAI can help to combat this growing threat. The telecoms industry is positioned to
to slash the time it takes for field teams help accelerate the digital transforma-
to install network elements, such as tion across all those verticals due to

48 49
The Fast Mode Trends and Outlook 2025

its role as the connectivity fabric, for can ensure high-quality connectivity, CSPs must leverage the vast volumes
example, supporting smart cities, hy- the telecom sector must also seeGe- of unstructured data outside their OSS/
brid work environments, and enterprise nAI as a driver of innovation and new BSS systems and integrate it with the
solutions. GSMA predicts a fourfold rise revenue opportunitiesfor the telecom structured data within. By bridging this
in mobile data traffic between now and sector itself. McKinsey sees that the gap, they can unlock the potential to
2030, withmonthly global mobile data telecom sector could experience 2.3- not only optimize their networks but
traffic per connection growing from 12.8 3.7% revenue and productivity gains also drive significant revenue growth.

AUTONOMOUS
GB in 2023 to 47.9 GB in 2030. But in- with GenAI, equivalent to $60-100 bil-
stead of focusing just on how networks lion globally. To seize this opportunity,

NETWORKS
BY BERNARDO GALVAO LUCAS

Bernardo Lucas is Mobileum’s Chief Marketing &

Strategy Officer, where he is responsible for strategy and
positioning, product marketing, corporate marketing and
sales development. Since joining Mobileum, Bernardo
led several functions within Marketing & Strategy, having
also led the company’s Risk Management Business Unit.
Bernardo has a wide experience working with Investment 1. Building Resilience: CSPs’ Path in 2025 to Hybrid Cloud, Autonomous
firms and Private Equity companies in growth programs Networks, and AI-Driven Security���������������������������������������������������������������������������������������������������������������������������������.52
and operational transformations. He holds a Master of
Business Administration (MBA) from INSEAD with Honors.
2. The Year Ahead: How AI, Automation, and Embedded Security Will Shape
Telecom in 2025������������������������������������������������������������������������������������������������������������������������������������������������������������������������������.55

3. Network Transformation in the Age of AI�������������������������������������������������������������������������������������������������������������� .58

4. How Generative AI will Transform Telcos in 2025����������������������������������������������������������������������������������������� .60

5. Charting New Territories in 2025: Navigating the Complexity of Future Networks

with Network-as-a-Service��������������������������������������������������������������������������������������������������������������������������������������������������.63

6. AIOps, APIs, & Automation: Key Trends Driving Telco Infrastructure in 2025�������������������.65

7. Agentic AI Can Change Networks If They Clear a High Hurdle�������������������������������������������������������.67

8. 5 Ways Agentic AI and Intuitive Networks Will Redefine Telco SaaS in 2025��������������������.69

9. Beyond the Hype: How AI Will Shape Telecom Infrastructure in 2025������������������������������������� .72

50 51
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

BUILDING RESILIENCE: CSPS’ PATH IN 2025 TO projected to account for 26% of the
projected $14.8 billion in CSP automa-
en security measures that enhance
their ability to detect and respond to
tions in a standardized way to devel-
opers so they can build and sell new

HYBRID CLOUD, AUTONOMOUS NETWORKS, AND

tion investments, according to Apple- threats swiftly. Meanwhile, quantum consumer, enterprise, and industrial
dore. CSPs that have already embraced computing adds another dimension of applications.
AI-driven service assurance are seeing complexity. The “Harvest Now, Decrypt

AI-DRIVEN SECURITY
real results, with incident reductions Later” (HNDL) threat is real; it involves Identity-based use cases - like device
as high as 75% over the past decade, cybercriminals stealing encrypted data status, positioning, and Quality of Ser-
according to a Tier 1 CSP in EMEA. today with the aim of decrypting it vice (QoS) - are expected to lead the
The takeaway here is that as CSPs when quantum computing technology way, especially in sectors like finance,
increase their automation maturity, is advanced enough. media, and entertainment. The value
Written by Jitin Bhandari they should benchmark their progress here is clear: a thriving API economy
Chief Technology Officer, Cloud and Network Services, Nokia and collaborate with industry leaders To prepare for these challenges, CSPs allows developers to tap into network
to build intent-driven services. Lever- should conduct regular audits, adapt capabilities, enhancing user experienc-
aging an open-source AI stack tailored to evolving regulations, and invest in es for both enterprise and consumer
As they harness the capabilities that Hybrid cloud models, in particu- #1: HOW AUTOMATION AND AI WILL SHAPE to telco needs will also enable CSPs AI-driven security tools. These proac- clients.
autonomous networks and network lar, strike a valuable balance: CSPs AUTONOMOUS NETWORKS to optimize operations and provide tive steps will help CSPs maintain legal
digital twins offer, communication ser- can keep mission-critical processes customers with a superior experience compliance, enhance system resil- CSPs should focus on preparing their
vice providers (CSPs) are refining their on-premises while leveraging the Automation and AI will continue to be based on real-time insights from 5G SA ience, and protect customer data. With networks for secure API exposure.
cloud adoption strategies. In 2025, this scalability of the cloud. This approach foundational as CSPs work toward fully networks. quantum technology on the horizon, Partnering with aggregators and devel-
will mean greater adoption of hybrid bolsters network resilience and sup- autonomous networks. As the indus- CSPs should also start investigating opers on real-world use cases will help
cloud models. ports new opportunities, especially try shifts to intent-based automation, #2: THE SECURITY LANDSCAPE REMAINS post-quantum cryptography to stay demonstrate the potential of network
for 5G standalone (SA) networks and CSPs will be able to tailor services for FULL OF CHALLENGES ahead of potential future threats. APIs and unlock new revenue oppor-
Right now, about $14.7 billion - around future 5G Advanced use cases. To fully specific verticals and deliver targeted tunities. Additionally, creating strong
4% of the industry’s capital expenditure capitalize on hybrid cloud, CSPs will solutions across diverse sectors. AI for Security will remain a top priority as #3: CSPS ARE LEVERAGING THE NETWORK API governance frameworks now will
- is directed toward network function need to invest in orchestration tools, IT operations, or AIOps, will empower CSPs explore generative AI (GenAI) API ECONOMY position CSPs for broader API deploy-
cloudification, according to Omdia. assess existing workloads, and map CSPs to use predictive analytics and and prepare for the emerging impact ments, ensuring they’re ready to scale
Why the focus on the cloud? Moving out a transition plan. These actions will observability, allowing them to speed of quantum computing. GenAI, while CSPs, motivated by massive invest- up and stand out in the market.
workloads to the cloud gives CSPs ensure that cloud adoption is strategic, up incident resolution and improve revolutionary, also introduces security ments in 5G, are increasingly focused
greater flexibility and operational resil- driving network performance while network health. risks, as it can be exploited to conduct on using network APIs to monetize #4: HOW CSPS ARE BUILDING A RESILIENT
ience, allowing them to scale services adapting quickly to customer needs. complex cyber-attacks. CSPs are coun- their network assets and core capabil- AND FUTURE-PROOF INDUSTRY
as needed and optimize costs. Consider this: by 2028, AIOps is tering this by investing in GenAI-driv- ities by exposing their network func-

52 53
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

CSPs that embrace hybrid cloud As CSPs and industry leaders gear up that’s prepared for whatever comes
models, invest in automation and AI, for these shifts, these predictions offer next—poised to deliver high-impact
and strengthen their security will be valuable insights to help navigate the services, improved customer experi-
well-positioned to thrive, enabling col- journey. Each of these transformations ences, and meaningful growth in 2025
laborative growth with partners. contributes to building an industry and beyond.

BY JITIN BHANDARI

Jitin Bhandari is the Chief Technology Officer of Nokia’s Cloud and Network Services (CNS) business group. Jitin
leads software technology strategy and disruptive innovation for the mid & long-term horizon, accelerating the
transformation of networks, operations, and analytics to cloud-native, 5G and beyond. He’s known as a commu-
nications visionary with end-to-end expertise in AI-driven networks, operations, and services. He has a proven
ability to not only build the big picture strategic focus but also to deep dive and execute. Jitin has 20+ years of
experience in the software/IT/telecom industry. A sports fanatic keeping true to his New England base, he loves to
run and experience the outdoors.

THE YEAR AHEAD: HOW AI, AUTOMATION,

AND EMBEDDED SECURITY WILL SHAPE
TELECOM IN 2025
Written by Pascal Menezes
Chief Technology Officer, MEF

As generative and agentic AI revolu- innovation and transformation can be three pivotal trends are poised to drive
tionize industries, telecom networks seen across the telecommunications the industry in 2025 and beyond.
must deliver unparalleled reliability industry. This year promises lead-
and low latency. Meanwhile, NaaS ing-edge advancements that will rede- #1: AI WILL DEMAND HIGHLY RELIABLE AND
continues to redefine network delivery fine connectivity and service delivery, PREDICTABLE NETWORKS
through automation and innovative reshaping how networks operate and
commercial models. Add to this the serve enterprises and consumers alike. The rise of generative AI and agentic
seamless embedding of cybersecurity From the expanding role of artificial AI, which emphasizes reasoning and
into networks, and 2025 promises to be intelligence to the evolution of on-de- autonomy, is set to transform appli-
a year of unparalleled innovation and mand networking and the integration cations across industries. These AI
growth for the telecom sector. of cybersecurity as a core feature, systems rely on a mix of foundational
these trends will significantly impact and specialized models to deliver
As we step into 2025, opportunities for the industry’s trajectory. The following results, necessitating networks that are

54 55
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

both highly reliable and predictable. contracts. This level of automation secure. Certified SSE and ZT solutions These trends highlight the telecom service providers, enterprises, and collaboration, and enabling innovation
Latency and quality of service (QoS) ensures real-time responses and seam- provide a standardized framework for industry’s opportunity to drive transfor- technology innovators—will be es- to help expand the connectivity and
will become critical for AI inference less connectivity, enabling users to implementing these advanced security mative change that benefits enterpris- sential to realizing the full potential of service delivery ecosystem. With a fo-
from the edge, while AI training of large provision and scale network resources models, helping service providers align es, consumers, and society at large. As these advancements. cus on robust infrastructure, seamless
models will require very large band- instantly. with best practices while ensuring networks evolve to meet the demands automation, and embedded security,
width to ensure efficient data exchange robust protection and operational effi- of AI, on-demand services, and inte- The future of telecom lies in its abil- the industry is well-positioned to thrive
during training cycles and advancing The implications are profound. Automa- ciency. Certification instills confidence grated security, the industry will foster ity to anticipate and adapt to rapid in 2025 and beyond, delivering value
AI capabilities. tion will empower service providers to in addressing emerging cyber threats, innovation, improve connectivity, and technological change. MEF continues that extends far beyond traditional
deliver a cloud-like experience for net- positioning service providers as essen- strengthen trust in digital ecosystems. to play a pivotal role in driving industry connectivity.
This shift means telecom operators work services, supporting on-demand tial for secure and reliable connectivity. Collaboration among stakeholders— standards and certifications, fostering
should invest in infrastructure that connectivity, multi-cloud integration,
meets AI’s demands. By enabling and superior application performance. This transformation enables service
efficient bandwidth, data routing and This evolution of NaaS will also unlock providers to offer high security pos-
BY PASCAL MENEZES
minimizing latency, networks will new commercial models, fostering tures while maintaining performance.
support the rapid growth of AI appli- innovation and simplifying network By integrating cybersecurity at the Pascal Menezes is a proven technology thought
cations, fostering innovation across consumption for enterprises worldwide. network level, providers can mitigate leader, sales evangelist, product manager and
sectors. AI’s expansion into areas like Enterprises will benefit from enhanced threats without compromising speed seasoned IP architect with decades of experience
healthcare diagnostics, financial mod- operational efficiency, observability, the or reliability. Enterprises benefit from in internetworking, next-generation information
eling, autonomous systems, and other ability to scale resources dynamically, seamless protection against malware, systems, and communication architectures. Pascal
critical industries, underscores the and simplified interactions with service exploits, and evasions ensuring data currently serves as Chief Technology Officer at MEF.
urgency for telecom providers to stay providers. By embracing NaaS, busi- integrity and user trust. For example, He is focused on SD-WAN, SASE, cloud scale archi-
ahead. Advanced network capabilities nesses can focus on growth and inno- integrated security measures pro- tectures, real-time media networks, Software Defined
will support AI’s technical requirements vation while leaving the complexities vide real-time monitoring and threat Networks (SDN), Network Function Virtualization
and empower industries to unlock new of network management to automated mitigation. This approach not only (NFV) and Lifecycle Service Orchestration (LSO).
efficiencies and transformative oppor- systems. enhances resilience against emerging
tunities. cyber threats but also instills confi-
As NaaS matures, it will drive new dence among enterprises navigating an
Additionally, as AI becomes embedded collaboration opportunities between increasingly complex threat landscape.
in more critical functions—from auton- service providers and enterprises.
omous vehicles to intelligent supply Partnerships will evolve to address in- As industries embrace digital transfor-
chains—the pressure on networks to dustry-specific needs, such as health- mation, the proliferation of connected
deliver uninterrupted and ultra-reliable care connectivity solutions or secure devices—from IoT sensors to industrial
connectivity will increase. The evolu- financial data transport. These tailored automation—amplifies the need for ro-
tion of 5G and the early adoption of offerings will further cement NaaS as bust cybersecurity. Embedded security
6G technologies will play a crucial role a transformative force in telecommuni- ensures that even the most complex
in supporting these advancements, cations. networks can adapt to new threats, of-
providing the quality of service and low fering a proactive defense mechanism.
latency essential for AI’s success. #3: CYBERSECURITY WILL BE EMBEDDED IN This shift reflects a broader industry
NETWORKS trend toward building trust and reliabili-
#2: NAAS WILL FOCUS ON AUTOMATION AND ty into every layer of network design.
COMMERCIAL MODELS Cybersecurity is becoming an intrinsic
part of network architecture. Instead of We can expect a pivotal year ahead for
Network-as-a-Service (NaaS) is rede- treating security as an add-on, service the telecom industry, with AI, auto-
fining how networks are consumed providers are integrating cybersecurity mation, and embedded cybersecurity
and delivered, placing automation at its measures directly into their network reshaping the landscape. Service
core. To function as a true as-a-service offerings. This approach includes providers need to adapt to meet these
offering, networks require end-to-end embedding Secure Service Edge (SSE) challenges, and ensure their networks
automation, including supply chain capabilities and Zero Trust policies, are equipped to handle the demands of
processes, API integration, and smart ensuring that networks are inherently next-gen applications and services.

56 57
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

NETWORK TRANSFORMATION IN THE AGE OF AI accurate data — across every domain

and network layer.
failures for improved performance and
productivity models, delivering better
coverage, capacity and QoE. The
network activity that are truly impact-
ful. Traditional service assurance will
be replaced by continuous network
Written by Joe Krystofik Gathering this data is complicated by resulting proactive approach delivers optimization and adaptation to meet
Head of AI Software Product Planning, multiple data feeds, conflicting network a vast improvement over the tradi- demanding service level agreement
Network Automation, Fujitsu events, duplicate data, legacy man- tional behavior of reacting to network (SLA) metrics.
agement software and ‘black holes’ faults, cyber security concerns and
of data that are overlooked or simply maintenance issues; thus helping to Ultimately, the combination of AI and
unused. CSPs face some monumental prevent costly network disruptions and automation will allow CSPs to shift to a
difficulties, such as establishing data outages. more service-oriented focus, applying
source consistency and determining AI to correlate and analyze multiple
what data is truly valuable, in order to #3: ON THE ROAD TO SELF-DRIVING data sources from across the net-
derive meaning from their data and NETWORKS work to support development of new
realize meaningful business process services and reduce customer churn.
improvements. As CSPs embrace AIOps, traditional In this way, AI becomes a valuable tool
network management will inevitably to advance organization growth and
With the implementation of automated evolve, easing operational burdens so future profitability, enabling a pivotal
analysis, CSPs can sort, organize and that engineers can spend less time change in mindset from reactive to pro-
elevate network data to the visibility on mundane tasks. In the future, we active, and boosting overall customer
necessary for decision-making, and eventually will see the transformation satisfaction.
use GenAI to sift through multiple of networks to become more autono-
knowledge bases, manuals, service mous, enabling simplified infrastructure Taking a leap of faith to AI
reports and past failure logs. This capa- planning, self-optimization of service
bility then enables the use of large lan- layers, proactive issue resolution and Recent AI advancements offer valuable
guage models (LLMs) to conduct root automated management of AI agents. opportunities for profound improve-
Across the shifting landscape of the Although some CSPs have adopted AI Taking this a step further, CSPs will cause analysis and predictive analytics, Although increasing adoption of AI and ments throughout the networking
telecommunications industry, keep- use in network operations to a certain implement AI-powered automation to allowing CSPs to implement advance automation will move us further down industry — spurring widespread oper-
ing pace with network evolution has extent, many are still in early, small- achieve real-time network optimization planning and stay one step ahead of this road in 2025, we are still a long ational changes in both the business
become increasingly challenging for to- scale trial phases. As these organiza- for maximum quality of experience changing conditions. An important way from that reality. model and across the network itself. As
day’s communications service provid- tions become more comfortable with (QoE). With the ability to continually step in this process will be developing CSPs adopt new tools and methodolo-
ers (CSPs). In addition to the ongoing leveraging technologies such as GenAI fine-tune network parameters using a unified AI/ML pipeline for continual In the meantime, CSPs will contin- gies that enable greater use of network
transformation of network infrastruc- and machine learning (ML), they likely GenAI, CSPs will not only save operat- GenAI training and inferencing. ue to integrate AI into the network automation, predictive analytics using
ture from one generation of technology will use AI to handle simple and mun- ing costs such as by lowering energy business model, wielding automation big data, and ML algorithms, they will
to the next, customer services and dane tasks at first, freeing up personnel costs, but also reduce capital expendi- In this way, tomorrow’s network will and data analytics to simplify day-to- be well-positioned to begin taking
expectations continue to evolve as well. for more important activities. tures by not over-provisioning. make intelligent predictions regarding day operations and glean insights into advantage of AIOps in 2025.

As a result, networks operations have Pairing AI tools with automation in #2: PREDICTING THE FUTURE WITH
become too complex for manual an AIOps approach allows CSPs to ANALYTICS
BY JOE KRYSTOFIK
processes alone. The need for rapid, eliminate manual and repetitive tasks,
real-time response to dynamic network optimizing network performance and In the realm of 5G and forthcoming 6G Joe Krystofik is Head of AI Software Product Planning,
conditions necessitates myriad new operations while realizing consider- technologies, network conditions are Network Automation, Fujitsu Network Communica-
tools and approaches that rely heavily able cost savings. For example, AIcan increasingly dynamic and service key tions. He is an expert in workflow automation, AI/
on automation and artificial intelligence be used to draft network specification performance indicators (KPIs) are more ML applications, microservice based solutions and
(AI). Further adoption of AI will solidify designs for engineers to review and strict than ever before. Greater adop- multi-vendor management and orchestration. Prior
its role in telecoms, empowering new, finalize, saving significant time for engi- tion of AI will help address these chal- to Fujitsu, Krystofik worked for Verizon. He received
value-added uses in communications neering staff. Likewise, implementation lenges with advanced tools for network an MSEE from Cal Poly with a focus in Digital Signal
networks during 2025 and beyond. of AI to improve fleet management management and orchestration. Yet, Processing and specialty on Digital Image Process-
efficiencywill help reduce resolution underpinning every AI decision is the ing. He holds a BS in Engineering Science from Trinity
#1: AUGMENTING HUMANS WITH AIOPS time when network issues occur. need for reliable, real-time access to University in San Antonio.

58 59
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

HOW GENERATIVE AI WILL pabilities to produce multiple designs

to assist network experts in making
Trends that redefine communication
experiences
conversations via phone calls, web-
sites, or mobile apps. Enhanced with

TRANSFORM TELCOS IN 2025

necessary trade-offs is a key trend to multi-language support and sophisti-
watch for. We will see the rise of AI Telcos and CSPs are entering a new cated natural language understanding,
Assistants interacting with humans and era where networks will do more than these assistants will provide consistent,
advising network design and config- connect—they will enhance, translate, personalized experiences that dramat-
uration changes to meet service and and optimize communication in real ically improve self-service resolution
Written by Natasha Tamaskar quality of experience requirements. time. In 2025, we predict CSPs will be- rates while reducing operational costs.
Head of Business Strategy and Marketing, Radisys gin embedding GenAI applications into
#2: OPTIMIZING ENERGY EFFICIENCY their networks, unlocking AI Assistance By bridging traditional voice channels
THROUGH PREDICTIVE TRAFFIC ANALYSIS that transforms how businesses col- with digital platforms, CSPs and com-
laborate, and individuals manage their munication Managed Service Providers
The Generative AI megatrend is trans- Trends that impact network demands while staying competitive in a With a wealth of KPIs and other metrics daily communications. With these ca- have a tremendous opportunity to
forming every industry, and telecom infrastructure fast-changing digital landscape. collected within the networks, telcos pabilities built directly into telecom in- provide a new generation of virtual as-
is no exception. While we may be just can use this data to generate various frastructure, companies of all sizes will sistants, allowing businesses to extend
becoming comfortable with GenAI-en- Network infrastructure is the back- #1: EMPOWERING NETWORK DESIGN WITH AI traffic scenarios and traffic distribu- experience seamless interactions that their reach to customers regardless of
abled chatbots, a deeper impact is bone of telecom operations, enabling ASSISTANTS tion patterns among users correlated drive productivity. At the same time, their preferred communication method.
expected over a period of time as seamless connectivity and high-quality with time of day and locations. RAN individuals will gain powerful personal
telcos usher in GenAI to deliver better service delivery. As networks grow Improving network coverage and ca- controllers and SON (Self Organizing assistants that organize conversations, #2: ELIMINATING LANGUAGE BARRIERS
services and network operations. more complex, Generative AI will be a pacity drives data and revenue growth Network) enabled products will lever- provide real-time support, and enhance THROUGH REAL-TIME TRANSLATION
age these capabilities to dynamically personal productivity - all creating new
Forward-looking telcos have been configure and reconfigure network revenue streams for operators. Language differences will no longer
investing in AI across a broad spectrum elements as per traffic demands and hinder global collaboration. CSPs will
of applications, from customer support without impacting customer experience #1: EXPANDING CUSTOMER REACH WITH adopt GenAI technologies that provide
portals and billing systems to network by monitoring necessary metrics. This OMNI-CHANNEL GENERATIVE AI AND AGENTIC real-time translation & transcription in
planning, design, and operations. is expected to bring significant savings AI live phone calls and video meetings,
These investments are driving greater in operation costs to telcos, especially complete with natural voice localiza-
automation, efficiency, and intelligence the energy costs in the radio networks. While enterprises are already adopting tion, lip-syncing, and captions. CSPs
across critical aspects of network man- AI-based virtual assistants to their cus- will help business customers expand
agement and maintenance. #3: AUTOMATING NETWORK OPERATIONS tomer experience centers to automate their reach effortlessly, making inter-
WITH PREDICTIVE MAINTENANCE self-service engagements, Generative national business as seamless as local
As telcos and CSPs deepen their AI and the emerging Agentic AI will conversations—all powered by the
investment in Generative AI, the impact While some telcos have already started transform how customers will inter- network.
on networks and customer experiences using GenAI in their field operations, act with business applications. The
will be nothing short of transformative. this year, we will see a wider adoption interaction with SaaS applications #3: TRANSFORMING PHONE CONVERSATIONS
From smarter, more efficient infrastruc- of GenAI in analyzing trouble tickets will evolve from web-based interac- WITH INTELLIGENT PERSONAL ASSISTANTS
ture to seamless, AI-enhanced interac- and automating and expediting the tions to voice-enabled AI agent-based
tions, these advancements will unlock resolution of issues. Predictive main- interactions. Both Generative AI and While we already see an emerging
new levels of performance, automation, tenance and network optimization Agent AI technologies will play a role adoption of AI personal assistants on
and innovation. become easier with such adoption. in this transformation. While GenAI will VoIP-based collaboration calls (like
With the deep and precise mapping of make it easy for AI assistants to have Teams, Zoom, and Google Meet), the
These trends fall into two major assets across the network and the vast immersive dialog with the customers, GenAI-powered personal assistants
areas — those that impact network amount of information captured, we will the Agentic AI will make the conversa- will also become constant companions
infrastructure and those that redefine game-changer, bringing automation, for telcos as subscribers consume also see telcos increasingly use Ge- tions proactive and drive autonomous in our phone communications, main-
communication experiences. The fol- intelligence, and efficiency to plan- newer services made possible with nAI-based internal tools and workflows operations on live conversations. These taining comprehensive records of all
lowing predictions highlight how GenAI ning, optimization, and maintenance. the advent of AI. The network planning to make informed decisions on network intelligent AI systems will seamlessly our calls. Unlike today’s fragmented
will revolutionize telecom in 2025 and AI-driven tools will streamline deci- and design is usually a complex and upgrades and Capex spends. integrate with business workflow and communication tools, these assistants
beyond, driving unprecedented value sion-making, reduce operational costs, manually intensive process. The use of multiple data sources to handle com- will seamlessly organize your entire
for businesses and consumers alike. and enhance network performance, digital twins to map network assets in plex customer queries through natural conversation history, enabling natural
ensuring telcos can meet evolving great detail and leveraging GenAI ca-

60 61
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

CHARTING NEW TERRITORIES IN 2025:

language queries like “When did I last engines. This integration will enable location, and traditional boundaries be-
speak with Susan, and what delivera- real-time decision-making based on tween providers and consumers fading
bles did I promise?” These assistants voice interactions, allowing sales teams as AI creates new value ecosystems.

NAVIGATING THE COMPLEXITY OF FUTURE

will transcend basic record-keeping by to adjust strategies mid-campaign and
understanding conversation context, product teams to prioritize features The year 2025 will mark a turning
tracking commitments, and proactively based on customer feedback. By point as the costs of integrating GenAI

NETWORKS WITH NETWORK-AS-A-SERVICE

managing follow-ups. This transition transforming every conversation into into telephony significantly decrease,
from passive communication channels structured data, voice becomes not just making these powerful technologies
to intelligent, personalized assistance a communication channel but a critical accessible to all. This cost & complex-
represents a fundamental shift in how source of enterprise intelligence that ity reduction will enable businesses
individuals manage relationships in drives strategic advantage. of any size to adopt and benefit from
both professional and personal con- GenAI capabilities. We will see mass Written by Ali Shaikh
texts. Transforming telecom: building adoption across industries as technolo- Chief Product Officer, Graphiant
intelligent networks and seamless gy becomes both more affordable and
#4: VOICE AI ANALYTICS: TRANSFORMING communication easier to implement through simplified
CONVERSATION DATA INTO BUSINESS integration paths.
INTELLIGENCE The merger of Generative AI with
telecom marks a defining shift in the The strategic path forward for telecom
In 2025, businesses will integrate voice industry’s trajectory. Throughout 2025, leaders is clear - those implementing
AI analytics directly into their opera- networks will evolve toward auton- GenAI will increase efficiency, but
tional workflows, transforming what omy—self-healing and self-optimiz- those reinventing their business mod-
was once ignored conversation data ing—while communication experienc- els around it will unlock entirely new
into actionable business intelligence. es break through current language, value categories. The future of telecom
Platforms will automatically extract, context, and channel barriers. isn’t just about enhanced networks
structure, and feed conversation- —it’s about creating an intelligent foun-
al insights into enterprise datasets. The most significant impact will come dation that amplifies human potential
Competitive intelligence and market through widespread access: small and enables innovations beyond our
trends identified during calls will flow businesses wielding enterprise-grade imagination.
seamlessly into business intelligence capabilities, individuals receiving per-
platforms and predictive analytics sonalized AI assistance regardless of

As we stand on the cusp of 2025, the trends. Each pushes the boundaries of #2: EDGE-ENHANCED AI SOLUTIONS
connectivity landscape is undergoing what networks can do and how com-
monumental changes. Technological plex they become. Edge-Enhanced AI Solutions combine
BY NATASHA TAMASKAR advancements are presenting chal- AI and edge computing to process data
lenges that the average enterprise #1: AUTONOMOUS NETWORKS closer to its source. This integration is
Natasha Tamaskar is Head of Business Strategy and Marketing at Radisys. Natasha heads corporate & product
simply cannot handle alone. As these vital for real-time data analysis appli-
marketing, business strategy, as well as the digital endpoints business for Radisys. She has over 20 years of
next-gen capabilities unfold, it’s be- Autonomous Networks are revolu- cations, such as automated manufac-
global telecom expertise in business strategy, product management and marketing and business development in
coming increasingly clear that sophis- tionizing network management with turing and smart cities. By reducing
cloud platforms, CPaaS, network intelligence, wireless core, and voice and data network security solutions. Prior
ticated Network-as-a-Service (NaaS) their self-optimizing, self-healing, and latency and enhancing processing
to Radisys, as the VP of Cloud Strategy, she was responsible for GENBAND’s Kandy.io PaaS strategy and business
providers are essential to designing, self-securing capabilities. These net- power at the edge, enterprises act on
development. Natasha also spearheaded and launched several of GENBAND’s key strategic solutions including
delivering, optimizing, and managing works adjust without human interven- insights faster than ever before.
Wireless Gateway, Network Security and WebRTC as the VP of Strategic Marketing.
these intricate systems. tion, reducing downtime and improving
efficiency. They represent a significant #3: QUANTUM ENCRYPTION
Embracing complexity: an overview leap forward, enabling networks to an-
of new network trends ticipate and react to issues in real time, Quantum Encryption is poised to
like having an internal team of network redefine secure communications. By
The future of networking is being engineers on standby, 24/7. leveraging the principles of quantum
shaped by several transformative mechanics, it offers unprecedented

62 63
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

levels of data encryption and secure

data transmission. This technology will
and the pace of technological change
shows no signs of slowing. For an en-
enterprises stay ahead of the curve.
AIOPS, APIS, &
AUTOMATION: KEY
disrupt current encryption standards, terprise, trying to keep up can feel like The expertise offered by NaaS vendors
providing security measures virtually running on a treadmill that’s constantly includes cutting-edge security solu-
unbreakable with today’s computing accelerating. tions, advanced analytics, and seam-

TRENDS DRIVING TELCO

capabilities. less integration capabilities. Consider a
The strategic shift to network-as-a- scenario where a company integrates
#4: DECENTRALIZED AUTONOMOUS service quantum networking through a NaaS
CONNECTIVITY

INFRASTRUCTURE IN 2025
provider, gaining access to ultra-secure
Many companies are turning to NaaS communications without developing
Decentralized Autonomous Connectiv- as the strategic solution in response to the technology internally. Or a business
ity is another groundbreaking trend, fo- these pressures. This model alleviates that adopts edge-enhanced AI solu-
cusing on creating secure, peer-to-peer the burden of managing complex net- tions via NaaS to boost its IoT oper-
interactions without a central authority. works by outsourcing these responsi- ations without investing in expensive Written by Morgan Stern
This approach enhances security and bilities to experts. infrastructure upgrades. Vice President of Automation Strategy, Itential
privacy and supports transparency and
user control, aligning with the growing NaaS provides flexible, scalable net- Conclusion
demand for decentralized solutions in a work solutions tailored to an enter-
data-conscious world. prise’s specific needs. By adopting As we navigate towards 2025, it is clear
NaaS, businesses can focus on their that next-gen networking technologies
Why enterprises can’t go it alone core operations, confident that their are not merely evolutionary—they are
network infrastructure is in capable revolutionary, demanding new ways
The evolution and complexity of hands. This model reduces operational of thinking and operating. For enter- Looking ahead to 2025, service provid- 5G, and enterprise services. Pressure tize opex reduction due to a number of
these technologies present significant overhead and ensures access to the prises looking to remain competitive, ers will be pressed to go beyond staff on margins and costs is forcing CSPs factors, most notably a lack of sig-
challenges for enterprises. Developing latest technologies and innovations embracing Network-as-a-Service is reductions and get serious about infra- to innovate to deliver the complex, nificant revenue generated by heavy
and maintaining advanced capabilities without constant internal investments. essential. NaaS allows businesses to structure and operations transformation customizable services that advanced investment in 5G infrastructure. At
internally requires specialized exper- harness the full potential of advanced efforts to achieve their opex reduction enterprise customers demand. The the same time, their largest and most
tise and substantial resources. Many The pivotal role of NaaS vendors networking technologies without being and revenue targets. All infrastructure ability to rapidly turn new infrastruc- advanced customers are demanding
organizations find that the technical overwhelmed by complexity. By part- investments and the adoption of new ture services into revenue-generating more complex, customized services
and economic barriers are too high to NaaS vendors are the linchpins in nering with NaaS providers, enterprises solutions will be evaluated from this products will rely on CSPs’ ability to at an increasing rate to support their
surmount. this new networking era. They bring confidently chart a course through the viewpoint. orchestrate processes across multiple evolving business needs. The way
together cutting-edge technologies digital landscape, ensuring they are networks and clouds and easily com- service providers view opex will need
Cybersecurity requirements are and deliver them as comprehensive, leading the charge into the future of GenAI will continue to be implement- pose and deliver these processes in a to be updated — measures like simple
becoming more intricate, network manageable services. These vendors connectivity.#1: How Automation and ed in certain use cases, especially standardized, consumable fashion. cost-saving, staff reduction, inventory
optimization demands are increasing, are partners in innovation, enabling AI Will Shape Autonomous Networks around customer interactions, but it audits, energy efficiency initiatives and
will do little to grow revenue near term, In addition, to support the delivery of decommissioning legacy devices have
so service providers will likely look to new products, telecommunications not moved the needle enough.
implement other AI-driven solutions service providers will continue to invest
BY ALI SHAIKH
alongside LLMs to drive growth. For in re-tooling the OSS stack to handle Significant opex reduction will be
Ali Shaikh, CPO, is responsible for the strategic direc- example, AI & ML will provide valuable faster changes to services and billing. driven by re-engineering the telco op-
tion for Graphiant, technical alliances, and driving the insights to providers looking to expand erating model around automation and
product roadmap. He led Systems Engineering for their business by better understanding Let’s dive deeper into the key automa- AIOps. Over the next 12 months, expect
Meraki & Viptela at Cisco and was the first Solution how their most profitable customers tion trends we’ll see in the telecommu- companies to make major strides to-
Architect and Customer Success Engineer at Viptela. are consuming their services. nications industry over the next year: wards multi-domain orchestration and
He was responsible for designing and deploying the the ability to embed AI-driven analysis
largest SD-WAN implementations globally. Automation and orchestration will #1: TELCOS WILL RE-ENGINEER OPERATING into every infrastructure change.
continue to grow, as they remain the MODELS AROUND AUTOMATION & AIOPS
keys to unlocking new revenue as #2: FOCUS WILL EXPAND BEYOND GENERIC
CSPs continue to invest in cloud, edge, Recent years have seen CSPs priori- FUNCTION APIS TO DRIVE REVENUE

64 65
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

Recently, CSPs have been focused on

the standardization of generic function
demands of advanced enterprise
customers, SPs must pivot. The next
changes across complex infrastruc-
ture required to deliver network as a AGENTIC AI CAN CHANGE NETWORKS
IF THEY CLEAR A HIGH HURDLE
APIs to enable more development and wave of innovation will be the delivery service.
innovation across the network — but of complex, dynamic, customizable ser-
this area of focus has not unlocked vices via consumable APIs to meet the To move forward, providers must
significant revenue. needs of rapidly innovating customers undertake significant re-tooling efforts,
and quickly grow revenue for providers. rethinking their foundational systems
While there are a few areas where and processes. This transformation Written by Ed Fox
these areas will deliver business #3: NAAS PRIORITY WILL DRIVE MAJOR RE- will be essential to unlocking the full Chief Technology Officer, MetTel
benefits, particularly with APIs related TOOLING OF LEGACY OSS STACKS potential of NaaS, enabling providers to
to SIM management, the industry has deliver flexible, scalable, and custom-
largely adapted around the limitations Network-as-a-Service (NaaS) remains er-centric services that align with the
of traditional service provider infra- a key focus for service providers, but demands of a rapidly changing market. The rise of agentic AI has the potential generation stacks and sophisticated be those that go the partnership route.
structure, introducing alternative solu- implementing it successfully is de- to transform boring static networks data architectures – will be a significant Successful implementations will rely
tions for basic network functions. manding greater transformation than into dynamic, self-optimizing entities. hurdle. The current hype surrounding heavily on collaboration with experi-
initially anticipated. Crucially, exist- Imagine networks that anticipate con- agentic AI, fueled by the desire to push enced vendors and systems integrators
Generic function APIs are not where ing service provider OSS stacks lack gestion, autonomously reroute traffic, generative AI beyond simple tasks, is who can navigate the complexities
the value lies. To meet the growing the flexibility to support the speed of and proactively defend against cyber leading to inflated expectations. Build- and deliver tailored solutions. Savvy
threats, all driven by AI agents pursu- ing these agents requires specialized businesses will recognize their current
ing user-defined goals. expertise and aligning diverse models limitations and leverage these part-
BY MORGAN STERN
which will not be easy. nerships to build innovative agentic
Morgan Stern serves as the Vice President of This transformation, however, is fraught networks, while others struggle to
Automation Strategy for Itential, where he is with challenges. The complexity of Organizations will try to build their catch up. We are still in a period of
responsible for assisting Itential’s global customers agentic architectures – requiring own agentic network architectures, but experimentation, but a greater, smarter
in developing, implementing and deriving value from intricate interplay between multiple AI I fear a lot of those will fail to hit the network is within our sights.
their automation strategies using Itential’s platform. models, several retrieval-augmented mark. The ones that will succeed will
For the majority of his career, Morgan has focused
on assisting large enterprise and service provider
organizations in maximizing the business impact of
new technologies as an architect, consultant, author,
and industry speaker.

66 67
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

#1: ENTERPRISES CHOOSE PREDICTIVE OVER

GENERATIVE AI
chains and accurate demand forecast-
ing. This shift will draw resources away
AI/network combination has already
shown immediate, positive results. 5 WAYS AGENTIC AI AND INTUITIVE NETWORKS
WILL REDEFINE TELCO SAAS IN 2025
from purely GenAI projects. However, A report from McKinsey found that
Right now, there is a near-equal split the most innovative organizations will applying GenAI to customer care
between predictive and generative AI recognize the symbiotic potential of functions could increase productivity
in enterprise applications, but I see a these two AI paradigms. Rather than by 30-45 percent. AI is deployed either
shift coming. While GenAI is the talk of viewing them as competing forces, as a virtual assistant to human agents
town, practical considerations will soon some will leverage the strengths of or as virtual agents interacting directly Written by Pramod Gummaraj
find enterprises choosing predictive each. Predictive AI can be used as the with the customer. Enterprises will have Founder & Chief Executive Officer, Aprecomm
AI over GenAI. Predictive AI has an engine to refine and enhance the out- options depending on how sophisti-
established history of success with put of GenAI. That synergy will promote cated their AI is. Those outcomes may
tangible results like fraud detection hybrid applications to pop onto the include real-time, precise answers to
and personalized recommendations. scene throughout the year. customer questions or routing custom-
The ROI is more assured and easier to ers to the correct call agent. It is now
quantify, its operation is controllable #2: GENAI’S RELEVANCE IN CUSTOMER CARE much easier to deploy these AI tools in
and transparent and it is easier to inte- the contact center. They require little
grate applications into legacy systems As mentioned, the priority on AI ROI to no coding experience and leverage
with less overhead coming from costs will cause enterprises to scale back existing company data to train the
and specialized skills. their network experiments. They will learning models. AI-powered contact
be looking at projects that deliver centers are an easy, safe lift we will see
Predictive AI investments will move to- real-world outcomes. In certain areas a lot of in 2025.
ward established applications like pre- GenAI can still deliver this. One is with
dictive maintenance, optimized supply AI-powered contact centers where the

BY ED FOX

Ed Fox is Chief Technology Officer at MetTel. Ed

is responsible for the planning, deployment, and
operations of MetTel’s broadband, data, and VoIP
network infrastructure. He is a member of the Forbes
Technology Council and a contributing author for
IDG influencer network. He has over 30 years of tele-
communications and network experience managing
massive organizational and customer growth at
major telecommunications providers.

The telecom industry is undergoing evolution from traditional networks to alized ways.
a profound transformation, powered more adaptive, self-learning systems
by advanced technologies such as marks the beginning of a new era. Tel- Here are five unique trends in Telco
Agentic AI and the rise of what we call cos will no longer operate in a reactive SaaS that will be propelled by Agentic
“Intuitive Networks”. As we approach manner but will leverage AI-powered AI and Intuitive Networks, setting the
2025, these innovations w ill reshape solutions that allow them to anticipate stage for a more agile, responsive, and
the telecom SaaS landscape in ways challenges, optimize services, and en- customer-centric industry.
that were previously unimaginable. The gage with customers in highly person-

68 69
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

#1: HYPER-CONTEXTUAL NETWORK telecom providers have relied on static customer needs. For example, an AI communication, Intuitive Networks As networks become more complex work behavior and adapt their security
AUTOMATION: THE BIRTH OF FULLY INTUITIVE models to forecast network usage and assistant could recognize a customer will dynamically allocate bandwidth and interconnected, traditional se- measures accordingly. For instance, if
NETWORKS allocate resources. These methods are experiencing a poor experience on a and prioritize traffic flows in real time curity systems struggle to keep pace AI detects unusual traffic patterns or
increasingly obsolete in a world where given application and autonomously to ensure the highest level of service. with evolving threats. Agentic AI and attempts to bypass security protocols,
In 2025, telecom providers will move customer behavior is unpredictable and resolve the issue by rerouting traffic or This will dramatically improve service Intuitive Networks will take a radically it will automatically adjust network
away from traditional network manage- demand for network services grows reconfiguring the network to improve quality and customer satisfaction, while new approach to network security defences, implement countermeasures
ment practices towards fully intuitive exponentially. service. allowing telcos to maximize their oper- by shifting from reactive to proactive like enhanced encryption, or isolate
networks—systems that autonomously ational efficiency. threat mitigation. affected areas to mitigate risk.
adjust in response to both external and Intuitive Networks, enhanced by Agen- These virtual assistants will act as intel-
internal conditions. These networks will tic AI, will leverage machine learning to ligent agents that understand not only #5: INTELLIGENT NETWORK SECURITY: AI-powered security systems will con- This proactive, real-time approach
monitor performance and actively learn predict network congestion, usage pat- the customer’s preferences but also the PROACTIVELY MITIGATING THREATS WITH tinuously scan the network for vulner- to network security will significantly
and adapt to changes in traffic pat- terns, and potential bottlenecks before nuances of the network and service AGENTIC AI abilities, detect anomalies, and predict reduce the risk of cyberattacks and
terns, demand fluctuations, and service they occur. By continuously processing conditions in real-time. This shift will potential security breaches before they ensure that telecom providers can
quality. real-time data, AI can anticipate future make customer interactions seam- Cybersecurity is one of the most press- happen. Intuitive Networks will enable maintain the integrity of their services
demand spikes, adjust routing paths, less, intuitive, and highly personalized, ing concerns for telecom providers. these AI systems to learn from net- without manual intervention.
Agentic AI will drive this transformation and dynamically reconfigure network dramatically improving satisfaction
by powering self-optimizing systems topology to optimize capacity. and reducing service costs for telecom
BY PRAMOD GUMMARAJ
that can detect network anomalies, providers.
predict issues before they occur, and For instance, AI-powered network in- Pramod Gummaraj is the Founder & Chief Executive
implement corrective measures without telligence will predict when certain cell #4: REAL-TIME RESOURCE ALLOCATION VIA Officer of Aprecomm. Pramod began his career with
human intervention. The most powerful towers or network segments are likely AI-DRIVEN DEMAND SENSING Qualcomm, a forerunner in wireless communications,
realization of this will come through to face higher demand and proactively where he pioneered Artificial Intelligence (AI) and
Intuitive Networks—network architec- deploy additional resources or adjust The ability to dynamically allocate machine learning (ML). In 2016, Pramod co-founded
tures that continuously use advanced configurations to ensure uninterrupt- resources based on real-time demand Aprecomm, leveraging his expertise in wireless tech-
AI algorithms to monitor and adjust ed service. This self-optimization will will be a game-changer for Telco SaaS. nologies and AI, in which he holds multiple patents.
network elements based on real-time significantly reduce operational costs Traditional methods of resource man- Aprecomm is already recognized for its leadership in
conditions. and enhance the user experience by agement, while effective, cannot match creating software to promote self-healing broadband
ensuring consistent service quality, the agility that is required in the fast- networks. Over 45 service providers use its customer
For example, Intuitive Networks could even during peak times. paced digital world. In 2025, Agentic AI, experience suite to manage over 7 million homes and
automatically detect network conges- integrated with Intuitive Networks, will businesses worldwide.
tion caused by an unexpected surge in #3: AUTONOMOUS CUSTOMER INTERACTION: revolutionize how telecom providers
traffic due to a live event, adjust capaci- THE RISE OF AI-DRIVEN VIRTUAL NETWORK manage and allocate resources.
ty in real-time, and allocate resources ASSISTANTS
dynamically to maintain optimal service With the power of real-time demand
levels, all without requiring manual AI increasingly drives customer sensing, Agentic AI will continuously
input. This radical shift will revolution- engagement in the telecom space. monitor customer activity, network
ize network reliability and efficiency, However, while AI chatbots have been health, and environmental variables
allowing telecom providers to scale used for years, Agentic AI will introduce to make instantaneous decisions on
rapidly without compromising service a new level of sophistication. In 2025, resource allocation. This means that
quality. telecom providers will deploy AI-driven network operators will no longer need
virtual assistants that can proactively to rely on pre-defined rules or outdated
#2: PREDICTIVE SELF-OPTIMIZATION: resolve customer issues, offer person- assumptions to scale their services.
REVOLUTIONIZING TELCO SAAS FOR alized plans, and assist with trouble- Instead, the network will “sense” de-
NETWORK CAPACITY AND UTILIZATION shooting—all autonomously. mand fluctuations and adjust capacity
automatically.
One of the most significant ways Intuitive Networks will support this
Agentic AI and Intuitive Networks will transformation, which will continuously For example, during a major
redefine Telco SaaS is by enabling pre- analyze customer data, network condi- live-streaming event or an emergen-
dictive self-optimization. Traditionally, tions, and service history to anticipate cy situation that requires immediate

70 71
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

BEYOND THE HYPE: HOW AI WILL SHAPE com data into actionable business in-
telligence. Network operators will gain
and more efficiently. From identifying
signaling storms to adjusting for band-
behavior and network performance to
spot and act on new revenue opportu-

TELECOM INFRASTRUCTURE IN 2025

new insights into customer lifetime width spikes from new device launches nities.
value and identify service opportunities or fraudulent network usage, AI will
that were previously hidden and uncor- optimize network performance while #4: AI WILL ADVANCE CUSTOMER
related in siloed data environments. determining cost-effective operational EXPERIENCE
strategies.
Written by Adil Belihomji However, this advancement comes AI’s role in telecommunications
Chief Technology Officer, OXIO with important caveats. Network oper- The industry acknowledges the risks customer service will expand in 2025,
ators must carefully validate AI-gener- in this transition. As telecom providers moving beyond today’s basic chatbots.
ated insights, as LLMs can hallucinate expand their AI capabilities, they must AI assistants will act as problem-solv-
with a high degree of inaccuracy. The balance automation with transparent, ers, analyzing usage patterns and
While many focus on AI’s role in ate massive amounts of data, but this data. By recognizing patterns across challenge of 2025 will be striking the unbiased decision-making that meets suggesting plan adjustments based on
customer service chatbots or network information remains trapped in silos, previously disconnected data points, right balance between AI automation regulatory requirements. individual needs.
monitoring capabilities, the underlying limiting its utility. In 2025, a significant these systems will enable proactive and human oversight in critical network
changes run deeper: AI will help alter shift will start to occur as Large Lan- maintenance and help predict network operations. #3: AI WILL RESHAPE TELECOM ECONOMICS These systems will graduate to ad-
how telecom services are structured, guage Models (LLMs) begin integrating failures, reducing service interruptions. dressing complex issues, including
delivered and optimized. Here are our across the entire telecom stack, chang- The technology will also optimize #2: AI WILL ENABLE AUTONOMOUS AI will enable new approaches to tele- billing discrepancies, device trouble-
predictions for 2025. ing how networks operate and adapt network resources automatically, using NETWORK OPERATIONS com pricing and delivery in 2025. Static shooting and service disruptions more
real-time usage patterns and predicted pricing and service models are shifting efficiently than current solutions. By
#1: AI WILL ENHANCE NETWORK LLM-powered systems will enhance demand to improve capacity allocation Self-healing networks will emerge in toward dynamic, AI-driven systems connecting with autonomous network
INTELLIGENCE network management through conver- and cost management. 2025 as AI takes on a larger role in that align with network capabilities and operations, these assistants can identi-
sational interfaces, allowing adminis- network operations and performance customer needs. fy and resolve problems in a contextual,
Traditional telecom networks gener- trators to interact naturally with their These systems will convert raw tele- management. Telecom providers are conversational, human-like interface
moving beyond basic automation This change is evident in how telcos and in most cases resolve them before
toward autonomous networks that can are adapting to bandwidth-intensive customers notice them.
predict, prevent and resolve issues with applications. Gaming illustrates the
minimal human intervention. trend: as 5G networks mature, pro- Mobile Virtual Network Operators
viders need pricing models suited to (MVNOs) will introduce AI co-pilots
The addition of AI systems, such as low-latency, high-throughput scenarios. that change how customers interact
Agentic AI, that interact with a combi- The implications extend beyond gam- with their providers. These assistants,
nation of other AI technologies includ- ing to XR (Extended Reality), enterprise available through mobile devices,
ing machine learning, natural language services, private 5G networks, cross will offer immediate support without
processing and automation technolo- border plans, IoT and V2X (Vehicle to requiring traditional customer service
gies, allows decisions around network everything) deployments. channels. They’ll function as service
optimization and performance to be advisors, using network data and
made autonomously without constant AI will guide this economic evolution economic insights to customize service
human guidance and attention. Agentic by helping carriers implement respon- recommendations.
AI will make decisions and adapt to sive pricing strategies. These systems
changes autonomously based on pre- will adjust pricing based on capacity Looking ahead
dictions and human behavior, learning demands, service quality metrics and
and improving from these interactions usage patterns, identifying revenue While these changes won’t fully mature
as it goes to proactively solve complex opportunities while optimizing network by 2025, the year will mark AI’s shift
network problems independently. resources. from experimental to established
technology in telecom infrastructure,
These types of systems mark an im- The transition moves telecoms from alongside a paradigm shift from AI as-
portant shift in network management, fixed service tiers toward flexible, sisting Humans to Humans assisting AI.
analyzing network data to prevent po- value-based models. AI systems will Success will come to organizations that
tential issues or resolving issues faster analyze market conditions, customer grasp AI’s broader potential: moving

72 73
The Fast Mode Trends and Outlook 2025

beyond automation to reshape how and AI validation requirements. Yet the As we approach 2025, the question be-
telecom services are delivered. benefits—improved network efficien- comes not whether AI will alter telecom
cy and personalized services—make infrastructure but how organizations
This evolution brings challenges, from adoption necessary. will adapt to and use these capabilities
privacy concerns to data governance to drive better business performance.

BY ADIL BELIHOMJI

CUSTOMER
Adil Belihomji is the chief technology officer at OXIO,
where he leads the technology vision and strategy
for the company’s global telecom-as-a-service (TaaS)
platform. He plays a pivotal role in driving product

EXPERIENCE
innovation while overseeing the development and
implementation of technology solutions.

1. The 2025 Forecast: Continued AI Domination Across the Global Telco

Landscape��������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������.76

2. Using AI to Extend Customer Care to Deliver Additional Business Value�����������������������������.79

3. How Will the Mobile Industry Shape Up in 2025?������������������������������������������������������������������������������������������.81

4. How Digital Identity, Messaging, and Satellite Communications Will Evolve in

2025������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������.84

5. From Faster to Smarter: The Telco Industry’s AI Revolution���������������������������������������������������������������.86

6. Five Business Communication Trends for 2025��������������������������������������������������������������������������������������������� .88

7. Generative AI for the Sceptics: Advice for 2025 from a Veteran Unbeliever�����������������������.90

74 75
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

THE 2025 FORECAST: CONTINUED AI port sales and finance teams as they
manage orders and renewals, enhance
exchange information between these
systems, offering enhanced flexibility
chatbots and co-pilots is rising, partic-
ularly as AI vendors, responding to the

DOMINATION ACROSS THE GLOBAL TELCO

revenue assurance and minimize reve- in developing comprehensive AI skills high costs of infrastructure-as-a-ser-
nue leakage. Increasingly, subscribers without extensive integrations. Think vice (IaaS), shift toward consump-
engage with chatbots and co-pilots AI talking to AI, but with the ability to tion-based pricing models. This chal-

LANDSCAPE
to address billing or usage inquiries action activities as well. lenge is compounded by growing user
or change plans - without waiting to adoption and comfort with AI-driven
speak with a human agent. For CSPs, agentic RAG-based AI offers solutions. While $1–$2 per AI engage-
numerous advantages, including agile, ment (a current pricing theme) may
In 2024, we started to see some vali- scalable and expansive capabilities that seem modest, the cumulative costs for
Written by Akil Chomoko dation of the effectiveness and value of enable both responsive and proactive CSPs with millions of subscribers can
Vice President of Product Marketing, Aria Systems chatbots within telecom environments, operations, particularly when data, quickly result in significant expenses.
with these tools demonstrating their events, and records are dispersed
ability to streamline processes, improve across various AI-enabled systems. In 2025, CSPs are likely to place
customer satisfaction, and optimize By monitoring extensive activities and greater emphasis on monitoring and
operations. In 2025, the deployment of diverse data patterns, and seamlessly controlling AI-related costs. This could
The rapid rise of generative and pre- Despite these advancements, however, #1: GENAI CHATBOTS AND CO-PILOTS WILL generative AI-based chatbots will shift connecting to intelligence sources via involve adopting stricter measures
dictive AI has been a dominant theme the telecom sector is still in the early PROVE THEIR WORTH from being a competitive advantage to RAG to bridge information gaps, agen- to ensure subscriber and employee
across the global business landscape, stages of its AI transformation, with the a fundamental necessity. CSPs yet to tic AI applications can proactively com- interactions with AI tools are limited
and especially within the telecommu- potential applications and impacts of AI The widespread adoption of AI-pow- adopt this technology will likely find it municate pertinent information and to high-value inquiries. For example,
nications industry, over the past year. still largely untapped. As we head into ered chatbots and co-pilots continued indispensable for meeting subscriber options to subscribers. This facilitates CSPs may choose to deploy more
Communication service providers 2025, the industry can expect further to accelerate this past year, particularly expectations and maintaining oper- a true concierge service, automatically streamlined, cost-effective natural
(CSPs) have begun to harness AI to AI-driven innovation, which will shape in the telecom industry. Chatbots are ational efficiency in an increasingly informing subscribers about network language processing models that
enhance customer support, drive sales how CSPs interact with customers, now integral to various functions. They AI-driven industry. outages, presenting personalized deliver essential capabilities within the
and business development, analyze optimize operations, and redefine their assist customer service representatives offers, suggesting optimal plans based telecom context, without incurring the
data for better decision-making, and value propositions. by providing critical information for #2: AGENTIC AI TO CHANGE THE DYNAMIC on usage, or alerting them to account expense of fully open systems.
achieve significant cost savings. timely inquiry resolution, and they sup- or service issues, often before they are
Historically, AI implementations have even aware there is a problem, using #4: REGULATION AND GLOBAL GOVERNANCE
maintained human oversight, with sys- the complete CSP ecosystem. CATCH-UP
tems like co-pilots or AI bots requiring
human intervention when the applica- In 2025, agentic RAG-based AI will Governments and regulatory bodies
tion encounters a challenge. However, become more prominent in the tele- often lag behind technological innova-
as AI’s accuracy and capabilities ad- coms world, offering a more engaging, tors, and this holds true for AI. These
vance, the industry is already evaluat- relevant, and personalized AI-powered organizations are working to keep pace
ing agentic AI - autonomous systems interaction for both subscribers and with AI advancements, trying to under-
that interact directly with customers, employees. stand its transformative power, while
stakeholders, and other AI entities assessing the associated risks, and
through natural language to perform #3: CONFRONTING THE FINANCIAL REALITIES determining how to best protect both
tasks, respond to inquiries, and initiate OF AI consumers and businesses.
proactive outreach based on customer,
service, and billing events – without We’ve established that AI will become The European Union took a signif-
humans in the loop. indispensable for CSPs in 2025, en- icant step forward in August 2024
abling streamlined operations, auto- with the introduction of the EU AI Act,
A notable development in this area is mated processes, revenue protection, the first comprehensive AI regulatory
Agentic Retrieval-Augmented Gen- enhanced customer satisfaction, framework. This legislation establishes
eration (RAG), which moves beyond and reduced churn. However, these uniform guidelines across all 27 EU
rigid API integrations or bulk data transformative benefits come with a member states for the responsible and
exchanges between a CSP’s systems. significant financial burden. ethical development of AI while safe-
Instead, it employs sophisticated natu- guarding against potential harm.
ral language AI functions to retrieve or The expense of running AI-powered

76 77
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

Historically, the EU has set a precedent

for creating regulations that influence
lated worldwide, and the EU AI Act is
expected to follow a similar trajectory.
themselves with the Act. Early prepa-
ration will enable them to protect their USING AI TO EXTEND CUSTOMER CARE TO
DELIVER ADDITIONAL BUSINESS VALUE
global standards, as was seen with In 2025 and beyond, other nations and businesses, employees, and custom-
the General Data Protection Regula- regions are likely to adopt comparable ers, while ensuring they capitalize on
tion (GDPR). The GDPR’s principles of AI regulations, so CSPs operating out- AI’s transformative opportunities in a
data privacy and security were emu- side of the EU are advised to familiarize responsible manner.

Written by Martin Morgan

BY AKIL CHOMOKO
Head of Digital Marketing, Qvantel
Akil Chomoko is Vice President of Product Marketing
at Aria Systems, where he is responsible for building
Aria’s go-to-market strategies. He also leads the
global marketing team and program. Akil brings 20+
years’ experience from the telecoms and technology
industry, serving most recently in senior product
marketing and management positions at MDS Global,
AsiaInfo and CSG.

At the recent AT&T analyst and inves- to measure the success of any new perience is a very wide term and could
tor day, AT&T CTO, Jeremy Legg out- technology always makes sense. All the cover every touch point (from web
lined how to measure the success of AI five key measures listed above can be site, to app, to customer care to the
and GenAI with five simple guidelines. impacted by customers’ experiences network) where a customer interacts
They are: when engaging with their CSPs. This with their CSP. What is interesting is
• Increase in number of customers would explain why the leading use of AI to look at where CSPs are investing in
taking convergent offers in telecoms is focused on the continu- GenAI to identify which specific areas
• Increase in net promotor score ous quest to deliver a better customer of customer experience optimisation
(NPS) experience. are highest up the priority list for CSPs.
• Increase in penetration The report showed that the majority of
• Increase in ARPU According to Nvidia’s 2024 State of AI CSPs (57%) investing in generative AI
• Increase in customer lifetime value in telecoms report 49% of respondents to improve their customer service and
are using AI to deliver a better custom- support.
Taking a business value approach er experience. However, customer ex-

78 79
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

#1: GENAI SUPPORTING SELF-CARE AND

ASSISTED CUSTOMER CARE
customer. This reduces the time for
the customer care rep to solve cus-
rep cannot provide the answers to the
questions that the customer has. HOW WILL
THE MOBILE
tomer problems and enable the rep to
GenAI customer care copilots are possibly use the opportunity to engage Extending customer care with AI tech-
being rolled out to support customer the customer for possible upselling nical care solves this problem. AI tech-

INDUSTRY
self-care and assisted (human) care. scenarios. nical care analyses network elements
Chatbots and other digital assistant to identify network issues and identifies
services are the first line of customer By using a combination of GenAI based the customers and services impacted.

SHAPE UP IN
support and a GenAI copilot enables copilot and human customer care The customer care copilot analyses the
a rich set of data to be analysed and agents first call resolution rates and customers affected and checks if any
used to answer questions and resolve call throughput can increase, which SLAs (service level agreements) have

2025?
issues via the chatbot. However, often has a positive impact on NPS, reduces been impacted and sends a notification
human (assisted) customer care is propensity to churn, increases upsell to the relevant customers. AI technical
also needed, and GenAI customer care opportunities and drives customer care care then identifies the root cause of
copilots can increase efficiency and de- costs down. the network problem and decides on
liver quality improvements for assisted what next best action to take. This ac-
customer care. #2: AI TECHNICAL CUSTOMER CARE – THE tion is then carried out via AI technical Written by Dario Betti
NEW FRONTIER care and OSS, and then the impact of CEO, MEF
When a customer engages with a the action is verified. When AI techni-
customer care rep, the GenAI custom- The next level of AI driven customer cal care sees that the issue has been
er care copilot can summarise and care is technical customer care, where resolved the customer care copilot then
process data (such as interactions in care reps can answer questions about notifies the customers that the network 2024 was a busy year for the telecoms
the chatbot and other channels) and complex network issues. Currently, in issue has been resolved. Also, the care sector but also a challenging one with
provide help for the care rep. This can many CSPs when a customer calls a reps are supported by AI driven cus- the continued high cost of living and
include providing routing of the call to care rep to complain about a network tomer care copilot to provide answers higher than hoped interest rates inhib-
the correct care rep, sentiment analysis or service issue, the customer service for any customers who get in touch to iting market growth.
(e.g. based on tone of language used rep typically sends a ticket to 2nd level ask about this network issue.
on the chatbot), summary of previous care. This is typically an engineer who So, will 2025 be a better year? And
interactions, intent recognition and works in the network operations centre Using GenAI customer care copilot to what might it hold for the mobile in-
templated responses / smart sugges- who will find out what the problem improve customer care performance dustry?
tions to the identified issue the cus- is and when it can be resolved. This can deliver the business value need-
tomer is likely to have. This provides can often be a complex and time-con- ed to help measure the success of AI. #1: RCS WILL REACH CRITICAL MASS,
analysis of the customer problem and suming process, involving a multitude Adding AI driven technical customer SLOWLY
a recommendation (resolution to the of data sources and root causes that care to the overall care process can
problem or next best activity) and is may need to be investigated to find out add even more business value and Rich Communication Services (RCS)
made available for the care rep to refine the source of the customer’s problem. deliver a better care experience for is set to achieve significant global
(if needed), discuss and present to the This means that the customer service customers. adoption, becoming the standard for
branded customer interactions, driven
by enhanced functionality and security
BY MARTIN MORGAN
features. However, this growth will be
Martin Morgan is the Head of Digital Marketing gradual rather than explosive.
at Qvantel. With 30 years’ experience in mobile
communications software, Martin has worked in As MEF board member and iTouch
mobile since the early days of the industry. He’s ran Chairman Waheed Adams aptly puts
the marketing teams for several BSS companies and it: “We are entering the first phase of
served on trade association and company boards. sector enablement, with more MNOs
(Mobile Network Operators) enabling
RCS on their networks and platforms
being refined. However, commercial

80 81
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

models remain uncertain, and the customer education, unclear bundling Wireless Access (FWA) will challenge Governments worldwide will enforce will surpass $2 billion in market value There is no doubt that 2025 will be
guest mode does not incentivise MNOs propositions, and a lack of effective traditional broadband providers by stricter regulations on data collection in 2025, with some analysts projecting another busy year for the telecoms
to fully support RCS. While aggregators pricing strategies. offering high-speed, cable-free connec- and sharing, driving the adoption of pri- it could rival the music industry by sector, and those who understand and
are raising market awareness, MNOs tivity, with adoption already reaching 10 vacy-first technologies. Businesses will 2030. Streaming platforms like Twitch prepare for coming trends and devel-
need to take a stronger role for rapid To unlock their full potential, the in- million US households. need to innovate responsibly, balancing and YouTube Gaming are attracting opments within the mobile ecosystem
adoption. That said, exceptions exist in dustry must better communicate API compliance with consumer trust. millions of daily viewers. Expect more will emerge as key market players while
markets where RCS is gaining traction benefits, develop tailored offerings, Meanwhile, satellite services, such as recognisable esports commentators, those who don’t, may be left behind.
faster. With iOS coming onboard, 2025 and experiment with innovative pricing. Direct-to-Device (D2D) connectivity, #10: THE RISE OF ESPORTS globally-viewed events, and an expan-
will likely be the year RCS reaches its Collaboration across the ecosystem will gain momentum, with more focus sion of sponsorship and advertising
potential, setting the stage for signifi- will pave the way for standardised on deployment than on early 6G dis- Esports, or competitive video gaming, opportunities.
cant industry growth.” frameworks, allowing APIs to power cussions. These developments will un-
new business models in industries like lock new opportunities for autonomous
BY DARIO BETTI
#2: REGULATORY INTERVENTION – THE banking, healthcare, and logistics. vehicles, smart cities, and immersive
GLOBAL INTERNET BECOMES MORE LOCAL experiences. Dario Betti is CEO of MEF (Mobile Ecosystem Forum)
#4: A GLOBAL PUSH TOWARD COMBATING a global trade body established in 2000 and head-
Regulatory activity in the telecom MOBILE FRAUD #7: MOBILE IDENTITY: EMERGING UNIVERSAL quartered in the UK with members across the world.
sector is poised to intensify, with major STANDARDS As the voice of the mobile ecosystem, it focuses
developments from the FCC (Federal Mobile fraud losses are expected to on cross-industry best practices, anti-fraud and
Communications Commission) in the surpass $100 billion in 2025. Howev- Universal mobile identity solutions monetisation. The Forum provides its members with
US and the EU’s Digital Services Act er, the industry is taking a proactive will become the leading standard for global and cross-sector platforms for networking,
(DSA) and Digital Markets Act (DMA). stance. AI will play a key role on both authentication, reducing reliance on collaboration and advancing industry solutions.
The FCC, following new commission- sides of the battle, with advanced tools traditional passwords. Technologies
er appointments, will likely enhance used to detect and counter fraudu- like Apple Wallet, now open to external
oversight in areas like robocall mitiga- lent activity. Stricter regulations and developers, are fostering innovation in
tion, spectrum management, and net fraud-prevention alliances will create digital identity and payments. As gov-
neutrality. Meanwhile, the EU’s regula- safer mobile environments, making ernments and enterprises adopt these
tions target gatekeepers, emphasising 2025 a pivotal year in fighting bad solutions, services such as e-gover-
transparency, curbing monopolistic actors. nance, banking, and healthcare will
practices, and ensuring stricter data become more secure and accessible.
privacy. #5: HYPER-PERSONALISATION VIA AI AND
MACHINE LEARNING #8: BAD NEWS FOR CASH: MOBILE
Cybersecurity and national security are PAYMENTS DOMINATE
now central concerns for regulators AI-driven personalisation will deliver
and telecom companies alike. These tailored content, services, and advertis- Mobile wallets are rapidly becoming
evolving regulations highlight the need ing with unparalleled precision. While the primary financial tool globally. In
for compliance strategies and collabo- user satisfaction will increase, height- the UK, cash use fell to 12% of total
ration between policymakers and the ened scrutiny over data privacy and payments in 2023 and is projected to
industry. intellectual property will accompany drop to single digits by 2025. Con-
these advancements. Expect a year tactless payments already account for
#3: THE MATURING OF NETWORK APIS of experimentation, with users enjoy- 38% of transactions, and a majority
ing increasingly intelligent AI-driven of smartphone users will soon rely on
Network APIs (Application Program- services. mobile wallets for daily payments. This
ming Interface) are increasingly vital for shift enhances financial inclusion while
enabling real-time services, enhanc- #6: 5G MATURES, AND SATELLITE PLANNING necessitating stronger fraud protection
ing user experiences, and combating TAKES CENTRE STAGE measures.
fraud through applications such as
authentication and traffic management. By the end of 2025, 5G promises to #9: INCREASED REGULATORY SCRUTINY ON
However, adoption remains limited deliver on its potential, with improved DATA PRIVACY
due to challenges such as insufficient speeds and reliability globally. Fixed

82 83
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

HOW DIGITAL IDENTITY, critical platforms. Tapbacks over iMessage and SMS.
Initially available in the U.S. and Cana-
represents a leap forward in achieving
universal accessibility, making it one of

MESSAGING, AND SATELLITE

This inertia will maintain the promi- da for iPhone 14 models and later, this the most exciting frontiers in telecom-
nence of Application-to-Person (A2P) move highlights the growing viability of munications.
SMS and MMS messaging in the near satellite communication using off-the-

COMMUNICATIONS WILL
term. As enterprises take a cautious shelf devices. Final thoughts
approach to upgrading their systems
to leverage RBM’s rich media and The real-world benefits of this tech- The year ahead promises transforma-

EVOLVE IN 2025
interactive capabilities, we can expect nology were demonstrated during the tive advancements in how we interact
continued growth in the A2P mes- hurricanes in Florida in the fall of 2024, with technology. Decentralized identity
saging market over the next 12 to 24 where satellite-enabled messaging ecosystems, the gradual evolution of
months. While the promise of RCS is provided critical connectivity in disaster enterprise messaging, and the prolif-
undeniable, the timeline for widespread zones when traditional networks failed. eration of satellite communications are
Written by Stephen Zitnik adoption and its full market impact Apple’s innovation underscores the three areas where we’ll see significant
Executive Vice President and Chief Technology Officer, Interop Technologies may stretch into 2026 and beyond. growing viability of satellite communi- progress. These innovations will not
cation using off-the-shelf devices, and only shape the tech industry but also
RCS will enhance, personalize, and it is only the beginning. deliver meaningful benefits to individu-
As we approach 2025, the technolo- one-time passwords (OTPs) over SMS #2: RCS MESSAGING WILL REMAIN ON THE transform all aspects of messaging, but als and organizations worldwide.
gy landscape is poised for significant or email. Instead, biometrics—such HORIZON AS A2P SMS/MMS HOLDS STRONG it will have to move an entire ecosys- As competition grows, the next logical
transformation across key areas that as facial recognition and fingerprint tem—and for that, we need patience. step will be the introduction of satellite As we navigate these changes, enter-
impact enterprises, consumers, and scans—will become the cornerstone of Rich Business Messaging (RBM), the voice capabilities, further enhancing prises that prioritize agility and innova-
mobile operators alike. From the evo- secure and seamless authentication. enterprise-grade iteration of Rich Com- #3: SATELLITE COMMUNICATION WILL global connectivity. The implications tion will be best positioned to leverage
lution of digital identity to messaging While this evolution is promising for munication Services (RCS), has been REVOLUTIONIZE CONNECTIVITY are profound: improved network re- emerging technologies for competitive
innovations and advances in satellite consumers and enterprises, it poses a long time coming in the U.S., with full silience, enhanced public safety, and advantage. The future is decentralized,
communications, these changes prom- challenges for wholesale SMS provid- nationwide enablement across carriers In September 2024, Apple expanded expanded coverage, particularly in interconnected, and powered by the
ise to redefine how we connect, trans- ers that rely on OTP traffic. and compatible handsets projected for its satellite messaging capabilities with underserved and remote regions. innovations we embrace today. Those
act, and secure our digital experiences. 2025. Despite this milestone, the adop- the launch of iOS 18, allowing users who act now will lead the way into this
Here are my top three predictions for The result? A future where users have tion of RBM capabilities by enterprises to send and receive texts, emojis, and Satellite-enabled communication transformative new era.
the year ahead: greater privacy and enterprises benefit will lag due to the deeply integrated
from lower fraud rates, while the tech- nature of existing messaging APIs
#1: THE DIGITAL IDENTITY REVOLUTION: nology ecosystem pivots to support within customer relationship manage-
DECENTRALIZED AND BIOMETRIC this paradigm shift. ment (CRM), billing, logistics, and other
ECOSYSTEMS WILL DOMINATE
BY STEPHEN ZITNIK
The digital identity landscape is on the Stephen J. Zitnik, Executive Vice President and Chief
cusp of a transformation, driven in part Technology Officer at Interop Technologies, combines
by Apple’s groundbreaking decision technical and financial expertise to lead software
to open its Near Field Communication development, engineering, and IT strategy. A respect-
(NFC) and Security Element (SE) func- ed authority in mobile messaging, he has chaired
tionality to third-party developers. This numerous working groups, task forces, and steering
move is a catalyst for a shift toward committees for prominent telecom standards bodies
decentralized identity infrastructures, over the past two decades.
enabling users to have more control
over their digital credentials.

One key impact of this shift will be

the accelerated adoption of passkey
technologies, which eliminate the need
for outdated, fraud-prone methods like

84 85
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

FROM FASTER consumers. that detect and mitigate scam calls,

providing customers with peace of
services and delivering high value ser-
vices to customers above and beyond

TO SMARTER:
While there are several AI-driven mind and safeguarding their commu- basic connectivity.
solutions already making waves within nications. By employing AI to actively
the telco sector, it now needs to be de- identify fraudulent activities, telco com- #4: BANDWIDTH IS OUT, AI IS IN

THE TELCO
ployed in a way that meets consumer panies are able to enhance security
needs and provides meaningful value. and trust for end users. The path forward for the telco industry
This falls to the AI tech stack. Deployed is clear: not only must service provid-

INDUSTRY’S AI
by cloud giants like AWS, Microsoft, Second is the creation of virtual AI ers shift their focus from bandwidth to
and Google, the AI tech stack utilizes assistants. These advanced agents can value, but they must embrace AI as a
massive data centers that are being fed handle customer inquiries, schedule catalyst for sustainable growth. Only

REVOLUTION
with massive power and connectivity appointments, and provide personal- then can telco companies unlock new
able to run the latest large language ized support, revolutionizing the small revenue streams, foster customer loyal-
models (LLM) and technology. This business’s ability to communicate with ty, and drive innovation.
technology platform is now available to their customers.
Written by Dag Peak the telecom sector as the foundation As telco’s north star evolves from
Chief Technology and Product Officer, to build applications with the power Lastly, AI can provide what we call bandwidth expansion to value creation
Alianza to enhance customer experiences, “deep insights” for regular telephone and customer experience, harnessing
streamline operations, and introduce calls. More than just call transcriptions AI’s transformative potential is critical
new services that were previously or generated to-do lists, deep insights to providing innovative solutions, rede-
Today’s telecommunications industry skyrocketed to $3.6 trillion, according their daily lives or the financial results unimaginable, including the hyper-cus- can provide a 360-degree view of voice fining offerings, and delivering services
is standing on the edge of a significant to Omdia Research. Without substan- of their business. As a result, service tomization of product offerings at-scale. communications, empowering small that resonate with modern consumers.
transformation. For decades, service tial revenue growth, the industry is providers need to expand the aperture business with information to make We’re currently standing at the cusp of
providers have concentrated their grappling with the reality that these in- of their focus, transitioning away from #3: PRACTICAL USE CASES HIT MAIN STREET more strategic decisions within their this AI revolution, but fortunately, the
efforts on expanding bandwidth, as- vestments are not providing the antic- simply offering faster connections to business. future of telecommunications looks
suming that faster and more extensive ipated financial rewards. Even further, providing services that genuinely en- There are three ways, in particular, promising — rooted in innovation,
connectivity was the key to success. this overemphasis on connectivity has hance the overall customer experience. that service providers are beginning to While just the beginning, these re- driven by technology, and focused
However, it’s becoming increasingly stifled innovation in other areas. Especially since the industry’s invest- capitalize on the AI tech stack to create al-world applications demonstrate our on delivering unparalleled value to
clear that this singular focus is no lon- ment in 5G and fiber has not yielded solutions that deliver tangible value. ability to leverage AI and build on top customers.
ger sustainable. The telco sector needs #1: SAME GAME BUT DIFFERENT expected returns, it’s time for telco to First is the development of systems of the AI tech stack, transforming telco
a pivot from this bandwidth-centric redefine its core offering from provi-
approach to one that prioritizes driving Beyond telco, other industries are ex- sioning bandwidth to value creation.
innovation and delivering tangible periencing this same existential crisis,
value to customers. This shift is not just specifically gaming. Over the past few #2: ONE TOOL TO RULE THEM ALL BY DAG PEAK
important—it’s necessary for survival decades, gaming executives and devel- Dag Peak is the Chief Technology and Product Officer
in the rapidly evolving competitive opers have been driven to consistently In navigating to this new north star, at Alianza, the world’s only cloud-orchestrated,
landscape. set new standards in graphical realism. telcos should reassess their strategies, AI-powered communications platform dedicated
However, we’re now beginning to see diversify investments, and look beyond to service providers. With a career rooted at the
This myopic focus on bandwidth is players prioritize unique experiences network infrastructure to prioritize crossroads of complex technology, innovative
both unnecessary – for most consum- and game play over technical specifi- value. Enter Artificial Intelligence product development, and forward-thinking industry
ers, the current levels of connectivity cations. (AI) — perhaps the most profound trends, Dag brings a unique perspective and vision
exceed their needs – and financially tech advancement that any of us has to his role.
unwise as shareholder returns in the This shift offers a valuable lesson for ever experienced in our lifetimes. AI
telecom sector have been unsatisfacto- telco. Just as gamers find joy in creative is a transformative force that has the
ry, and in some cases, negative. In fact, gameplay and social interaction, telco potential to unlock unprecedented op-
over the past decade, telco revenue customers are also looking for some- portunities for value creation across the
has stagnated, declining to $1.98 trillion thing beyond just a faster connection. telco ecosystem and empower service
in 2014, from $1.99 trillion in 2022. They’re looking for meaningful inter- providers to deliver innovative services
And, during that same period, CapEx actions and services that improve that actually resonate with today’s

86 87
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

#5: IMPLEMENTING MORE COMPLEX

FIVE BUSINESS COMMUNICATION

and new requirements set by cyberse- to reassess existing communication in-
curity regulations such as FISMA and frastructure that usually consists of dis- COMPLIANCE REQUIREMENTS
CISA mean that companies are now tributed legacy systems. These systems

TRENDS FOR 2025

facing a whole raft of imposing chal- can be modernized by highly scalable Companies must also deal with new
lenges. As a consequence, they need cloud messaging solutions to enable technologies in a conscious, respon-
to respond to these growing threats by more efficient communication process- sible manner. This applies both to
re-evaluating their risk and IT security es. With the help of a SaaS provider the way they are used and the strate-
management regimes and taking steps specializing in messaging, communica- gic investment they represent. They
Written by Martin Hager to strengthen them where necessary. tion channels such as SMS, email and consequently need to meet the most
President & CEO, Retarus This is especially true for business-crit- fax can be easily connected and adapt- demanding security and data protec-
ical communication – to provide the ed to changing conditions such as tion requirements. Solutions suitable
best protection possible, companies fluctuating transmission volumes, up- for digital corporate communication
At Retarus, we have identified five digital fax and automated capture of business communication, companies vitally require secure, transparent dated data protection regulations and provide high levels of transparency and
key trends set to shape digital busi- documents by which unstructured data would be well advised to consider the solutions that comply with all relevant new technologies. Modernizing legacy control while also fulfilling stringent
ness communication over the coming is captured and transformed into struc- specific use case for their business. regulations and guidelines. infrastructures also provides significant compliance requirements like HIPAA
year and beyond. Not only do these tured data. In these ways, companies Retarus deliberately only includes AI cost savings compared to self-operat- (the Health Insurance Portability and
developments reflect technological are able to reduce costs significantly in its products where it generates real #4: HYBRID CLOUD ENVIRONMENTS LEAD TO ed messaging infrastructures, which Accountability Act), all backed up by
advances, but also corporate social and ensure they remain competitive added value for customers. RISING COMPLEXITY are resource intensive and have high clearly defined service level agree-
responsibility and accountability. over the long term. running costs. Ideally, cloud messaging ments (SLAs). This enables companies
#3: MINIMIZING SECURITY RISKS Hybrid IT landscapes are common in products are billed based on usage. A to meet ever-increasing governance
#1: AUTOMATION OF BUSINESS-CRITICAL #2: TARGETED USE OF ARTIFICIAL many companies, but to fully har- broad interface portfolio such as REST and data protection requirements and
COMMUNICATION INTELLIGENCE Cyberattacks are on the rise and ness the innovation potential of new APIs, SOAP, XML, SMTP adapters and creates a sound, trustworthy basis for
becoming increasingly sophisticated, technologies, seamless integration much more offer the flexibility needed employing new technologies.
Companies are under substantial Artificial intelligence (AI) is one of the with AI playing a major role in both across these diverse IT environments is to integrate hybrid environments.
pressure from rising costs in the driving forces behind digital transfor- enhancing security measures and en- essential. This presents an opportunity
current challenging economic climate. mation. At the same time, considering abling malicious actors. In fact, 93% of
More than ever, businesses now need the dizzying array of AI-based solutions organizations experienced two or more
to ensure that their processes are on the market, it is not always clear identity-related breaches in the past
designed to be efficient. At the same whether AI provides any additional year, underscoring the urgent need for
BY MARTIN HAGER
time, essential resources are lacking benefit in every case. When opting improved defenses. The growing pro-
due to the general shortage of qualified for an AI-supportedsolution for their fessionalism of cybercriminal activity Martin Hager founded Retarus in 1992 and is its
staff.This is especially critical as the U.S. President and CEO. Under his leadership, the com-
labor force participation rate remains at pany has seen healthy, steady growth, transforming
a decreased 62.7%, still below pre-pan- from a Bavarian start-up to a global leader in B2B
demic levels. Automating work pro- cloud communications solutions. In his youth, Martin
cesses has thus become vital for suc- tinkered with electronic circuits, even making a living
cess. Innovative messaging solutions at a young age by developing software. He gained
support companies in achieving digital experience through working for various IT companies
transformation and automation in ways in development, support, consulting, sales, and
that meet their specific needs – laying product management before founding Retarus.
the foundations for leaner, more effi-
cient processes. Messaging solutions
play a crucial role in facilitating the
various integration scenarios along the
supply chain – the beating heart of any
business. In addition, an ongoing trend
towards cloud technologies and APIs
means that modern communication
solutions can be employed with greater
efficiency. This includes migrating to

88 89
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

GENERATIVE labor-saving device. the possibility to process vast amounts

of data, is making the Gen AI a really
those that publish fake news, AI pro-
vides actionable insights. Training AI

AI FOR THE
It doesn’t believe like the Electric Monk useful tool. to prioritize credible sources and avoid
in all the things the modern world misinformation ensures that businesses
expects us to believe in, but it can pro- In our telecommunications sector, operate with clean, trustworthy data.

SCEPTICS:
cess enormous amounts of data and while Gen AI won’t lay cables or build This is critical for making informed de-
produce human-usable summaries. physical networks, it excels at data cisions in today’s information-saturated
Desktop research and similar tasks are analysis and pattern recognition. This environment.

ADVICE FOR
ripe for replacement. Gen AI will do pattern-matching ability – one of the
them better. human brain’s core functions – is #4: ANALYZING NETWORK PERFORMANCE
what makes AI truly remarkable. I was

2025 FROM
I’ve already seen the backlash. Two personally surprised by how quickly Network analysis and root cause
accountants left because we auto- humanity achieved this. Pattern match- diagnostics are other areas where AI
mated too much of their work. I’d tell ing is fundamental to both human and exceeds human capabilities. Traditional

A VETERAN
them what I’d tell the Luddites of the animal brains – it’s how we distinguish systems often rely on predefined rules,
19th century: “Get a life and a job that threats from opportunities, and how we which can miss nuances. Genera-
requires human creativity and think- learn and adapt. tive AI tools, on the other hand, learn

UNBELIEVER
ing.” The world, and for that matter, from every interaction and improve
any business, needs that and needs it #2: TRANSFORMING DATA FORMATS over time. By analyzing patterns and
badly. relationships, AI can identify anomalies
One of the most transformative uses of and suggest actionable solutions faster
From a practical perspective, genera- AI in our business is handling customer and more accurately than humans.
Written by Neven Dilkov tive AI excels in areas like network per- inquiries that arrive in unstructured This continuous learning makes it an
Founder and CEO, Neterra formance analysis, root cause analysis, formats. Each client has their own invaluable tool for maintaining and
marketing analytics, and transforming format—often subject to arbitrary optimizing network performance.
information from one human-readable changes—which makes quick respons-
format into another. These are domains es challenging. AI tools like ChatGPT The limitations: creativity and goal
I have been accused of not trusting sophisticated and better funded ($80 Gently’s Holistic Detective Agency by where AI is not just faster than humans have revolutionized this process. With setting
people. It happens when I say, “Show billion?!) than most. Douglas Adams. The book is brilliantly but also significantly more accurate. minimal training, these tools can take
me”, more often than usual. My de- funny, and no, I thought, there’s no way Let’s delve into these applications. unstructured data and reformat it into But let’s be clear about AI’s limitations.
sire to understand what is happening We humans haven’t had many original we could come up with anything better our system’s structured requirements,
around me has made me inquisitive, ideas. Apart from the wheel on an axle, than that. #1: THE GREATEST ACHIEVEMENTS dramatically increasing productivity. While AI excels in many areas, there
and bad experiences have made me most innovations are inspired or direct- are tasks it’s ill-suited for. Creativi-
sceptical about what many consider ly copied from nature. Human science So, when IT specialists claimed to pro- Gen AI’s greatest achievement is its Previously, our colleagues wasted valu- ty—true, original creativity—remains a
common sense. doesn’t fully understand the brain; if it duce something that mimicked human ability to understand and generate able time manually formatting, reading, domain where humans have the edge.
did, we’d know how to treat autism. brains, I laughed. I mean, give me a human forms of expression, including and copying information between for- AI-generated poems may win contests,
So naturally, I have been in the slow break! speech, music, and images. Before, we mats. Now, AI performs these repetitive but they’re no match for Shakespeare.
adopter’s guild. The first iPhone I We can’t even describe what intelli- struggled to translate human instruc- tasks efficiently, allowing our team to Success in creative tasks often reflects
bought was the iPhone 6. I did not gence is. For curious readers, I recom- I changed my mind about Gen AI now. tions into computer language and focus on strategic work. This shift is the weakness of human competitors
use Facebook until after a year of it mend reading the dialogue between My curiosity was piqued when I saw an back again. Now, AI bridges this gap revolutionary, enabling us to respond to rather than AI’s brilliance.
becoming a global WOW. And I did not Richard Noonan and Valentin Pillman AI-generated fictional dialogue about effortlessly. This also explains why one inquiries quickly and consistently.
believe in this AI thing when it came in a bar, featured in the philosophical our company’s services. Around the of the first implementations for Gen AI Similarly, goal setting is beyond AI’s
out. I was fairly confident that it was yet and highly entertaining 1972 science same time, I heard AI-generated music was to write computer code. The com- #3: MARKETING ANALYTICS: SMARTER reach. Ask it, “What should I do now?”
another marketing gimmick, just like fiction novel Roadside Picnic by the that sounded good. Within the same puter language is actually quite simple INSIGHTS and it will provide methods to decide,
machine learning was a euphemism Strugatski brothers (read the book, not week, two literary contests in Bulgaria compared to human language. Once but not the decision itself. Critical
for “good quality statistical analysis” the movie). were won by AI-generated poems. you solve the more complex problem of AI also shines in desktop marketing thinking is another domain where
and now AI was going to be the better understanding human speech, translat- analysis. By tracking global trends, AI is competent but insufficient. As
“machine learning”. And for those who believe that imag- So, I did my research. Here’s what ing it into computer programs becomes analyzing vast amounts of data, and Edward de Bono famously said, critical
ination is a key intelligence trait, I I’ve learned, fellow sceptics: Gen AI, elementary. This ability combined with distinguishing reliable sources from thinking is excellent but not enough;
I took everything as a joke, albeit quite recommend the Electric Monk in Dirk as Douglas Adams would say, is a

90 91
The Fast Mode Trends and Outlook 2025

solving problems also requires creative AI integrated into a cable-laying ma- a powerful tool when used wisely. Its
thinking. chine could analyze data to determine strengths lie in processing and summa-
the optimal cable routes. Similarly, it rizing data, recognizing patterns, and
The future: a balanced perspective can support construction by providing automating repetitive tasks. However,
data-driven insights. it’s essential to approach AI with clean
Generative AI’s ability to recognize data, clear instructions, and an under-
patterns is one of its most powerful The real breakthrough lies in Gen AI’s standing of its limitations. Don’t rely on
features. This skill, akin to the innate ability to communicate in human lan- it for tasks requiring genuine creativity
abilities of even the simplest animal guage. This achievement, more than its or human judgment.
brains, has profound implications. AI capacity to summarize information, has
can analyze global data and provide made AI a truly revolutionary tool. By For sceptics like me, the key is to
human-readable summaries. It can bridging the gap between human and recognize AI for what it is: a labor-sav-

DATA CENTERS
transform human instructions into machine, AI enables users to interact ing device. It’s not a replacement for
computer-readable code—a task once with technology more intuitively than human ingenuity but a complement to
reserved for highly paid programmers. ever before. it. Used correctly, it can free us to focus
on what we do best—thinking creative-
In telecommunications, AI won’t lay Embrace AI with caution ly and solving problems in ways no
cables or build physical networks, but machine can replicate.
it can enhance processes. For example, Generative AI is not a panacea, but it’s
1. How the Evolution of the AI Factory Will Ignite GenAI Transformation������������������������������������.94

2. The Next Generation of Data Centers: Power, Performance, Precision����������������������������������� .97

BY NEVEN DILKOV
3. How 400G ZR/ZR+ Transceivers Are Revitalizing Decades-Old Tech To Support

Neven Dilkov is a European entrepreneur with nearly

AI-Driven Connectivity�������������������������������������������������������������������������������������������������������������������������������������������������������������.99
30 years of experience in IT and telecoms. Founder
and CEO of Neterra, he also established NetIX, Sofia
4. How Generative AI Is Reshaping Data Centers in 2025������������������������������������������������������������������������ .101
Data Center, Neterra.Cloud, NetFleet, and Neterra.T V.
Neven chairs the ecta Board and is a member of the 5. GenAI and the Infrastructure Renaissance������������������������������������������������������������������������������������������������������� .104
Forbes Business Council.
6. Transforming Networks to Enable GenAI in 2025����������������������������������������������������������������������������������������.107

7. AI is Driving a Connectivity Revolution������������������������������������������������������������������������������������������������������������������ .110

92 93
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

centers are cooled. Hyperscalers and computational intensity of AI workloads rural data centers. Building in dense-
enterprise network operators are facing while minimizing their environmental ly populated urban areas is not only
a stark reality—legacy cooling systems footprint. It’s a win-win for both innova- cost-prohibitive but also riddled with
built for air-cooled environments can tion and efficiency. logistical challenges like high real es-
no longer keep pace with the heat gen- tate prices, elevated operational costs,
erated by today’s high-density GPUs #3: RETHINKING DATA CENTER PLACEMENT and energy shortages. Rural areas,
and next-generation compute stacks. on the other hand, offer an appealing
Keeping servers cool is not just an op- The location of next-generation data combination of affordability and access
erational necessity but also a financial centers has traditionally been dictated to renewable energy sources such as
and environmental imperative. by their proximity to population hubs stranded power and wind energy.
and client bases. However, advance-
Enter direct-to-chip liquid cooling, a ments in network designs and AI Speed-to-power—the ability of data
technology primed for center stage in technologies are shifting priorities. For centers to quickly access the energy
2025 to revolutionize how data centers AI workloads, the critical challenge lies needed for operation— is equally crit-
manage heat. While liquid cooling not in external latency but in internal ical. Urban areas often face delays in
has existed on a smaller scale in the data center latency—ensuring ultrafast securing power due to limited infra-

HOW THE EVOLUTION OF THE AI FACTORY WILL

past, the data center industry is now communication between GPUs and structure, strict regulations, and grid
preparing for large-scale liquid cooling compute nodes to handle complex issues. These slowdowns don’t match
adoption. Unlike air-cooled systems, tasks efficiently. the fast pace of AI innovation, which

IGNITE GENAI TRANSFORMATION

which often falter under extreme ther- demands quick scalability.
mal conditions, liquid cooling transfers This shift opens new opportunities
heat directly away from critical compo- for data centers to be placed in less Rural areas often have the advantage of
nents, delivering optimal performance densely populated areas, where untapped energy sources, like wind or
with greater efficiency. lower operational costs and access to stranded power, and data centers can
Written by Wes Cummins renewable energy offer clear advan- accelerate the setup process by work-
CEO, Applied Digital However, advancing to liquid cooling tages. However, applications like online ing with local energy providers. This
in data centers comes with signifi- gaming, video streaming, or banking— faster access to energy boosts inno-
cant challenges. Retrofitting existing which require quick responses—may vation and keeps businesses compet-
air-cooled facilities for liquid cooling is still benefit from being closer to users itive in a rapidly changing technology
Artificial intelligence (AI) is not just #1: MAXIMIZING GENAI PERFORMANCE cycles. These cycles enable models to often cost-prohibitive, requiring either to reduce latency. landscape.
knocking on the door of the future— learn and adapt to live data streams, entirely new data center designs built
it’s kicking it wide open. Powered by At the heart of GenAI’s power lies an resulting in more reliable systems over for liquid cooling from the ground The key to balancing these demands is Locating AI Factories in rural com-
neural networks, machine learning unyielding need for performance and time. This push for real-time comput- up or entirely new purpose-built “AI integrating next-gen technologies with munities doesn’t just make economic
(ML) algorithms, and large language speed. Training AI and ML models— ing is driving advancements in data Factories.” Yet the benefits outweigh strategically located infrastructure. By sense—it also drives significant local
models (LLMs), generative AI (GenAI) especially the expansive ones driving centers, which are now built to handle the hurdles. Large-scale liquid-cooled designing AI Factories that prioritize development. These facilities bring
is revolutionizing industries with inno- state-of-the-art applications like hu- faster speeds and larger workloads. systems use significantly less power ultra-fast internal communication and high-tech infrastructure, job opportu-
vative, rapid, and highly personalized manoid robots—demands vast com- These centers are adopting energy-ef- than air-based systems, lowering ener- leveraging advanced networks, we nities, and investments into historically
solutions. However, the true power of putational muscle. These sophisticated ficient technologies that cut waste and gy consumption and achieving better can accommodate the needs of both overlooked or underserved areas. From
GenAI lies in the infrastructure that models require billions, if not trillions, of boost performance to meet the energy power usage effectiveness (PUE) AI workloads and real-time consumer boosting local economies to expanding
powers it, presenting both opportu- parameters to process and predict out- demands of AI tasks powered by GPUs. ratings. More notably, technologies like applications. This approach opens up the talent pool for AI and IT profes-
nities and challenges. Purpose-built comes effectively. Whether it’s Tesla’s Modern data centers are helping GenAI closed-loop liquid cooling also con- opportunities to build data centers in sionals, rural AI Factories are poised to
“AI Factories,” designed to meet the humanoid advancements or deploying reach its full potential across industries serve water, addressing global sustain- rural areas, bringing new advantages, create sustained value far beyond their
unique demands of AI, will be pivotal AI to enhance enterprise decision-mak- by combining powerful computing with ability concerns. such as lower costs and renewable en- operational mandates.
in transforming the GenAI promise into ing, the ability to compute in real-time sustainable practices. ergy, which leads us to the next trend
tangible outcomes. Here is a look at and at scale is imperative. This transformation will be pivotal to to watch. The decentralized nature of this shift
four significant developments driving #2: THE YEAR LIQUID COOLING GOES LARGE the long-term success of GenAI infra- also ensures long-term benefits for
AI transformation and how they will One key element reshaping how structure. By solving the heat problem, #4: BUILDING UP RURAL COMMUNITIES enterprise networks. By strategically
help reshape networks, data centers, enterprises manage networks is the The explosion of AI workloads is also direct-to-chip liquid cooling ensures situating data centers in less populated
and communities. emergence of continuous inference driving a seismic shift in how data data centers can keep pace with the AI is opening the door to a new era of areas, organizations can tap into re-

94 95
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

newable resources, enhance grid resil-

ience, and build scalable infrastructures
maintain the systems that support
these technologies.
beyond, building such systems feels
well within reach. THE NEXT
GENERATION OF
to accommodate future AI workloads.
Improving computing power, expand- AI has opened the door to reshaping
The next chapter of AI ing advanced cooling systems, and how organizations innovate, connect,

DATA CENTERS:
transformation locating AI-capable data centers in and grow. With the right investments
rural areas are just some of the steps and strategies, GenAI will become the
The rise of GenAI is both exciting and we’re taking. These innovations show driving force for transformation across

POWER,
challenging. For businesses, tele- how next-gen AI Factories can tackle industries, including telecom and enter-
com providers, and researchers, it’s a today’s challenges while opening the prise networks. The door is open—and
chance to reshape how we connect, doors to AI’s limitless possibilities. To those who harness the full power of

PERFORMANCE,
engage with customers, and boost truly unlock the power of AI, we need AI today will shape the innovations of
efficiency. But making this transforma- infrastructure that blends top-level tomorrow.
tion a reality requires a fresh approach performance with sustainable prac-

PRECISION
to how we design, build, power, and tices—and as we move into 2025 and

BY WES CUMMINS

Wes Cummins is Chairman and CEO of Applied Written by Scott Willis

Digital, a company he co-founded. With more than Chief Executive Officer, DartPoints
20 years of experience as a technology investor,
Wes has a robust background in capital markets,
having held positions at investment banks and
institutional asset management firms. His expertise The data center landscape is chang- high-performance computing (HPC). high-density workloads. Innovations
and leadership in technology investment have been ing faster than ever. Just a few years The average rack power density has such as direct-to-chip cooling, liquid
instrumental in guiding Applied Digital’s strategy and ago, data centers were primarily used significantly increased, with some immersion, and hybrid systems are
operations. for supporting storage and basic hyperscale data centers operating at transforming how facilities manage
connectivity. Today, they’re the back- densities upwards of 50 kW per rack. heat.
bone of business operations, enabling For many organizations, the question
high-performance workloads, hybrid IT, is no longer about how to add capacity Not only do these methods tackle heat
and seamless cloud integration. As we but how to do so efficiently. more effectively, but they also lower
see rising power requirements, increas- energy consumption, blending perfor-
ing densities, and innovative cooling High-density solutions, designed for mance with sustainability. For busi-
techniques taking center stage, the maximum efficiency and scalability, nesses striving to excel operationally
data center industry is transforming. are stepping up to help manage these while meeting environmental targets,
increasing demands. They focus on these new cooling techniques are
This evolution is driven by the need for smarter power management strategies becoming essential tools.
smarter, more sustainable, and scalable and sustainable practices, keeping data
solutions. Here’s a closer look at the centers resilient and future-proof while #3: HYBRID IT
challenges facing modern data centers handling increasingly complex work-
— and five predictions for the future of loads. Hybrid IT—a blend of private, public,
this essential infrastructure in 2025. and on-premises cloud solutions—is
#2: COOLING INNOVATION: BEYOND becoming the new normal for enter-
#1: RISING POWER DEMANDS AND DENSITIES TRADITIONAL prises. Organizations require seamless
integration between these environ-
Power consumption in data centers With rising densities comes the need ments, demanding high-speed, low-la-
has never been higher, driven by the for new cooling methods. Tradition- tency connectivity across providers
growing adoption of artificial intelli- al air-cooling systems are no longer and regions.
gence (AI), machine learning (ML), and sufficient for the heat generated by

96 97
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

To achieve this, flexible colocation

services and private cloud connectivity
data centers. This focus on intercon-
nectivity ensures businesses can oper-
infrastructure.
HOW 400G ZR/ZR+
TRANSCEIVERS ARE
play pivotal roles. This infrastructure ate at peak efficiency in an increasingly The path ahead
empowers enterprises to deliver fast, complex digital ecosystem.
secure, and consistent experiences The next generation of data centers
#5: SECONDARY MARKETS: THE NEW

REVITALIZING DECADES-OLD
while keeping a firm grip on essential is all about smarter, more efficient
operations in an interconnected eco- FRONTIER FOR GROWTH designs built to meet the challenges
system. of modern workloads. By focusing on

TECH TO SUPPORT AI-DRIVEN

While major metro areas continue to high-density computing, sustainable
#4: CONNECTIVITY: BEYOND THE EDGE dominate the data center conversation, operations, and advanced connectiv-
secondary markets are experiencing ity, the industry is well on its way to

CONNECTIVITY
As the industry evolves, the focus has explosive growth. These regions offer shaping the next era of data center
moved from simply placing infrastruc- cost advantages, improved sustain- innovation.
ture closer to end users to enabling ability opportunities, and proximity
strong provider connections. High- to emerging customer bases, making As businesses and industries evolve,
speed, private connectivity is now es- them ideal for data center expansion. data centers will remain essential to
sential for supporting hyperscale, retail, their success. Whether it’s enabling Written by Frank Yang
and hybrid IT environments. Investing in these secondary markets cutting-edge hybrid IT or scaling Senior Product Line Manager, Legrand
ensures businesses in these regions operations in secondary markets, data
Leading infrastructure providers can access the same high-perfor- centers are setting the stage for a more
are bridging these gaps by offering mance infrastructure as their metro sustainable and innovative 2025 and
high-capacity, low-latency connections counterparts. This expansion supports beyond. Advances in chips and GPUs have scalability — requirements that the requires installing additional under-
that enable seamless integration be- local economies while addressing the helped to spur the Artificial Intelligence fragmented, hardware-heavy legacy ground fiber infrastructure between
tween networks, cloud platforms, and growing need for distributed digital (AI) revolution and corresponding approach cannot efficiently deliver. facilities. Until recently, IPoDWDM was
surge in demand for processing power The answer lies in a decades-old limited to 10G or 25G connections.
currently impacting data centers. An optical networking technique that has, However, the introduction of 400G ZR/
equally important piece of infrastruc- until recently, been limited in capacity. ZR+ and 100G ZR digital coherent
BY SCOTT WILLIS ture that will help to truly unlock their Next-generation transceivers coupled (DCO) transceivers brings renewed in-
Scott Willis is DartPoints’ Chief Executive Officer and potential going forward are data center with multichannel passive multiplex- terest in the technology by unleashing
serves on the company’s Board of Directors. With interconnects (DCIs). ers/demultiplexers have brought new higher bandwidth capacity.
over 30 years in the communications and technology life to Internet Protocol (IP) over Dense
industries, Scott has a strong history working with DCIs are the backbone of modern dis- Wavelength Division Multiplexing IPoDWDM, powered by 400G ZR/ZR+
Private Equity Sponsors. Scott holds a B.B.A. in tributed data infrastructures. Reliable (IPoDWDM) as a leading-edge DCI and 100G ZR transceivers, supports
Finance from the University of Oklahoma and an and high capacity DCIs are essential solution. By unifying the IP and optical superior DCI by directly integrating IP
MBA from Wake Forest University Babcock Graduate for the performance and operations layers, IPoDWDM provides the simplic- routers and switches into the Optical
School of Management. of data centers. The sales of DWDM ity, scalability and efficiency needed to Transport Network (OTN). This stream-
optical transceivers to DCI market is handle modern AI-driven workloads. lined approach eliminates the need
expected to grow from $1.7 billion to for intermediary transport equipment,
$4.9 billion in 2029 at a CAGR of 23%, #1: NEW TRANSCEIVERS REVIVE IPODWDM reducing capital expenditures (CapEx),
according to LightCounting. Similarly, power consumption and network com-
data center clusters require greater Legacy DCI optical networks rely on plexity. Instead of relying on multiple
connectivity to support AI. a substantial amount of equipment, layers of networking hardware, IPoD-
including IP routers and switches WDM allows data centers to transmit
But the surge in data-intensive applica- and grey and color fiber optics. This signals over existing fiber infrastructure
tions like AI has exposed the limitations approach has significant drawbacks: with only two optical fiber strands,
of traditional DCI architectures. These transponders and muxponders add extending up to 80 kilometers. This
workloads generate unprecedent- extra layers of hardware, increasing innovation results in up to 75% cost
ed data volumes that demand high costs, power consumption and physical savings in CapEx alone compared to
bandwidth, low latency and seamless footprint. Scaling up these systems traditional optical networking.

98 99
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

The advantages of IPoDWDM go work traffic is east-west. This places to 3.2 Tbps over just two fiber strands
beyond reducing equipment. One of its an immense burden on DCI infrastruc- using 8-channel passive DWDM multi-
most significant benefits is its ability ture, as GPUs need to coordinate in plexers/demultiplexers as an example,
to maximize fiber utilization. Deploy- real time to process massive amounts optimizing network efficiency.
ing new fiber infrastructure is both of data. Traditional optical networks
time-consuming and cost-intensive, struggle to accommodate these traffic #4: THE FUTURE OF NETWORKING
requiring trenching, conduit installa- demands efficiently, requiring costly,
tion and lengthy permitting processes time-consuming and power-hungry As AI adoption accelerates, data
that can delay network expansion by scaling. IPoDWDM delivers ultra-high centers must evolve to support the
months. IPoDWDM allows data center bandwidth and scalability while main- exponential growth in workloads.
operators to multiply their bandwidth taining an efficient, low-power foot- Legacy networking models are not
capacity instantly over existing fiber print. equipped to handle the scale, speed
without the need for disruptive con- and cost-efficiency requirements of
struction work, leading to additional #3: SCALABLE AI GROWTH AI-driven infrastructure. IPoDWDM,
cost savings. powered by advanced transceiver
The ability to scale efficiently is crucial technologies, is emerging as the key to
#2: MANAGING TRAFFIC BETWEEN DATA for AI growth. Hyperscalers and en- unlocking high-performance, scalable
CENTERS

HOW GENERATIVE AI IS
terprises deploying AI workloads need DCI solutions that meet AI networking
a networking architecture that can demands. The integration of transceiv-
AI is dependent on “east-west traffic,” expand with rising compute demands ers with passive multiplexing creates

RESHAPING DATA CENTERS IN 2025

where massive amounts of data move without needing significant infra- an end-to-end solution that simplifies
laterally between GPUs and compute structure overhauls. With IPoDWDM, deployment, minimizes failure points
nodes within and across data centers. organizations can scale to 100G and/or and enhances overall network perfor-
Traditional client-server applications 400G speed using modular upgrades, mance.
rely more on “north-south traffic,” ensuring that their DCI keeps pace
where data travels from one destination with AI’s growth. IPoDWDM eliminates unnecessary
to the next. AI training and inferencing, network layers, reduces power and
on the other hand, require high-band- Additionally, IPoDWDM leverages footprint costs, and leverages existing Written by Wes Swenson
width, low-latency interconnects to passive optical multiplexing, which fiber infrastructure. This cutting-edge CEO, Novva Data Centers
facilitate seamless communication further enhances scalability by aggre- solution is positioned to support the
between distributed processing units. gating multiple wavelengths on a single next wave of AI innovation while lower-
fiber pair. This allows data centers to ing costs, offering greater flexibility and
In AI data centers, up to 90% of net- achieve bandwidth capacities of up increasing efficiency. Equipped with enhanced security fea- This example of independent robots necessary transformation to keep pace
tures and armed with predefined tasks, is just one example of how GenAI is with burgeoning AI demand, here are
the fleet of AI-powered robotic dogs is transforming data centers in fascinat- four key trends we expect to dominate
hard at work in the data center. They ing ways. As data centers become the the data center industry conversation
move methodically from room to room, backbone of powering the AI revolu- in 2025.

BY FRANK YANG scanning server racks, monitoring tion, the need for efficiency, reliability,
temperatures, and ensuring machines and security has never been greater. #1: MICROGRIDS FOR ENERGY RESILIENCE
Frank Yang is a senior product line manager at Leg-
run smoothly — all without human AI-driven automation in data centers
rand. He manages a team responsible for the product
intervention. This isn’t science fiction heeds the call by transforming in- As energy grids face increasing strain
lines of optical transceivers, passives and more. He
or the plot of a Philip K. Dick novel; frastructure management, reducing and data centers strive for greater
worked at CommScope and Dell prior. Frank has ex-
it’s the present-day reality of modern downtime, automating routine main- energy resilience and sustainability,
perience in product management, strategic planning,
data centers. As AI, automation, and tenance and optimizing energy use in microgrids are emerging as a lifeline
product marketing and development engineering in
robotics reshape infrastructure man- ways that were unimaginable just a few for data center operators who want to
the technology industry.
agement, the question isn’t just what’s years ago. create their own resilient, sustainable
next — it’s how quickly the future is power-sourcing strategies.
becoming the present. As AI-driven applications evolve and
data center infrastructure undergoes These localized power systems en-

100 101
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

hance resilience by integrating renew- from robotics’ promise of enhanced efficiency. How? By analyzing real-time demands of an AI-driven world. While so are the opportunities. As we look to meet the needs of an increasingly
able energy sources like solar and wind automation. According to a previous data on workloads, cooling needs, and the challenges—such as rising power ahead to 2025 and beyond, the conver- digital world.
with battery storage and backup gener- prediction from Gartner, by the end of power usage, GenAI can dynamically consumption and staying on track with gence of AI and data center innovation
ators, allowing data centers to operate this year, half of cloud data centers are adjust operations to minimize waste sustainability goals —are significant, will only grow stronger and evolve
independently from the traditional expected to leverage advanced robots and improve efficiency. It can also
power grid when necessary. Microgrids with AI and machine learning capabili- predict and recommend the best times
also allow operators to maintain uptime ties, which will reportedly increase data to utilize renewable energy sources
BY WES SWENSON
during grid disruptions while reducing center operating efficiency by 30%. or, when possible, shift workloads to
reliance on fossil fuels, aligning with regions with lower carbon intensity. As Wes Swenson is a founder, investor, and the CEO of
overall industry sustainability goals. The implications of this shift extend AI-driven automation advances, data Novva Data Centers, as well as the former CEO and
far beyond operational efficiency. With centers will move closer to a self-opti- founder of Utah-based C7 Data Centers (acquired
Microgrids are primed for greater a growing gap between the number mizing model where meeting sustain- by DataBank in 2017). With 35 years of experience
adoption in 2025 and will benefit from of servers and storage volumes at ability targets becomes an intelligent, working in the tech industry — including 15 years in
increased AI integration in the future. data centers and capable workers to continuously evolving process. So, the data center industry — Wes has a deep bench of
With smarter monitoring capabilities, oversee them, deploying robots will be although AI places great strain on data tech knowledge including semi-conductors, desktop
AI-enabled microgrids will be able to critical in coming years. With AI han- centers, it can also be a catalyst for in- software, network software and hardware, web
produce real-time energy forecasting, dling routine monitoring and trouble- creased sustainability in the long-term. security, consumer products, and data centers.
demand-response automation, and shooting, human technicians can focus
predictive maintenance. But can they on higher-value tasks such as strategic #4: CONTINUED INTEREST IN SECONDARY
keep up with the enormous amount of planning, security, and innovation. This MARKETS
power needed to power AI data cen- evolution also raises important ques-
ters? The results have yet to be seen, tions: How will automation reshape the This final trend hits close to home.
but some research indicates that yes, workforce? What new skills will be in Driven by the rise of AI and continued
according to testing, microgrids can be demand? And as data centers continue cloud and hyperscale market growth,
viable for powering AI data centers. to expand, how will AI-driven systems demand for new data centers is soar-
keep pace with growing complexity? ing. As a result, colocation capacity in
#2: AI-ENHANCED MONITORING FOR the U.S. has doubled since 2020 and
OPERATIONAL EFFICIENCY #3: SUSTAINABILITY GOALS BACK ON TRACK behind leading markets like Northern
Virginia, New York City and Silicon Val-
As hinted at in this article’s opener, There’s no doubt the meteoric rise of ley, secondary markets are growing.
AI-powered monitoring tools — and generative AI technologies has put
even robotic assistants or autonomous great demand on data centers and In the year ahead, we’ll continue to see
drones — are transforming data center unfortunately has also thrown a wrench construction continue at a rapid pace,
management and automation. These into many operators’ sustainability with more secondary markets increas-
tools can conduct routine inspections, goals. Luckily, the industry’s not going ing their data center capacity. GenAI’s
monitor hardware conditions, detect to give up so easily on their promises of role here will be to analyze historical
anomalies in real-time, and even going green. Sustainability has become trends and future demand forecasts to
predict maintenance needs, minimiz- a defining priority for data centers and assist data center operators in planning
ing downtime and improving overall amid growing public concern over to meet capacity demand.
efficiency. As AI adoption accelerates, environmental impacts, operators are
data center operators are leveraging investing in renewable energy, liquid As AI continues to drive the evolution
machine learning to create smarter, cooling, and carbon footprint reduction of data centers, its impact is being felt
more adaptive facilities. strategies to meet stricter regulations across energy management, auto-
and corporate sustainability goals. mation, sustainability, and expansion
It’s estimated that 90% of businesses into new markets. From AI-powered
globally plan to implement some form GenAI has a transformative role to play microgrids enhancing energy resil-
of robotic automation by 2030, an in this regard, as harnessing its abilities ience to autonomous monitoring tools
increase from 20% in 2021. Data cen- can help operators optimize everything improving operational efficiency, the
ters in particular are primed to benefit from energy consumption to hardware industry is rapidly adapting to meet the

102 103
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

GENAI AND THE INFRASTRUCTURE RENAISSANCE better power efficiency and scalability
than x86 architectures. Additionally,
the open-source RISC-V architecture
infrastructure (HCI) is set to play a
pivotal role in supporting this develop-
ment.
data centers and public clouds but also
provides the flexibility to move AI work-
loads as needed. For generative AI,
offers flexibility for developing custom which might involve training models in
Written by Lori MacVittie AI-focused processors. HCI simplifies scalability by consolidat- one cloud environment and deploying
Distinguished Engineer & Chief Evangelist, F5 ing compute, storage, and networking them in another, this seamless connec-
#2: DPUS ESTABLISH A FOOTHOLD resources into a single, software-de- tivity is crucial.
fined infrastructure. In an HCI envi-
One of the specialized accelerators ronment, as the need for compute #5: THE IMPACT ON THE NETWORK IS JUST
As organizations prepare to leverage to storage and HPC, now dominates ly large language models and deep gaining significant traction are Data or storage increases, new nodes are BEGINNING
generative AI, many are realizing that AI factories, meeting generative AI’s learning applications, require signifi- Processing Units (DPUs). This traction added, and the infrastructure automati-
their networks are unprepared for the network demands. Apple continues to cant computational power and parallel is growing because the rise of gener- cally balances workloads. This scalabil- While significant changes are coming
demands it brings. Just as with previ- enhance chipsets for AI performance in processing capabilities. This demand is ative AI is promoting heterogeneous ity is ideal for AI factories, which need to data center and enterprise architec-
ous technology shifts - from internet consumer products, while Intel’s data driving the adoption of specialized ac- computing, where multiple types of to handle fluctuating and sometimes tures in the near term, generative AI is
access to mobile phones to digital center dominance faces new compe- celerators like GPUs, TPUs, and custom processors (CPUs, GPUs, TPUs, and unpredictable demand as they process far from reaching its final form. As the
services - the initial excitement quickly tition from RISC architectures proving AI chips (such as those from NVIDIA, FPGAs) are used together to optimize vast amounts of data and continually technology matures, the refinement of
shifts to the need for robust techni- their value in critical markets. Google, and other vendors), which are performance. train and deploy new models. Small Language Models (SLMs) is be-
cal capabilities to meet performance often more efficient than traditional x86 coming a priority. SLMs are emerging
challenges. The network is in transition, and the CPUs for AI tasks. DPUs take over many network and #4: MULTICLOUD NETWORKING NO LONGER as a more efficient alternative to Large
changes wrought by this infrastructure storage tasks that would otherwise OPTIONAL Language Models (LLMs), offering
Thus, we stand upon the precipice of renaissance will be felt for decades. We ARM-based processors, already consume valuable CPU resources. In the benefits of faster response times,
an innovation cycle driven by the need present four distinct changes we see gaining traction in mobile and cloud high-performance environments, DPUs While most AI-enabling technologies, reduced computational demands, and
for performance and scale of the infra- being driven into data center networks environments, are emerging as com- can handle tasks like data encryption, such as GPUs, TPUs, and DPUs, are improved accuracy when trained on
structure that supports generative AI. by generative AI. petitive alternatives in the data center, compression, packet routing, and indeed data center-centric, the reality specialized datasets.
especially with companies like Amazon storage management, freeing up CPUs for most organizations is that they
This infrastructure renaissance will be #1: GENAI WILL BREAK X86 DATA CENTER (AWS Graviton processors) and Apple for other compute-heavy tasks. This operate within hybrid, multicloud envi- Additionally, the need for AI to evolve
as dramatic as generative AI’s en- DOMINANCE leading the charge. ARM’s efficiency is particularly valuable in AI and ML ronments. Enterprises today often rely beyond transformer-based architec-
trance via ChatGPT into our collective and customizability make it attrac- workflows, where CPUs need to focus on a mix of on-premises data centers, tures is inevitable. While transformers
consciousness. InfiniBand, once limited Generative AI workloads, particular- tive for AI-driven workloads, enabling on model training and inference, not public clouds, and even edge comput- have been revolutionary in advancing
data movement or application delivery ing locations to deploy and run their AI natural language processing, their lim-
and security. This trend allows data applications and models. Generative itations in scalability, energy efficiency,
centers to adopt a more flexible, work- AI further compounds this complexity, and interpretability are becoming more
load-specific approach to hardware as it requires massive amounts of data, apparent as AI applications proliferate.
and networking as workloads can be processing power, and seamless ac- Emerging architectures and method-
distributed based on matching work- cess to diverse data sources to function ologies, such as retrieval-augmented
loads to hardware resources to opti- effectively. For organizations, this hy- generation (RAG), recurrent neural
mize for performance and utilization. brid setup introduces a range of chal- networks (RNNs) revisited for better
lenges, especially around connecting sequence handling, and one-bit or
#3: THE RETURN OF HYPERCONVERGED applications, models, and data sources sparse models, are being explored to
INFRASTRUCTURE across the internet and multiple cloud address these constraints.
providers.
The separation of duties in these This ongoing evolution will not only
heterogenous environments does not To overcome these challenges, shape the models themselves but
obviate the need for low-latency, high organizations will increasingly adopt will also impact the infrastructure
throughput networking, especially multicloud networking solutions. that supports them. Data centers and
in the construction of AI Factories - Multicloud networking provides a uni- cloud environments will need to adapt
large-scale, integrated environments fied, consistent networking layer that to support these next-generation AI
optimized for building, training, and spans on-premises, cloud, and edge architectures, optimizing for flexibility,
deploying AI models - to support gen- environments. This approach not only lower latency, and energy efficiency. As
erative AI. The rise of hyperconverged simplifies the connectivity between a result, we can expect to see shifts in

104 105
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

hardware requirements, with a rise in

specialized chips and optimized stor-
AI models. In this dynamic landscape,
staying adaptable will be crucial for
full potential of generative AI’s next
chapter. TRANSFORMING
NETWORKS TO
age solutions tailored for these refined organizations aiming to harness the

BY LORI MACVITTIE

Lori MacVittie is a Distinguished Engineer & Chief

Evangelist in F5’s Office of the CTO with an emphasis
on emerging architectures and technologies includ-
ENABLE GENAI
ing cloud and edge computing, digital transforma-
tion, and application delivery. MacVittie has over 25
years of industry experience spanning application
IN 2025
development, IT architecture, and network and
systems’ operation. She holds an M.S. in Computer Written by Mattias Fridström
Science and particularly strong opinions about Vice President and Chief Evangelist,
cheese and gaming systems. Arelion

Telecommunications has been abuzz

since GenAI’s mainstream debut. This
powerful technology dominates head-
lines for its potential to streamline op-
erations and automate time-consuming
tasks. Most conversations focus on the
data centers enabling GenAI appli-
cations, with these sites supporting
critical data processing and replication
for AI model training. There are already
enormous amounts of traffic within
data centers and over short distanc-
es between adjacent data centers to
support AI use cases. However, internet
carriers have yet to see high amounts
of long-distance AI traffic, and we’re
unsure if AI will create significant traffic
loads for carriers any time soon. Many
industry trends point in this direction,
but we have yet to see real-world proof
currently.

But, if that traffic does hit carriers soon-

er than we expect, how will GenAI’s
operational needs spur continued the
transformation of the backbone net-
works that provide underlying connec-
tivity to these data centers? While Ge-
nAI is relatively novel, its connectivity
requirements are similar to cloud com-
puting and other bandwidth-consump-
tive technologies, albeit heightened.

106 107
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

In 2025, internet carriers will continue limitation. These additional innova- make critical decisions concerning geospatial monitoring systems and a single point of failure in the event evolve to enable these emerging ap-
to maximize scalability, performance, tions will include further integration of customer traffic, their data must be better anticipate natural disasters. of a natural disaster. These physical plications. Although rapid innovation is
security and other familiar qualities extended frequency bands like L-band trustworthy and accurate. More careful evaluation of potential infrastructure improvements will prove exciting, network transformation occurs
through strategic network investments, technology. Typically, the C-band networking sites will also prove crucial. crucial in enhancing network reliability in small steps first rather than in short-
helping enterprises effectively leverage cannot improve due to Shannon’s Network automation is crucial for serv- Traditionally, operators viewed forests to ensure GenAI applications’ uninter- term leaps and bounds. Operators will
GenAI applications. Limit, and other bands are too ineffi- ing GenAI’s networking demands, and as relatively safe and remote locations rupted operations. maximize familiar connectivity qualities
cient to use with modern networking operators are working steadily to make for placing fiber optic cables. Howev- in their network transformation strat-
#1: AT THE SPEED OF LIGHT: ACCELERATED technologies. But the often-underuti- their networks smarter, faster and more er, with the rise of forest fires, service Leaps and bounds: network egies, but the stakes are higher now
OPTICAL INNOVATION lized L-band can give internet carriers efficient. But we still need some level of providers may become more strategic transformation’s outlook as AI/ML applications demand more
a short-term push within the limits of human intervention in the loop, so it’s in their fiber locations or bury cables capacity and reliability to facilitate
As it did during the rise of cloud hyper- Shannon. But what optical network unlikely we’ll see completely self-heal- deeper in vulnerable regions. While transformation is a constant real-time functionality. No matter how
scalers, optical networking innovation transformations can we expect in the ing networks in the near future. Howev- force in telecommunications, we’ll networks evolve, these qualities will re-
will again prove vital for supporting distant future? er, operators may draw better conclu- Amid these developments, some things never predict its exact trajectory. When main vital for enabling GenAI applica-
GenAI use cases by enhancing data sions from their massive data sets in stay the same. Operators will still con- looking ahead, we can leverage the tions across diverse industries in 2025
center interconnection, capacity and Hollow-core fiber is promising for inter- 2025, even if it just means improved tinuously improve network diversity by lessons of past transformation trends and beyond.
other capabilities. GenAI applications net carriers as GenAI accelerates. This insights on equipment replacements building additional routes that eliminate to understand how the industry will
consume massive amounts of high-ca- cable type allows operators to transmit and related network operations issues
pacity bandwidth to enable extensive light through an air-filled core rath- that can hinder GenAI’s real-time func-
BY MATTIAS FRIDSTRÖM
computations and real-time opera- er than glass fibers, enabling higher tionality.
tions. Optical innovation persists as speeds, lower loss and other enhance- Mattias Fridström is the Vice President and Chief
the primary method for transforming ments for supporting the connectivity #3: FLOODED SERVERS AND FLOODED Evangelist for Arelion. Since joining Telia in 1996, he
networks to meet these needs in the requirements of emerging applications. INFRASTRUCTURE: WEATHERPROOFING has worked in several senior roles within Telia Carrier
GenAI era. Hollow-core fiber currently faces pro- NETWORKS (now Arelion), most recently as CTO. He has been
duction scalability and cost challenges Arelion’s Chief Evangelist since July 2016.
Some service providers have been which inhibit widespread implemen- Our sector has made recent progress
early adopters of open optical network- tation. It’s a fascinating technology in fostering collaboration between net-
ing models, and we will continue to see but it comes with a hefty price tag if work operators, infrastructure owners,
further integration. Open optical net- used to “replace” all large networks. regulatory authorities and industry
working often presents additional com- Still, the industry may overcome those traceback groups in the fight against
plexities initially, but many operators obstacles eventually and begin larger DDoS attacks. While these efforts seek
are learning that its long-term benefits deployments by 2030. to combat malicious traffic flooding
outweigh its immediate challenges. network servers, what about the
This networking model allows internet #2: TRUST ISSUES AND THE DATA natural disasters flooding our physical
carriers to integrate 400G coherent op- ECOSYSTEM infrastructure? These events signifi-
tical pluggables, empowering them and cantly disrupted global communica-
their customers with higher capacities, While operators are transforming their tions in 2024, particularly in Florida and
enhanced scalability and improved networks to support GenAI and AI other regions along the East Coast of
power efficiency. Aside from increased applications, these technologies may the United States. Unfortunately, many
integration of 400G pluggables, we’ll also enhance network operations. service providers are only beginning
also likely see enhanced utilization of AI holds great promise for service to discuss weatherproofing network
800G ZR pluggables by more internet providers when leveraged through infrastructure, and even fewer are
carriers as bandwidth demands contin- predictive analytics, while GenAI can actively implementing weatherproofing
uously accelerate. assist in streamlining customer service strategies.
processes through chatbots. These
However, our industry has reached analytics can help carriers predict As climate change escalates, these
Shannon’s Limit, which represents the traffic patterns, identify anomalies and strategies will enter the spotlight in
maximum capacity that can fit on a improve automation. However, creating 2025. While we still have much work
fiber optic cable. As a result, we can accurate, high-quality data ecosystems to do in this area, these enhancements
expect more ingenuity in addition to is essential to enable AI-driven network may include further collaboration
pluggables to overcome this physical operations. If we want machines to with authorities to leverage advanced

108 109
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

AI IS DRIVING A CONNECTIVITY REVOLUTION revolutionized optical line systems.

In 2025, look for a new generation
of compact modular OLS to debut,
to deliver the right capacity at the
right time and at the right cost. While
ICPs may obtain dark fiber and light
than 95% of all international traffic. In
2025, we expect to see more holistic
and synergistic collaboration between
marked by three key enhancements: 1) their own private networks, doing so is subsea and terrestrial networks to meet
Written by Rob Shore increased transmission spectrum per fi- not the right answer in all cases. With the growing demands of data centers
Senior Vice President, Global Marketing, Infinera ber that includes Super C, Super L, and MOFNs, ICPs can drive the network and AI. To minimize CLS space and
combined C+L (and Super C + Super architecture and vendor selection, but power, subsea networks will continue
L); 2) high density, high-port-count they can also leverage the resources to glass-through the CLS and termi-
(>60 ports) direct-attach ROADMs to and operational expertise of wholesale nate/originate optical waves at data
2025 promises to be an exciting year While still nascent, AI-related traffic de- ters scale up and out – with 100,000 support hyperscale IPoDWDM with CSPs. In 2025, look for a large increase centers when possible. But when traffic
for telecommunications networks. Ar- mand is accelerating, helping to sustain power-hungry GPUs now enabling the coherent pluggables deployed directly in the number and variety of MOFN needs to be delivered to multiple data
tificial intelligence will be at the center an already hot data center interconnect largest computing complex. Intra-data into routers; and 3) enhanced software networks in Asia, the Middle East, centers or waves must be terminated
of most networks. AI is not only driving (DCI) market with an annual growth center connectivity is moving forward automation that continuously optimizes Africa, and around the globe, ranging at the CLS, look for 800G coherent
growth in connectivity demands – it is rate of greater than 50%. In addition with 800 Gb/s pluggables followed transmission settings on a per-wave- from multi-fiber bundles to dedicated pluggables utilizing Super C spectrum
also creating new connectivity patterns to the need for more capacity per fiber closely by 1.6 Tb/s solutions seeking to length basis to maximize OSNR and fiber pairs and multi-tenant spectrum to provide cost-effective, low-power
as data center operators are forced and improved cost per bit per kilometer increase speed and reduce power and capacity-reach in multi-vendor environ- sharing. terrestrial backhaul for subsea traffic. In
to get increasingly creative in the of transmission, lowering power con- costs. Look for 2025 to be a year filled ments. addition, expect compact modular op-
geographical placement of new data sumption is becoming a driving force with 1.6 Tb/s experimentation, including #5: SURF AND TURF – SUBSEA AND tical platforms to increasingly support
centers due to limitations on power for connectivity solutions. Look for DSP-based retimed optics, linear-drive #4: CONNECTIVITY COLLABORATION – THE TERRESTRIAL NETWORKS COLLABORATE TO a superset of terrestrial and subsea
availability. In addition to AI creating commercial adoption of three related pluggable optics (LPO), linear receive RAPID GROWTH OF MANAGED OPTICAL FIBER POWER DATA CENTER GROWTH functionality, including advanced soft-
challenges and opportunities with con- solutions to significantly aid operators optics (LRO), and co-packaged optics NETWORKS ware automation and intelligent power
nectivity solutions, network operators in 2025: 1) 800G coherent pluggables (CPO), for both serial and parallel fiber With the number of data centers management to simplify planning, de-
need to be looking for ways to leverage with 3-nm DSPs that reach 1,500 km or applications. Managed optical fiber networks expected to double in coming years ployment and operations across metro,
new AI processing capacity to increase more, 2) 1.2+ Tb/s embedded optical (MOFNs) involve close collaboration and AI driving increased power long-haul, and subsea.
efficiency and further automate their engines that transmit 800 Gb/s “ev- #3: LINING UP FOR SCALE – A NEW among internet content providers requirements, modular, distributed
operations. Below are the five key erywhere,” and 3) Super C and Super L GENERATION OF OPTICAL LINE SYSTEMS HITS (ICPs), communication service pro- data centers are becoming increas- All in all, 2025 will be an interesting
industry trends that we expect to see expanded spectrum that transmits up THE MARKET viders (CSP), and optical transport ingly common. In addition, many data year as AI deployments ramp up,
emerge in 2025. to 27% more bits per second per fiber. vendors. With the number of data centers that are at the heart of driving bringing new challenges and opportu-
With mix-and-match capabilities, centers expected to double in the next increased traffic demands will not be nities for network operators and driving
#1: AI PROVIDES A BOOST TO NETWORK Inside the data center, connectivity support for 300-mm and 600-mm few years and growing demand for colocated in the immediate vicinity of increased innovation in networking
OPERATIONS AND MAINTENANCE speeds and power efficiency also footprints, and pay-as-you-grow modu- high-capacity optical transport among subsea cable landing stations (CLS) designs and optical connectivity solu-
remain top of mind as GPU clus- larity, compact modular platforms have them, ICPs are looking for new ways – but subsea networks carry more tions.
Throughout 2024, AI was at the center
of trade shows and customer conversa-
BY ROB SHORE
tions, with the entire ecosystem work-
ing to understand its full potential and Rob Shore is the Senior Vice President of Infinera’s
possible impacts to optical networking Global Marketing organization. He is responsible for
traffic both inside and outside the data Infinera’s Global Marketing efforts, including solu-
center. In 2025, we expect vendors to tions, content, campaigns, digital and social media,
leverage their growing AI knowledge trade shows, advertising, and sales enablement. He
and skillsets to deliver solutions that has held roles in engineering, system testing, market
help network operators enhance the management, account management, technical sales,
efficiency and capabilities of their oper- and marketing.
ations and maintenance teams.

#2: GROWING AI DEMAND DRIVES A

CONNECTIVITY REVOLUTION – OUTSIDE AND
INSIDE THE DATA CENTER

110 111
 eBook: The GenAI leap: how networks are reinventing themselves

WHAT DOES AI BRING TO THE TABLE FOR

TELECOMS SECURITY?
Written by Dmitry Kurbatov
CTO & Co-Founder, SecurityGen

SECURITY
1. What Does AI Bring to the Table for Telecoms Security?����������������������������������������������������������������������.113

2. How AI Is Rewriting Best Practices for Network Infrastructure and Security������������������ .116

3. AI Is Reinventing Network Security����������������������������������������������������������������������������������������������������������������������������.118

4. The Evolving Ransomware Threat: Challenges and Innovations���������������������������������������������������.121

5. Driving Business Value with Responsible GenAI in 2025��������������������������������������������������������������������.124

6. The Evolution of Connectivity: Emerging Trends in Telecommunications��������������������������.127

7. GenAI in Telecom 2025 – Machine Learned Friend or AI Generated Enemy������������������� .129

In 2024, multiple incidents demonstrat- the Snowflake Customer Data Breach. ic protocols and technologies, targeting
ed the alarming capability of hackers vulnerabilities in areas once deemed
to infiltrate telecom networks, remain These events reveal a significant shift obscure and secure.
undetected for extended periods, exfil- in the threat landscape: attackers are
trate sensitive data, and spread laterally no longer limited to exploiting exposed Methods such as leveraging trust-
to other systems. Notable examples IT services, web applications, or phish- ed relationships between roaming
include Salt Typhoon’s Espionage ing campaigns. They now possess a partners via IPX/GRX and exploiting
Campaign, the AT&T Data Breach, and deep understanding of telecom-specif- unmonitored segments within telecom

112 113
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

infrastructures enable attackers to 5G standards. However, in real-world increasingly unsustainable. Focusing on these overlooked areas and being prepared for swift incident long-standing issues, streamline
bypass defenses and execute sophisti- deployments, the actual level of protec- will be critical for building resilience response and remediation—are not processes, and achieve measurable
cated data theft operations. tion often appears overestimated. In 2025, telecom companies must and safeguarding critical infrastructure new. However, applying these princi- improvements in their security posture
pivot toward dynamic, intelligent, in an increasingly hostile cyber envi- ples to telecom networks has always in a cost-efficient manner.
Looking ahead to 2025, the challenges Recent incidents of hackers exploiting and adaptive defense strategies. This ronment. been a formidable challenge due to the
for telecom infrastructures are poised vulnerabilities have made it clear that transformation will require moving immense scale of infrastructure and the Telecom networks are critical to the
to intensify. The likelihood of uncov- telecom operators must reevaluate beyond static designs to incorporate Summary high associated costs. functioning of societies, governments,
ering previously undetected compro- their risk exposure across all domains. real-time threat detection, automated and countries. The time has come to
mises will increase, and attackers will The MITRE FiGHT database identi- response mechanisms, and measurable The fundamental principles of cyber- What is new is the ability to overcome take bold, decisive steps to secure
continue to refine their tactics. With the fies approximately 150 techniques for security improvements—all enabled by security—knowing your attack surface, these challenges at scale, thanks these vital systems, ensuring they re-
root problems now more apparent, the compromising 5G networks, with even advanced AI technologies. implementing effective protections, to advancements in AI. By leverag- main resilient, adaptable, and prepared
telecom industry must act decisively— more targeting legacy technologies performing continuous monitoring, ing AI, telecom operators can tackle to meet the challenges of the future.
analyzing vulnerabilities, addressing from 2G to LTE. While these findings By adopting these new measures,
root causes, and devising a compre- are concerning, they also serve as a telecom operators can demonstrate to
hensive plan for security improvements valuable roadmap for directing security stakeholders that they have learned
BY DMITRY KURBATOV
to safeguard critical networks. investments. from past missteps and are committed
to fortifying their networks against fu- Dmitry Kurbatov, the CTO and Co-Founder, is a
#1: COMPANIES WILL BE TASKED WITH AI-powered assessments are be- ture threats. This shift is not just an op- world-renowned telecom security expert with over 15
UNCOVERING HIDDEN THREATS coming essential tools for identifying erational necessity but also a strategic years of experience in researching 5G, LTE network,
vulnerabilities, prioritizing threats, step for restoring trust and resilience in SDN/NFV, and IoT.
Telecom networks pose a unique and devising efficient and impactful the sector. A telecom security veteran with a sharp focus on a
challenge to cybersecurity due to their mitigation strategies. By leveraging research-based approach - Dmitry, along with his
vast scale and complexity. The exten- AI, telecom operators can transform a #4: ADDRESSING OVERLOOKED AREAS IN R&D team, have led several innovative global telecom
sive attack surface, countless assets, daunting challenge into a systematic, SECURITY MONITORING security deployments. He has spearheaded the de-
and intricate communication channels data-driven approach to enhancing velopment of an automated security analysis system
create an environment where achieving their security posture. In today’s security paradigm, resilience and telecom security research, which includes iden-
comprehensive control has historically is key—not just preventing breaches, tifying vulnerabilities in network equipment, errors in
been nearly impossible. #3: ADOPTING DYNAMIC APPROACHES FOR but minimizing their impact by detect- data transfer, network design, protection of signalling
NETWORK FORTIFICATION ing and disrupting the attack kill chain protocols (SS7, Diameter, GTP), and IoT security.
However, the emergence of advanced and ensuring critical operations can
AI technologies is transforming this Moody’s has ranked the telecom sector continue despite an attack. Dmitry graduated from Moscow Technological Uni-
landscape. AI can now process mas- among the highest-risk industries, versity (MIREA) with a degree in information security
sive volumes of network data, detect citing “weaker defense practices” Historically, telecom companies have of telecommunications systems.
patterns that would otherwise go compared to other critical sectors. This struggled with maintaining sufficient
unnoticed, and analyze anomalies. By designation not only highlights sys- security oversight, leaving many net-
leveraging AI-driven insights, telecom temic vulnerabilities but also imposes work segments unmonitored. Hackers
operators can perform retrospective significant business challenges, making have exploited these gaps, making
analyses, uncover hidden threats, it essential for the industry to chart a it imperative to address them as a
and answer the critical question: Are new path forward. priority.
hackers currently inside the network, or
have they been there undetected? This Prior to 2024, many telecom defens- In 2025, AI-powered intrusion detec-
step is essential to fortifying networks es relied heavily on systems deemed tion and security monitoring systems
against future attacks. “secure by design” and standardized will offer the ability to analyze network
by 3GPP. However, this approach often communications in real time, learn nor-
#2: REASSESSMENT OF RISK EXPOSURE resulted in static security measures mal patterns of behavior, and instantly
ACROSS ALL TELECOM DOMAINS ill-equipped to counter the dynamic flag anomalies. These capabilities
and evolving tactics of adversaries. As enable swift incident response and
The protection of communications has cyber threats grew more sophisticat- significantly reduce the window of op-
always been a fundamental pillar of ed, these traditional methods proved portunity for attackers to cause harm.

114 115
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

HOW AI IS REWRITING BEST PRACTICES FOR impossible for recipients to recognize

malicious intent, leading to a sharp rise
pure prevention. Once primarily used
by large enterprises, incident response
effective as the experts who train and
deploy them. The shift toward a more

NETWORK INFRASTRUCTURE AND SECURITY

in successful breaches and signaling plans and third-party retainers will robust AI-powered toolset will also
the need for advanced detection meth- become priorities for businesses of all call for cybersecurity teams to use the
ods to counteract these highly targeted sizes. time gained to adapt swiftly to evolving
attacks. threats. Developing new expertise in
#4: AI WILL TAKE ON THE BASIC INCIDENT threat intelligence interpretation, com-
Written by Erik Nordquist Most concerning is AI’s ability to lower RESPONSE ROLES, BUT SECURITY PROS WILL plex analysis, and strategic threat re-
Global Managed Security Product Director, GTT the barrier of entry into the world of SUFFER IF THEY DON’T EVOLVE QUICKLY sponse will be critical. If the demand for
cybercrime. You don’t have to be a skilled, affordable experts far exceeds
skilled coder, for example. Anyone As it has been laid out thus far, AI the supply as cyberattacks inevitably
with malicious intent and AI-assisted will become an integral part of both become more complex, it will leave a
The rapid evolution of AI is set to dra- programming tools can now produce technological innovation and security significant gap in oversight and expose
matically transform network architec- powerful malware quickly at scale strategies. Its influence will redefine businesses to serious risks.
tures, cybersecurity strategies, and the how businesses operate and protect
nature of digital threats in 2025. The To defend against all AI-powered their networks. Conclusion
demand for more agile, responsive, threats, organizations must match
and secure networks will intensify as the speed and agility of attackers or With an increased focus on rapid de- The continued rise of AI will deep-
AI processing shifts to a decentralized risk falling behind in this new era of tection and agile response, data-driven ly influence the future of network
model. Concurrently, cybercriminals are accessible cyberweaponry. Educating and AI-infused automation will serve architectures and cybersecurity in the
operationalizing AI for their own pur- employees to recognize these attacks as the primary frontline defense. Such coming year. From meeting the needs
poses, lowering the barriers to launch- is essential. systems will act instantly and auton- of decentralized AI processing and
ing increasingly sophisticated attacks. omously, analyzing data patterns to defending against the growing threat
Organizations must adapt to evolving #3: NETWORK SECURITY BUDGETS WILL combat threats without human inter- of AI-powered cyberattacks, organi-
threats in this shifting landscape and PIVOT FROM PREVENTION TO INCIDENT vention. They will also allow networks zations must evolve their strategies to
invest in advanced detection and RESPONSE to be more adaptable as automated stay ahead. While AI-driven automation
response strategies to protect against defenses learn from every new incident will play a key role in detecting and
these new risks. As AI continues to While endpoint and firewall protection to evolve in real-time. This will allow responding to threats faster than ever,
shape both technological innovation will remain critical, the network attack organizations to be faster, smarter, and human expertise will remain essential
and cybersecurity practices, the role to AI performance and responsiveness. ease of use. According to the Microsoft surface is expanding due to growing more precise in responding to emerg- in optimizing these systems and in-
of human expertise in managing these Digital Defense Report 2024, cyber- reliance on hybrid workforces, IoT, ing threats. terpreting complex security scenarios.
systems will remain crucial. As networks’ capacity to handle AI criminals are developing and using and cloud services. Threats are also To stay secure in this rapidly changing
workloads grows, we’ll see AI appli- emerging attack techniques powered growing more sophisticated with the Despite AI’s advances, the human landscape, businesses must invest
#1: DECENTRALIZED AI PROCESSING WILL cations assist network operators in by AI. In 2025, one can expect to see help of AI. That is why, in 2025, security element will remain irreplaceable, as in both AI technologies and skilled
REDEFINE NETWORK ARCHITECTURES fine-tuning the performance output of more threat actors using AI to create budgets will increasingly focus on rapid skilled professionals will be critical to professionals who can harness their full
its architecture. With AI-driven moni- sophisticated malware. detection and agile response to contain unlocking its full potential. To begin potential.
Advances in chip technology have toring and adaptive optimization, net- breaches, reflecting a shift away from with, AI-enabled defenses are only as
given user devices, such as smart- works will become smarter and more Narrowing the focus of AI-powered
phones and laptops, the increasing efficient, ready to support the high cyber threats, expect a surge in “deep
BY ERIK NORDQUIST
ability to process AI workloads locally. demands of modern AI applications. phishing” attacks this year—AI-pow-
Less reliant on data centers to process ered campaigns that take spear phish- Erik Nordquist is responsible for the Managed Secu-
AI workloads, the new decentralized AI #2: AI WILL LOWER THE BAR FOR ing to a new level of sophistication. rity portfolio, including Managed Firewall, Managed
processing paradigm will require net- CYBERCRIMINALS TO LAUNCH EFFECTIVE Unlike its traditional counterpart, deep DDOS and, Managed SASE, GTT Secure Connect. After
works to support significant device-to- SOCIAL ENGINEERING AND CYBERATTACK phishing leverages synthetic media to supporting customers in field operations, network
cloud traffic. As AI applications operate CAMPAIGNS create hyper-realistic impersonations of integrity, DDoS mitigation, and security management
more and more in real-time, compa- trusted individuals, such as executives for more than a decade, Erik brought his customer-fo-
nies will be pushed to build networks Many organizations have and will or colleagues, engaging targets with cused approach to product management, where he
optimized for machine-to-machine continue to leverage AI to optimize personalized audio or video messages delivers unsurpassed customer support for best-in-
communication, where latency and sin- their business performance, but there that build a false sense of security. This class network and security services.
gle-hop environments become critical is a cautionary tale with AI’s scale and AI-driven approach will make it nearly

116 117
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

AI IS REINVENTING NETWORK SECURITY intervention. Generative AI can auton-

omously adjust network configurations,
routing, and bandwidth allocation in
#3: SELF-HEALING NETWORKS: A NEW ERA
OF RESILIENCE
that network infrastructure is built to
withstand evolving cyber threats - all
while maintaining optimal performance.
real-time, reducing downtime and Beyond predicting failures, AI-driven Moreover, AI can streamline network
Written by Erez Tadmor enhancing overall efficiency. These self-healing networks can take immedi- design processes, reducing the time
Field CTO, Tufin AI-driven optimizations allow networks ate action to address issues in real time and cost required to implement com-
to dynamically adapt to traffic fluctua- without human intervention. These net- plex infrastructures.
tions, ensuring seamless performance works leverage AI-powered algorithms
- even under varying loads. to reroute traffic, patch vulnerabilities, #5: DIGITAL TWINS: SIMULATING NETWORKS
and neutralize threats dynamically. BEFORE DEPLOYMENT
With the ability to quickly analyze mas-
sive volumes of data, AI can identify For example, when an AI system Generative AI is enabling the creation
inefficiencies and automatically fine- detects unusual network activity, it can of digital replicas of networks - oth-
tune network parameters. This level of isolate affected nodes, redirect traf- erwise known as digital twins. These
automation not only reduces opera- fic through secure paths, and deploy virtual duplicates allow for extensive
tional costs but also enhances the user security patches automatically. This testing, development, and simula-
experience by minimizing latency and rapid response capability significantly tion before implementing changes in
congestion. Furthermore, AI can con- reduces the impact of cyberattacks, real-world networks. By replicating
tinuously learn from network behavior, ensuring business continuity and data network behavior in a controlled envi-
making iterative improvements that integrity. ronment, organizations can evaluate
result in long-term stability, security performance, security risks, and optimi-
improvements, and efficiency gains. While the closest we have to this zation strategies without disrupting live
today are AI-powered assistants that operations.
#2: PREDICTIVE MAINTENANCE: PREVENTING can recognize issues and lead a user
ISSUES BEFORE THEY OCCUR to where a solution may be located, Digital twins provide a powerful tool
self-healing networks will eventually for refining network configurations,
One of AI’s most significant contribu- take advantage of newer AI technology validating security measures, and opti-
tions to network security is predictive and become even more autonomous, mizing resource allocation. They ensure
maintenance. AI-powered models can dramatically reducing an organization’s that any network modifications are
analyze patterns within network traffic, reliance on manual oversight and thoroughly tested before deployment,
hardware performance, and historical intervention. reducing risks and improving overall
data to predict potential failures or con- reliability.
gestion points before they happen. This #4: AI-CREATED ARCHITECTURES: DESIGNING
proactive approach ensures seamless NETWORKS FOR THE FUTURE As AI continues to evolve, digital twins
operation by addressing issues before will become even more sophisticated,
they escalate into outages or security AI is not just optimizing networks - it providing deeper insights into network
vulnerabilities. is also designing them. With AI-gen- behavior and potential vulnerabilities.
erated architectures, networks can
Modern network security measures uncovers them. Here are some areas where AI is sig- For instance, AI can detect anomalies be designed and architected based #6: DISTRIBUTED AI PROCESSING: BRINGING
need to adhere to today’s highly dy- nificantly changing the way networks in the data flow that indicate potential on business goals and optimization INTELLIGENCE TO THE EDGE
namic and scalable environments by Network security needs to be able to are managed, the way potential risks hardware degradation or cyber threats. targets. AI can assess vast datasets to
quickly adapting to change - and main- keep up with the pace of innovation are addressed and how scalability can By predicting these failures in advance, recommend the most effective topol- The future of network security also lies
taining the highest level of security as being used in these attacks. But it isn’t be improved: organizations can implement preven- ogies, configurations, and security in distributed AI processing, where
they do so. For example, think of adap- easy, especially when teams have few tive measures, reducing downtime and frameworks. AI-driven decision-making occurs at
tive malware that can alter its code, resources to apply to the problem. #1: AI-DRIVEN OPTIMIZATION: ENHANCING ensuring a resilient network infra- the edge, rather than solely on central-
execution patterns, or communication That’s where Artificial Intelligence (AI) NETWORK EFFICIENCY structure stays that way. Additionally, This AI-driven approach enables orga- ized cloud systems. By embedding AI
methods in response to the environ- can be most beneficial - as a tool that predictive analytics can help optimize nizations to construct networks that are models within edge devices, networks
ment it encounters during an attack. can level the network security playing AI is revolutionizing how networks are resource allocation, ensuring that inherently secure and scalable, adapt- can achieve real-time processing, re-
Because of this, it can dynamically (and field. With AI, network security can optimized, enabling them to function networks are always operating at peak ing seamlessly to emerging challenges. duce latency, and improve security.
quickly) exploit misconfigurations as it scale to meet these threats as needed. more efficiently with minimal human efficiency. AI-generated architectures ensure

118 119
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

This shift to decentralized AI process-

ing allows for faster anomaly detection,
protocols that standardize communi-
cation across diverse systems. Tools
#8: TOMORROW’S NETWORK SECURITY WILL
RELY ON AI THE EVOLVING RANSOMWARE THREAT:
CHALLENGES AND INNOVATIONS
quicker threat response, and more like Google’s AutoML and OpenAI’s
efficient data handling. Some modern Codex automate API creation, reducing As cyber threats continue to evolve, AI
devices are already leveraging AI to the complexity of integrating diverse will be a critical enabler of next-gen-
filter and analyze network traffic locally, network components. eration network security. In fact, from
reducing the burden on centralized optimizing configurations and predict-
infrastructure while enhancing overall By automating configurations, AI-driv- ing failures, to enabling self-healing Written by Jim McGann
security. Additionally, edge AI enables en protocols reduce human error, networks and designing architectures, VP of Strategic Partnerships, Index Engines
devices to operate autonomously in enhance security, and streamline AI is already fundamentally reshaping
remote or disconnected environments, operations. IBM’s Watson AIOps, for how networks operate - and how they
improving resilience in critical applica- example, predicts and resolves network defend against advanced, aggressive
tions. issues before they escalate, improv- threats. Ransomware has emerged as one and offense, the challenges posed by Ransomware attacks have become
ing resilience. As AI advances, these of the most persistent and evolving ransomware require businesses to commonplace and insidious, with
#7: AI-GENERATED APIS AND PROTOCOLS: intelligent APIs will further optimize As AI improves, its role in network threats in the cybersecurity landscape. rethink their strategies. The focus has attackers using increasingly advanced
ENHANCING SEAMLESS INTEROPERABILITY communication in increasingly complex security will only become more pivotal, This malicious software has grown shifted from prevention to recovery and methods to stay ahead of an organiza-
network environments, driving efficien- redefining the way we protect and in power, employing increasingly resilience, as cybersecurity adjusts to tion’s defenses. One popular method
AI is transforming network interopera- cy and security at scale. manage an organization’s environment. sophisticated tactics to compromise new realities. is intermittent encryption, a tactic that
bility by generating adaptive APIs and systems and disrupt operations. From corrupts the content of data while
deploying shadow encryption modes #1: THE RISE OF SOPHISTICATED leaving metadata intact. This allows
which are difficult to detect to the dual RANSOMWARE ransomware to bypass traditional
role of artificial intelligence in defense detection tools, which often rely on sur-
BY EREZ TADMOR

Erez Tadmor holds a two-decade career in the

information security field, with expertise in cloud and
network security, automation & orchestration, IAM,
fraud detection, and prevention. As Tufin’s Field CTO,
he bridges the gap between customers, marketing,
and product teams, educating stakeholders on
network security technologies, cybersecurity best
practices and solutions.

120 121
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

face-level checks like unusual changes more sophisticated ransomware vari- curity. businesses will gain confidence in their emphasis on recovery, businesses a recovery-first mindset and ensuring
in compression rates to identify com- ants, and deploy attacks on a broader ability to recover clean data, minimizing must adapt to stay ahead of the curve. operational continuity.
promised files. scale. AI enables them to adapt their First, AI-powered detection systems operational impact and downtime. By adopting proactive monitoring,
methods in real-time, outpacing some will take center stage in combating leveraging AI-powered tools and As ransomware continues to evolve, so
This approach can create a false defensive systems and exploiting ransomware. Unlike traditional tools Building a resilient future fostering a culture of collaboration, must our defenses. The key lies in em-
sense of security for organizations. weaknesses faster than ever. that rely on indicators such as metada- organizations can build resilience bracing innovation, prioritizing recovery
Files appear untouched to the system, ta, these systems analyze data content The evolving nature of ransomware against ransomware threats. Looking and recognizing that cybersecurity is
yet their content is quietly rendered We all know that the cybersecurity and behavior over time to identify demands a paradigm shift in how ahead to 2025, it’s clear that AI-pow- no longer just about prevention, it’s
unusable. And as we know, the ram- industry is facing a pressing challenge, anomalies. organizations approach cybersecurity. ered Machine Learning, unified cyber about building a resilient, adaptive
ifications can be severe, as recovery to leverage AI as a force for good while From the rise of shadow encryption to resilience efforts and continuous data future.
becomes exponentially more com- staying ahead of bad actors who are #4: COLLABORATIVE CYBER RESILIENCE the dual role of AI and the increasing integrity monitoring will be critical to
plex and time-consuming. Extended using it for financial and disruptive gain.
downtimes, increased financial losses, The only way to counter this threat is Cybersecurity will no longer be the ex-
BY JIM MCGANN
and long-lasting operational impacts by utilizing equally advanced AI tools clusive domain of IT or security teams.
are now common consequences. and technologies to defend against Instead, it will become a collaborative Jim McGann is VP of Strategic Partnerships at Index
Organizations must recognize that these evolving risks. effort involving storage teams, execu- Engines. Jim is a globally-experienced marketing
their existing detection methods may tive leadership and other stakeholders. and business development executive instrumental in
already be obsolete. Their emphasis #3: FROM PREVENTION TO RECOVERY This unified approach will require developing key relationships and brand development
must shift toward tools that monitor breaking down organizational silos and at Index Engines. Jim is experienced with both large
data content and behavior over time to The focus of cybersecurity strategies fostering a culture where cybersecurity established software firms and emerging startups
identify anomalies. Without proactive has shifted from prevention to recovery. is viewed as a shared responsibility. and is a frequent writer and speaker in the areas of
detection and monitoring, recovery With ransomware attacks becoming ransomware recovery, cyber resilience and unstruc-
from ransomware attacks will become more frequent and complex, it’s no By involving all stakeholders in design- tured data management.
an even greater challenge. longer a question of “if” but “when” an ing and implementing resilience plans,
organization will be targeted. This shift organizations can better mitigate risks
#2: AI IS A DOUBLE-EDGED SWORD necessitates a greater emphasis on and recover more effectively. Compa-
resilience, that is the ability to recover nies that embrace this “team sport”
Artificial intelligence is transforming the quickly and effectively from attacks. mentality will not only reduce the
cybersecurity battlefield, serving both Modern ransomware doesn’t just likelihood of attacks but also strength-
as a shield and a sword. On one side, encrypt files, it targets critical systems en their capacity to respond swiftly and
AI-powered tools are proving invalu- and data, making recovery a multi-step cost-effectively.
able for strengthening defenses. These and complex process. For businesses,
tools can predict threats, detect anom- this underscores the importance of #5: DATA INTEGRITY AS A STANDARD
alies, and automate responses to ma- having robust cyber resilience plans
licious activity, significantly enhancing that integrate proactive monitoring, A non-negotiable from this day forward
an organization’s ability to prevent and AI-driven tools and rapid recovery is continuous data integrity monitoring.
mitigate attacks. For example, AI-driven mechanisms. It is absolutely an essential compo-
detection systems can identify subtle nent of cybersecurity strategies. With
signs of data corruption and ransom- Organizations that prioritize recovery ransomware becoming harder to detect
ware activity that traditional methods as part of their cybersecurity strategy and recover from, businesses will adopt
miss. By continuously analyzing data can reduce downtime, financial impact tools that analyze how data changes
patterns and behaviors, these tools and reputational damage. Those that over time. These tools will provide an
enable organizations to address threats fail to prepare risk significant disrup- additional layer of security by focusing
and execute an intelligent recovery tions, data loss and potentially crippling on the actual content and behavior of
process to minimize the impact. ransom payouts. data, rather than surface-level indica-
tors. This shift will allow organizations
At the same time, cybercriminals are As we move into 2025, several trends to proactively identify corruption and
also adopting AI to identify system are set to redefine how organizations recover quickly. As data integrity mon-
vulnerabilities more efficiently, create approach ransomware and cyberse- itoring becomes a standard practice,

122 123
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

DRIVING #3: BALANCING INNOVATION WITH

CYBERSECURITY TO PREVENT
VULNERABILITIES
and manual tasks, many employees
fear displacement, which can lead to
trust with customers, regulators, and
investors while minimizing the risks

BUSINESS
increased turnover and decreased associated with AI adoption. Audit your
morale. AI systems regularly to ensure com-
Cybersecurity threats are expected pliance and mitigate potential biases.

VALUE WITH
to rise as AI becomes increasingly 38% of U.S. workers are worried that AI This will help businesses stay ahead
embedded in business operations. Ac- could eventually take their jobs. Among of ethical challenges and protect their
cording to the World Economic Forum, these, 51% report work-induced mental brand reputation.

RESPONSIBLE
this trend has already begun: in 2023, duress, and 46% are considering seek-
data breaches scaled globally, with a ing new employment. #6: LEVERAGING AGILE FRAMEWORKS TO
72% increase in data compromises NAVIGATE TECHNOLOGICAL COMPLEXITY

GENAI IN 2025
compared to the previous year. The key to mitigating these disruptions
lies in reskilling and upskilling the AI integration is a complex and multi-
The increased reliance on AI technol- existing workforce. Enterprises need faceted process, particularly for large
ogies exposes businesses to potential to invest in AI literacy programs and enterprises with legacy systems. As AI
Written by Justin Hutchens vulnerabilities, including data breaches, hands-on training to empower employ- tools become more sophisticated, busi-
Innovation Principal, Trace3 unauthorized access, and malicious at- ees to adapt to the evolving landscape nesses face the challenge of integrating
tacks. Organizations that fail to address and ensure the company’s long-term these technologies with their existing
these risks may face significant finan- success. Equipping employees with the infrastructure. This complexity can slow
In the rapidly evolving landscape of goals, causing poor return on invest- of AI implementation is data quality. cial losses and reputational damage. tools to navigate AI shifts fosters loyalty adoption and impede progress without
artificial intelligence (AI), businesses ment (ROI) and low stakeholder buy-in. Poor data management can derail and reduces turnover. a flexible, agile approach.
are racing to harness the power of AI initiatives, resulting in inaccurate To sidestep these challenges, com-
generative AI (GenAI) to transform As the initial frenzy passes and the wa- predictions, biases, and unreliable panies must prioritize cybersecurity #5: ENSURING ETHICAL AI USE TO NAVIGATE To overcome these hurdles, companies
their operations, maximize efficiencies, ters calm down, enterprises will need outcomes. In fact, Gartner predicts that as part of their AI integration strate- LEGAL AND REPUTATIONAL RISKS should adopt agile methodologies that
and unlock unprecedented growth. to better define their goals before im- by 2025, 80% of organizations trying gy. This includes aligning with data allow for iterative testing and rapid
Yet, with all of AI’s promise comes the plementing AI solutions. This involves to scale their digital business will fail protection regulations such as the As AI continues to evolve, ethical con- adaptation. By breaking down AI proj-
challenge of navigating its risks. As engaging cross-functional teams in due to inadequate data and analytics General Data Protection Regulation cerns surrounding its use are becom- ects into smaller, manageable phases,
networks across industries embrace AI, workshops that explore the “art of the governance. (GDPR) and the California Consumer ing increasingly prominent. Issues such businesses can address challenges
sidestepping pitfalls becomes crucial possible,” identifying AI use cases that Privacy Act (CCPA), as well as employ- as data privacy, algorithmic bias, and as they arise and adjust their strategy
for sustainable success. How can net- align with specific business needs and To mitigate this risk, enterprises must ing advanced security measures like the lack of transparency in AI deci- accordingly. This approach accelerates
works reinvent themselves with GenAI goals. develop robust data governance encryption, network segmentation, and sion-making processes pose signifi- the integration process and minimizes
while addressing unforeseen imple- frameworks that ensure data integrity, endpoint protection. cant risks for businesses. Missteps in the risk of costly setbacks.
mentation risks? Pilot projects can serve as testing security, and privacy. This involves im- this area can lead to legal challenges,
grounds for refining AI models and plementing scalable architectures like Businesses can maintain trust and pro- regulatory penalties, and reputational Agility is key in AI implementation. The
#1: ALIGNING AI WITH STRATEGIC BUSINESS determining the feasibility of large- data lakes or warehouses to efficiently tect their AI infrastructure from cyber damage. fast pace of innovation demands that
GOALS TO AVOID MISALIGNMENT scale implementation. It’s vital to select store and organize diverse data types. threats by adopting ethical AI frame- businesses stay agile and ready to
use cases that offer clear, measurable Additionally, organizations should works and implementing regular audits To address these risks, companies pivot when necessary. By embracing
According to a recent survey, 83% of outcomes. Focus on projects that deliv- employ data profiling techniques to to detect potential biases and vulnera- must develop and adhere to ethical agility, companies can navigate the
companies reported that AI adoption is er quick wins to build confidence and identify inconsistencies and anomalies bilities. Solid security systems like the AI frameworks that prioritize trans- technological complexities of AI inte-
a top priority in their future plans. Yet, secure further investment. By aligning before feeding the data into AI models. NIST Cybersecurity Framework and parency, fairness, and accountability. gration and ensure long-term success.
data suggests that 70% to 80% of all AI AI initiatives with strategic goals, busi- ISO/IEC 27001 ensure comprehensive This includes implementing oversight
projects fail. nesses can sidestep the risk of wasted Investing in reliable data assessment, protection for AI-driven operations. mechanisms to monitor AI systems Preparing for the GenAI leap
resources and foster a stronger founda- cleansing, and validation is crucial to and ensure that they are free from
In the past months, the allure of AI’s tion for AI-driven success. the success of AI projects. By prioritiz- #4: RESKILLING THE WORKFORCE TO biases. Additionally, businesses should As businesses across industries
potential has led businesses to dive ing data quality, businesses can ensure MITIGATE AI-INDUCED DISRUPTIONS communicate clearly with stakeholders reinvent themselves with the power of
headfirst into adoption without a com- #2: STRENGTHENING DATA GOVERNANCE TO that their AI models perform optimally about how AI is used and ensure data GenAI, the opportunities are bound-
prehensive strategy. In many cases, ADDRESS DATA QUALITY RISKS and deliver the desired outcomes. AI is often seen as a double-edged protection measures are in place. less. However, realizing these opportu-
this resulted in misalignments between sword, offering productivity gains while nities requires careful planning and risk
AI projects and overarching business One of the most overlooked aspects simultaneously causing workforce By taking a proactive approach to management. From aligning AI projects
disruptions. As AI takes over repetitive ethical AI use, companies can build with business goals to reskilling the

124 125
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

workforce and ensuring ethical AI use,

companies must proactively address
By developing robust data governance
frameworks, strengthening cyberse-
growth in the age of generative AI. The
key to success lies in balancing innova- THE EVOLUTION OF CONNECTIVITY: EMERGING
TRENDS IN TELECOMMUNICATIONS
the unforeseen challenges accompany- curity measures, and embracing agile tion with responsibility, ensuring that AI
ing AI adoption. methodologies, businesses can side- serves as a transformative force rather
step the pitfalls of AI implementation than a disruptive one.
and position themselves for sustained

Written by Steven Griggs

VP of Sales Engineering, Rajant Corporation
BY JUSTIN HUTCHENS

Justin “Hutch” Hutchens, the Innovation Principal at

Trace3, is a technology leader and visionary. He is
the author of the book “The Language of Decep- Over the past few decades, telecom- rise of quantum computing also poses In a hyperconnected world, busi-
tion: Weaponizing Next Generation AI”, the creator munication technology has become potential risks to current encryption nesses are more hesitant than ever
of Sociosploit, a research blog which examines increasingly integral to our lives. From standards, making it imperative for the to share proprietary knowledge. The
exploitation opportunities on the social web – a social media and video chatting to industry to invest in quantum-resistant risk of intellectual property theft, data
confluence of his interests in both technical hacking IoT systems and edge computing, a cryptographic techniques. breaches, and competitive espionage
and social psychology, and the co-host of Cyber reliable and robust infrastructure is has led to a more guarded approach
Cognition, a podcast focused on the rapidly evolving needed to transfer data from point A Moreover, contingency planning for within the telecommunications sector.
technical landscape of artificial intelligence and to point B. As data demands continue threat response is no longer an option This hesitancy has been exacerbated
machine learning. Hutch is an award-winning public to rise, so too do the complexities of but a necessity. Telecommunication by geopolitical tensions and growing
speaker and has spoken at multiple Universities and securing these networks, managing companies are enhancing their inci- concerns over data sovereignty.
global conferences to include the RSA Conference, proprietary knowledge, and ensuring dent response frameworks to mitigate
and DEF CON. efficiency through visionary approach- breaches effectively while minimizing One notable development is the role of
es like edge computing. The industry is downtime. As cyberattacks become artificial intelligence (AI) in foundation-
at a crossroads, balancing innovation more regular and costly, regulatory al coding. AI is now being leveraged
with security and operational effective- bodies are also stepping in, enforcing to automate and streamline software
ness. stringent compliance measures to safe- development, making coding more ef-
guard data integrity and user privacy. ficient and scalable. However, this also
#1: INCREASED NETWORK SECURITY AND presents new challenges, as AI-driven
CYBERTHREAT CONTINGENCY PLANS #2: HESITANCY TO SHARE PROPRIETARY coding tools rely on vast datasets that
KNOWLEDGE IN THE GLOBAL MARKETPLACE may include proprietary information.
With the proliferation of digital services,
telecommunication networks have
become prime targets for cybercrim-
inals. Cyberattacks are growing in
frequency and sophistication; therefore,
network security measures must evolve
as well. Telecommunication providers
are increasingly investing in proactive
cybersecurity strategies, such as threat
intelligence, real-time network monitor-
ing, and AI-driven security protocols.

One of the main concerns is data and

network traffic privacy. As personal and
business data traverse global networks,
end-to-end encryption and secure
access for users are paramount. The

126 127
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

Companies are increasingly scrutiniz-

ing AI models to ensure they do not
One prominent example of this trend is
the use of platforms like Cowbell, which
score a broader industry shift toward
greater resilience and operational effi- GENAI IN TELECOM 2025 – MACHINE LEARNED
FRIEND OR AI GENERATED ENEMY
inadvertently expose confidential algo- facilitates edge-based security and ciency. Telecommunications providers
rithms or operational strategies. operational intelligence. By processing must strike a delicate balance between
data at the edge, telecommunication security, innovation, and global collab-
The reluctance to share knowledge networks can significantly reduce oration.
is also reflected in stricter data gov- bandwidth consumption, lower costs,
ernance policies. Telecommunication and enhance data security. This ap- Cyber threats are evolving, and compa- Written by Fergus Wills
firms are implementing advanced ac- proach also minimizes the risk of data nies must prioritize advanced security Strategic Marketing Director, ENEA
cess controls, zero-trust architectures, interception, as information remains frameworks and robust contingency
and blockchain-based verification localized rather than traveling across plans to be prepared. Simultaneously,
systems to ensure that only authorized potentially vulnerable long-distance the protection of proprietary knowl-
personnel can access critical informa- networks. edge is becoming more critical in an
tion. This shift is essential to maintain- era where AI-driven coding and global
ing a competitive edge while mitigating Additionally, the rise of 5G and IoT connectivity present new risks. Lastly,
the risks associated with data leaks ecosystems has fueled the adoption the rapid adoption of edge computing
and cyber espionage. of edge computing. With millions of is revolutionizing data processing,
connected devices generating vast making networks more agile and
#3: INCREASING RELIANCE ON EDGE amounts of data, traditional cloud in- responsive and, thus, operations more
COMPUTING frastructures struggle to keep up. Edge streamlined.
computing ensures that latency-sen-
Edge computing has emerged as a sitive applications—such as real-time The telecommunications landscape will
critical solution for optimizing data pro- medical diagnostics or smart city continue to transform, driven by the
cessing and reducing network latency. management—can function seamlessly dual forces of technological innova-
Rather than transmitting all data to without the bottlenecks associated tion and security imperatives. Industry
centralized cloud servers, edge com- with centralized processing. leaders who proactively address these
puting enables data to be processed trends will be better positioned to nav-
closer to its source, at the edge. This #4: A SHIFT TOWARD GREATER RESILIENCY igate the challenges of the digital era
shift significantly enhances operational while unlocking new opportunities for
efficiencies, particularly in industries These three trends—heightened net- growth and efficiency.
that rely on real-time analytics, such as work security, increased hesitancy to
healthcare, manufacturing, and autono- share proprietary knowledge, and the
mous vehicles. expansion of edge computing—under-

BY STEVEN GRIGGS

Steven Griggs is the VP of Sales Engineering at In 2025, AI will continue to be a pow- personal - and effective exploits at extreme volumes.
Rajant Corporation. Griggs holds a business man- erful force for change. It will re-shape • Data Charging Bypass Fraud - The
agement degree from Saint Leo University and an how we think about network and Need For AI & Machine Learning As a result, cybersecurity professionals
electrical engineering technology degree. He served personal security, and how we respond who headed into 2024 expecting to
nine years as an Electronics Technician in the US to security threats. #1: ATTACK OF THE SCRIPT KIDDIES focus on combatting ultra-sophisticat-
Navy, including time on the USS Manitowoc and the ed AI attacks have found themselves
US Naval Special Warfare Group Two. To explore what lies ahead in 2025, Script kiddies are low-skilled hackers fighting a flood of low-quality AI-gen-
we will look at AI from four different who use tools and exploits developed erated attacks.[1] Human analysts
perspectives: by others to launch cyberattacks, and can’t fight such voluminous attacks
• Impact on network security - they’ve been having a grand time with manually. They need AI. At the same
beyond attack of the ‘Script Kiddies’ the deluge of malicious AI resources time, much-feared advanced AI attacks
• Regulation - will first mover being traded on the dark web. They’ve have indeed emerged, ones which are
advantage ‘Trump’ oversight? shown a particular fondness for GenAI essentially impossible for humans to
• Making consumer fraud ever more tools that can churn out low-quality detect, and require AI-based defenses.

128 129
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

#2: AI REGULATION – EXPECT THE on the Internet. It is information that #4: DATA CHARGING BYPASS FRAUD – THE access will become more intelligent Conclusion The good news is that these are ca-
UNEXPECTED is accessible to fraudsters but very NEED FOR AI & MACHINE LEARNING and reactive in 2025, even in the face of pabilties that can be adopted for both
difficult (if not impossible) for us to an ever changing app space. The fact that AI is getting more sophis- defense and offense, but one thing is
In AI as in other technology domains, manage. There has been a lot of focus on the ticated and personal is not a surprise. for sure – ignore it at your peril, as it is
the competing mantras of “move fast different types of fraud that can happen How quickly it is being adopted and effective, even in the hands of ‘script
and break things” and “move slow and In 2025 scammers will leverage such in the telecom environment from iden- becoming a reality in our world is, how- kiddies’.
fix things” are at the epicenter of ap- infomation and new AI tools to push tity theft, privacy breach, subscription ever. In 2025, we will see a new level of
proaches to AI regulation. While many their scams to new extremes, creating (application) fraud, etc. Regular lists activity.
nations worldwide are developing and highly personalized messaging and of top 10 fraud activities have been
implementing AI governance legisla- voice call attacks. The default defense published but a type of fraud which References
tion, the EU is the world’s regulatory mechanism today is government gets little attention is data charging [1] Compare the 2023 survey report “Artificial Intelligence in Cybersecurity” at https://info.enea.com/enea-cybersecurity-insiders-ai-in-
first mover and takes a risk-based, awareness efforts through consumer bypass – in effect users finding ways to cybersecurity-report-pdf to the “2024 State of Network Threat Detection” survey report at https://info.enea.com/network-threat-detection-
product-safety approach to AI policies. warnings. However, the only practical circumvent rating and charging rules in report-pdf.
filter point for defending against these the network to get free data access or [2] See https://aiindex.stanford.edu/vibrancy/
In contrast, the United States has attacks is the network. That said, such to access geo-restricted services. [3] CNN - This bank says ‘millions’ of people could be targeted by AI voice-cloning scams
followed a more fragmented path, with AI scams are significantly more sophis- [4] Price of GB: https://www.visualcapitalist.com/the-cost-of-1-gb-of-mobile-data-worldwide/
policies mainly taking the form of dis- ticated and challenging to detect with In 2025, in our view, this will change,
crete presidential executive orders and traditional firewall technology. and there will be a focus on this type of
legislative acts. The US is, however, misappropriation of data. In a cost-sen-
considered the world’s current leader in One particularly concerning trend is sitive environment, CSPs can’t afford to BY FERGUS WILLS
AI investment and development [2] and for scammers to use AI to harness ex- leak revenue. Users are also becoming
Fergus Wills is a Strategic Marketing Director at
the regulatory path it chooses in 2025 tensive data, often scraped from social more adept in using VPN tools like
ENEA, and has over 20 years of experience in
with respect to AI will have an outsized media, to craft convincing messages Psiphon or more simply HAProxy to
telecoms, 15 of which have been spent in product
impact on the regulatory environment that can mirror someone’s communica- disguise traffic as being sourced from
management, primarily of large scale server-side mo-
for AI worldwide. And at present, the tion style, or include personal informa- ‘zero rated’ domains. Using behavioural
bile internet data products and value added services.
explicit goal of many Trump Adminis- tion that makes them highly relevant techniques to compare access across
During his time at Enea, Fergus has been heavily
tration nominees and appointments is and credible. Such deep personaliza- multiple users will only tell a CSP a
involved in the development and implementation of
to “move fast and break things” across tion significantly boosts the success long time after the event that data ac-
new standards for mobile data access, development
many domains, including government rate of phishing attempts, underscoring cess has been misused. In this year, we
& adoption of NFV and 5G Infrastructure evolution.
regulations. However, at the time of this the gravity of the situation. will see new techniques to recognise
article’s drafting, the only given with the traffic appropriately for charging and/
new US Administration is to expect the Another alarming trend is the use of or regulatory enforcement. In effect,
unexpected. AI for voice cloning and deepfakes. we are moving beyond recognising
Creating a voice clone only takes a server names in TLS key exchanges,
#3: MOBILE SCAMS – AI & MACHINE brief audio clip (as short as 3 seconds into advanced heuristics and pattern
LEARNING GETS PERSONAL [3]) of the target talking. This meth- learning – a perfect environment for
od has been used in various scams, new AI capabilities.
A theme of 2024 has been the rise of including fake kidnapping and grand-
AI-based fakes and deep fakes, which parent scams, where the cloned voice The value of combating this fraud
are increasingly used as tools by scam- of a loved one is used to extract money starts at closing revenue leakage (e.g.,
mers to target individuals. It is a type of from victims. Together with number every GB not charged for is a dollar
fraud that exploits the trust within our spoofing, deepfakes make scams very lost – the world wide value of a GB is
regular communications with friends hard, not to say impossible, to discern. currently $2.59 [4], and with 5G users
and family. And the ploys are moving consuming 20+ GB a month, losses
beyond relatively simple ‘I need help’ As we look ahead to 2025, we will see can quickly mount up). It is also a
messages to more sophisticated forms increasing sophistication in smishing means of providing additional network
of fraud. This more personal form of and vishing attacks as AI use spreads security – i.e., ensuring that in a shared
fraud increasingly incorporates the trail among scammers. Simultaneously, network, resource usage is accounted
of information we leave in our wake as cybersecurity vendors will continue to for, and appropriate access controls are
we interact with websites and services harness AI to mitigate these threats. enforceable. In essence, network data

130 131
 eBook: The GenAI leap: how networks are reinventing themselves

FOUR KEY TRENDS IN AI NETWORK

INNOVATION: AN INSIDE PERSPECTIVE
Written by Stephen Douglas
Head of 5G Strategy, Spirent

EDGE & CLOUD

1. Four Key Trends in AI Network Innovation: An Inside Perspective������������������������������������������� .133

2. Edge Computing for the Rest of Us�������������������������������������������������������������������������������������������������������������������������� .136

3. IT Modernization Will Be Critical to Networking in 2025��������������������������������������������������������������������� .139

High-tech has always been a fast- more incredible. AI evolve from all those perspectives
paced industry, but what we’re seeing and many others.
with the evolution of AI is truly unprec- As a leading provider of test and
edented. It seems like every time you simulation solutions, we work directly Last year, we conducted hundreds of
refresh your browser, you’ll find news of with the companies driving innova- engagements with hyperscale cloud
yet another advanced large language tion across the AI value chain, so we providers, telecoms, network equip-
model (LLM), another new GenAI get to watch these trends unfold from ment manufacturers (NEMs), and
benchmark, another startup applying the inside. Not just inside one cloud others shaping the future of AI innova-
AI to new enterprise use cases. As provider’s data center, or one telecom tion. Based on this work, here are the
quickly as this space is moving for enabling distributed enterprise AI, or biggest trends we’ll be following this
outside observers, however, what’s one vendor developing next-gen AI year.
happening behind the scenes is even network technologies; we’re watching

132 133
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

#1: AS PORT SPEEDS AND PROTOCOLS Today, operators are using advanced workload optimally, combining the best While conventional automation frame- increasingly complex and dynamic focus and investment in advanced net-
EVOLVE, ETHERNET WILL SOON BECOME THE protocols like RDMA over Converged of both public and private worlds. works are the first step in optimizing operational tasks. We expect growing work testing. Even in more controlled
DOMINANT AI NETWORK TECHNOLOGY Ethernet (RoCEv2) to achieve the re- complex network environments, AI use of AI agents that can adapt and re- semi-autonomous implementations,
quired performance. But the new Ultra #3: TELCO AI NETWORK INFRASTRUCTURE- provides a multiplier effect, making spond to changing network conditions, automating testing capabilities become
Data centers are the workhorse for AI, Ethernet Transport specification should AS-A-SERVICE (NIAAS) WILL SEE them more intelligent and adaptive. autonomously or semi-autonomously, essential for effective governance of
and data center operators are investing be finalized this year, and we expect to SIGNIFICANT GROWTH This process has already begun, as to accelerate service delivery and opti- AI agents. It’s only through continuous
billions to transform their architec- start seeing early adoption in 2026. telecoms apply AI to expand foun- mize quality. testing that human network engineers
tures to accommodate exponential Telecom operators have been following dational automation in labs, testing, can validate the efficacy of AI optimiza-
AI growth. That includes deploying #2: MORE ORGANIZATIONS WILL SHIFT AWAY the AI “cloud repatriation” trend with and network management. Now, with As more operators push the enve- tions, facilitate ongoing training, and if
thousands of new graphical processing FROM PUBLIC CLOUD AND INTO PRIVATE AND interest, and many see it as a major the rise of agentic AI in 2025, telcos lope of what’s possible with network needed, quickly roll back changes.
units (GPUs) and other hardware ac- HYBRID AI DEPLOYMENTS opportunity. By providing on-demand will use automated agents to handle automation, we will also see increased
celerators (xPUs) to support ever-larg- regional and localized AI networking
er AI models with billions of dense In the initial explosion of GenAI ser- resources, they believe they can offer
parameters. And it entails building out vices, most enterprises went all-in on enterprises better performance, with
BY STEPHEN DOUGLAS
dedicated back-end networking fabrics public cloud for AI workloads. As the greater control and governance for
to interconnect all those xPUs with the business and regulatory landscape many AI implementations. Stephen is the Head of 5G Strategy at Spirent.
high throughput, low latency, and loss- around AI evolves, however, many are Stephen works for Spirent’s strategy organization
less performance that GenAI inferenc- finding that this approach is no longer There are good reasons for this opti- helping to define technical direction, new innovative
ing and model-training require. the best fit. mism. Telecom networks already reach solutions, and market leading disruptive technologies
most locations where enterprises want which make a real difference. With close to 20 years
Those data center network trends First, governments and regulators are to extend AI and already comply with experience in telecommunications Stephen has been
began in earnest last year. What’s new implementing stringent data sover- local regulations. Telcos also have at the cutting edge of next generation technologies
is how quickly they’re accelerating, eignty laws, mandating that sensitive proven expertise providing multitenant, and has worked across the industry with multiple
especially when it comes to the use of data remain within national boundaries. scalable, and secure networking and service providers, start-ups and Tier 1 OEMs helping
Ethernet for back-end AI networking Organizations seeking to apply GenAI hosting solutions. By offering AI-ready, them drive innovation and transformation. Stephen
fabrics. Given the sensitivity of AI work- in sectors like healthcare, finance, and optimized NIaaS, they can provide: is an ardent believer in connected technology and
loads to any excess latency or packet government also face strict security • Scalable, on-demand AI networking strives to challenge, blur, and break down the silos
loss, many operators used lossless and data privacy requirements. Sover- and GPU resources for multiple which prevent innovation and business success.
InfiniBand for the first wave of AI net- eign and on-premises AI systems offer industries
works. Moving forward, however, more greater control and auditability to meet • Secure enterprise and government
operators would prefer to use Ethernet these mandates. AI communication via secure
due to its standardization and mass service edge (SSE), quantum
market adoption. Some organizations are also bump- secure, and private edge
ing up against technical limitations of distributions
When will Ethernet take the lead in AI public cloud, especially for meeting the • Optimized network performance
networks? Sooner than almost, anyone demanding performance and latency and ultra-low latency for AI
expected. Today, more than 80% of requirements of real-time applications applications using 5G, edge
back-end AI networks use InfiniBand, like AI video monitoring. Many enter- computing, and dedicated wireline
but Ethernet is rapidly catching up. prises are also concerned about high and wireless network slices
According to Dell’Oro Group, it will costs for public cloud resources as they
overtake InfiniBand by 2028. Ethernet scale up, as well as the risk of getting #4: AGENTIC AI WILL UNLEASH AN
port speed evolution is accelerating locked into a single provider. AUTOMATION REVOLUTION IN TELCO
too. While 400G ports have dominated NETWORKS
AI fabrics, most Ethernet ports shipped Given all these factors, shifting to in-
this year for these networks will be house resources for GenAI is becoming Telecoms have long anticipated that AI
800G. By 2028, the majority will be increasingly attractive. Don’t expect and ML advances would help automate
1.6-Terabit Ethernet. most enterprises to abandon public network operations to reduce costs
cloud entirely. Rather, many will begin and optimize service delivery. With the
Finally, we’re following the evolution pursuing a hybrid AI model that allows rise of GenAI, the timeline for these
of Ethernet itself in response to AI. them to address the needs of each expectations has accelerated.

134 135
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

EDGE COMPUTING FOR THE REST OF US analysis makes any real-time decision
making impractical or infeasible. Edge
computing has matured from a cloud
Real-world applications showcase the
practical benefits: computer vision
systems interpret visual data for safety
professionals. This approach helps
maintain operational excellence as
skilled workers become scarcer while
extension into an independent technol- and quality control. At the same time, keeping sensitive operational data se-
Written by Nick Chang ogy platform that addresses challenges predictive maintenance analyzes cure and locally processed. Even minor
VP of Customer Experience, ZEDEDA locally while integrating with cloud operational data to prevent equipment miscalculations in drilling trajectories
systems to support scalable operations. failures and optimize maintenance can significantly impact production
schedules. capacity, making this AI-assisted pre-
#2: AI INFERENCE AT THE EDGE: MOVING cision crucial for operational efficiency.
TOWARD MAINSTREAM ADOPTION In retail environments, edge computing Organizations prioritizing data manage-
transforms existing infrastructure into ment and security find edge computing
AI inference at the edge demonstrates smart systems. AI analysis of standard particularly valuable for meeting both
its value through specific applications security camera feeds can monitor cus- operational and compliance require-
across industries, from image process- tomer queue lengths, suggest when to ments.
ing and personal protective equipment open additional checkout counters, and
(PPE) detection to oil well operations track store maintenance needs while #4: NAVIGATING EVOLVING REGULATIONS
optimization. These implementations maintaining existing security functions. SHAPING THE EDGE COMPUTING LANDSCAPE
leverage local data processing for fast- These implementations demonstrate
er, more precise results. how edge computing adds value to Data privacy, AI oversight, and sustain-
existing investments without requiring ability regulations are reshaping edge
Consider a major U.S. railway’s imple- extensive new infrastructure. computing requirements. Fortune 1000
mentation of edge AI for safety: edge companies must address data sover-
computers with sensors scan pass- #3: DRIVING FACTORS FOR ENTERPRISES eignty mandates while ensuring robust
ing trains every three miles along the ADOPTING EDGE COMPUTING data protection across jurisdictions.
track to detect potential wheel bearing These requirements demand infrastruc-
failures before they cause derailments. Edge computing delivers essential ben- ture alignment with both local laws and
With tens of thousands of monitoring efits driving enterprise adoption: lower global standards.
stations deployed across its network, latency, reduced bandwidth usage and
the system examines each train con- higher productivity through automa- Healthcare organizations exemplify
tinuously during transit. The AI sys- tion. By moving the computing power this regulatory challenge. Medical
tem learns from normal operations to closer to where the data is created, facilities must balance innovation with
establish baselines and then identifies initial processing can occur on site and HIPAA compliance and data privacy
potential failures by detecting anoma- only the necessary data transferred to requirements. Edge computing enables
lies such as failing wheel bearings that the cloud for advanced analysis, elimi- advanced applications like AI-powered
could lead to catastrophic derailments. nating cloud transfer bottlenecks. medical imaging analysis while keeping
This real-time detection system pro- sensitive patient data within regulated
cesses data locally to identify problems The energy sector clearly illustrates boundaries. This local processing ap-
Edge computing adoption has acceler- spending to reach $228 billion by 2028. #1: CONVERGENCE OF EDGE AND AI within minutes rather than hours or these benefits: oil and gas companies proach satisfies technical and regulato-
ated significantly, with well-recognized Amid evolving regulations and global days, enabling immediate intervention use edge computing to capture and ry requirements without compromising
use cases in manufacturing, logistics challenges, executives seek solu- Artificial intelligence and machine when safety issues arise. The imple- preserve critical operational expertise. care quality.
and IT. Non-technical professionals will tions that can provide transformative learning workloads are growing at the mentation demonstrates how edge in- At drilling sites, AI systems operating
increasingly encounter edge comput- business change, such as competitive edge, enabling real-time decision-mak- ference delivers immediate operational at the edge observe experienced drill Edge computing’s local processing
ing implementations in healthcare, advantage, improved efficiency, or ing in energy, healthcare and retail. value while meeting new federal safety masters controlling complex opera- capabilities enable regulatory com-
retail and consumer services, demon- increased safety and security. This Why? Because a tremendous amount mandates for rail operations. tions—managing equipment that must pliance while reducing cloud security
strating its essential business value in change is data driven and enterprises of data is generated along the physical navigate 10,000 feet down and up to risks. New standards governing AI
diverse industries. are increasingly turning to edge com- supply chain, from sourcing, process- Adoption accelerates through three 30,000 feet horizontally with precision. transparency, bias mitigation, and
puting to deploy new use cases closer ing, transporting, and distribution. The key factors: advances in edge-specific The AI runs alongside human oper- governance highlight the importance of
Industry analysis supports this expan- to this data, driving powerful results cost and latency of moving sensor data hardware, maturing ecosystems, and ators in learning mode, comparing edge-based AI management. Industry
sion: IDC projects edge computing across sectors. and camera feed data to the cloud for regulatory compliance requirements. its decisions to those of experienced initiatives like LF Edge and the Margo

136 137
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

Project support this transition by de-

veloping interoperability standards to
ate operational benefits. Furthermore,
these platforms simplify scaling and
Trust security model to ensure system
integrity across global operations. IT MODERNIZATION WILL BE CRITICAL
TO NETWORKING IN 2025
prevent system fragmentation. managing solutions in distributed envi-
ronments, reducing the need for onsite Conclusion
#5: GLOBAL CHALLENGES IMPACTING THE IT staff and enabling local operators
EDGE COMPUTING INDUSTRY to focus on tasks that deliver superior The growing need to process data
business value. locally to energy, manufacturing, trans-
Economic uncertainty and cybersecu- portation, and retail operations is driv- Written by Paul Savill
rity threats drive demand for efficient, Rising cybersecurity threats, particu- ing edge computing adoption. Howev- Global Practice Leader for Network and Edge Computing, Kyndryl
secure computing solutions. Organiza- larly state-sponsored attacks, require er, addressing security and operational
tions facing budget pressures find edge comprehensive edge security mea- challenges are critical to unlocking the
computing platforms deliver compel- sures. Modern implementations must time, cost and productivity benefits of
ling ROI through reduced data transfer integrate zero-trust security, encryp- edge AI. Implementing a solution like
costs and optimized operations. Supply tion, and real-time threat detection an edge computing platform can solve
chain disruptions and tariff impacts while complying with evolving data these challenges and enable transfor-
further encourage the adoption of sovereignty requirements. An edge mational business change for modern
edge solutions that provide immedi- computing platform must leverage Zero enterprises.

BY NICK CHANG

Nick Chang is ZEDEDA’s VP of Customer Experience,

overseeing the customer relationship lifecycle, in-
cluding onboarding, enablement, adoption, support,
retention, and expansion. He brings 30+ years of
experience in worldwide field services, customer
success, professional services, portfolio manage-
ment, and business strategy at companies including
Aryaka, Palo Alto Networks and HPE.

As AI begins to infuse everything from Before any of that is possible, how- Here are three trends that will impact
operations to personnel functions, it’s ever, companies and public-sector the networking industry in 2025:
helping us develop strategies for han- entities must modernize their IT
dling repetitive tasks and eliminating estates. They’re not going to have to #1: NETWORK SECURITY WILL BE TOP OF
human error. We’re also starting to use “rip and replace,” as in past eras. But MIND
AI to identify network risks in real time organizations will need to co-create
and make suggestions for mitigation with experts to make the most of their While the rapid adoption of IoT devices
and recovery. Of course, some de- current IT investments, upgrade when has increased security threats, many
gree of human intervention remains and where necessary, enhance cyber- companies lack the policies and proce-
necessary, but eventually, AI will help security and regulatory compliance dures to manage these incidents. The
us become much more proactive in the capabilities, and make everything work cybersecurity chain is only as strong as
ways we manage networks. together. its weakest link, and inadequate prepa-

138 139
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

ration and training have made humans predictive analytics and real-time Meanwhile, mainframes continue to more proactive by leveraging AI. The enhancing operational reliability and deployment and governance. To help
that link. But even without human error, insights to proactively prevent hazards. be central to many enterprises’ hybrid result is an IT infrastructure that is able providing a superior customer experi- us manage this tremendously powerful
bad actors can infiltrate a company’s Edge AI also will be able to deliver ben- IT environments. And interest in using to minimize downtime by anticipating ence, having a modern IT infrastructure tool, people need to complement the
operations and freeze, manipulate, or efits to manufacturers seeking to man- generative AI on the mainframe ac- problems, automate routine tasks and is a must. use of AI with communication across
steal intellectual property and custom- age power consumption and emissions celerating. Moving forward, modern- operate more efficiently. And when it internal constituencies, and adoption of
er data. A company’s “front door” may more effectively. It is a valuable tool to izing mainframes with quantum-safe comes to helping ensure consistent Ultimately, people must always have a culture of security.
be bolted shut, but what about Trojan enable real-time processing that opti- networks will be crucial as business performance for AI-driven applications, the last word when it comes to AI
Horse deliveries or “open windows” mizes resource consumption, reduces leaders address complex technological,
such as seemingly innocuous connect- defects and improves logistics. security and regulatory challenges.
BY PAUL SAVILL
ed devices?
And across industries, organizations The time for network modernization Paul Savill is the Global Practice Leader for Network
For large and remote work sites, private will benefit from using various types is now and Edge Computing at Kyndryl. With expertise
networks are often the best choice for of AI to focus on the minutiae that can spanning edge, IT services, and optical networking,
connectivity, offering wider coverage indicate looming trends and affect None of the developments in security, he previously led product management and services
and enhanced security. While private proactive and corrective measures. For 5G, edge AI or networking as a service at Lumen Technologies. His deep industry knowledge
5G networks enable new exposures, example, digital twins – when com- will realize their full potential until helps businesses harness value from advanced
they also allow companies to have full bined with Industrial Internet of Things companies modernize their network network and edge strategies. Paul has spoken at key
control over their network infrastructure (IIoT) devices – will help manufacturers architectures. You can’t drive a race car industry events including Mobile World Congress,
– providing better security, data privacy simulate the unique characteristics of on a dirt road. Accordingly, the expo- EDGE Computing World, and Connected Manufactur-
outcomes, and flexibility when imple- their facilities for planning, process de- nential increase in connected devices ing Forum.
mented correctly. However, embracing velopment and execution. But edge AI and modern workloads such as AI will
a zero trust integrated approach that will require seamless integration with require an update of the digital super-
reduces siloed components will be the networks that enable operations highway that many companies have
critical for organizations to minimize across distributed computing environ- been relying upon for decades. Accord-
network vulnerabilities, especially in ments. ing to a recent survey exploring the
the era of private 5G. A zero trust inte- readiness paradox, 44% of enterprise
gration, supported by the creation of #3: QUANTUM NETWORK DEVELOPMENT IT infrastructure is at the end of its life
multiple visibility points and automated WILL CONTINUE cycle, posing imminent and significant
detection and response, helps ensure risks to operational and data security.
the protection of critical applications, The telecom industry will continue
data and systems, and can reduce an investing in the development of the Traditional networks were not designed
organization’s attack surface. quantum networks that will enable ze- with AI’s vast requirements in mind, so
ro-latency communication. But the time as businesses integrate AI into their
#2: EDGE AI WILL “FLATTEN” ONSITE to availability is uncertain. systems more deeply, network modern-
DECISION MAKING ization will become essential to meet-
Quantum computing is another emerg- ing the data and bandwidth demands
Traditional networks were not de- ing disruptor which demands proactive of AI-driven applications. A multitude of
signed with AI’s vast requirements in action now. When quantum becomes forces – including the rapid evolution of
mind, and as businesses integrate AI fully operational, bad actors may find technology, the increasing sophistica-
more deeply, network modernization ways to use its speed and predictive tion of cyberattacks, heightened market
becomes essential to meeting the data capabilities to bypass traditional com- competition, evolving regulations, and
and bandwidth demands of AI-driven puting safeguards. This means quan- critical skills gaps in AI and cyberse-
applications. Deploying AI at the edge tum computers could break traditional curity – will drive the need for network
will help optimize functions and detect encryption, posing serious risks. Pro- modernization.
patterns as they start to form, all while tecting networks against “harvest now,
improving energy efficiency. decrypt later” threats with technologies AI is transforming innovation and
such as datacenter interconnect net- growth in our sector, which is why
In manufacturing, for example, edge works will be crucial. 73% of telecom leaders are investing
AI will be able to provide workers with in it. Network management becomes

140 141
 eBook: The GenAI leap: how networks are reinventing themselves

2025: A DEFINING YEAR FOR THE INDUSTRY

Written by Eugina Jordan

MOBILE
NETWORKS
1. 2025: A Defining Year for the Industry������������������������������������������������������������������������������������������������������������������� .143

2. Standardising Open RAN, Open APIs and Network Automation�������������������������������������������������.147

3. Rampant Evolution Will Transform the Telecoms Industry in 2025������������������������������������������� .150

4. Growing Innovation and Adaptability in the Cellular Connectivity Industry�������������������� .154

5. How mmWave Technology Will Support Network Evolution to GenAI���������������������������������� .156

6. Cloud, AI & 5G SA: Driving the Evolution of Mobile Core Networks��������������������������������������� .159

7. UK Telecoms: Key Trends to Watch in 2025������������������������������������������������������������������������������������������������������ .161

After years of uncertainty, 2025 marks a Multimodal AI will revolutionize cus- phone camera, and an AI-powered
critical juncture for the telecom indus- tomer engagement, delivering seam- assistant overlays step-by-step AR
try. We’ve faced unprecedented chal- less, hyper-personalized service while instructions on their device. No wait
lenges - declining revenues, unfulfilled cutting operational costs. times, no endless calls.
promises of 5G, and sweeping layoffs.
But this year is about something more Let’s face it - traditional customer This eliminates miscommunication,
profound than just recovery; it’s about support is becoming obsolete. In 2025, reduces the time spent resolving
recalibrating, finding solid ground, and multimodal AI will take center stage, issues, and enhances the customer ex-
setting the stage for long-term resil- reshaping how telecom operators perience. Similarly, voice-enabled bots
ience. Here are seven predictions that interact with their customers. Unlike could handle real-time billing queries
will define the year ahead. earlier AI systems, these models can or troubleshoot network problems with
process text, voice, images, and video natural, context-aware interactions,
#1: MULTIMODAL AI: REDEFINING CUSTOMER simultaneously. Picture this: a customer further reducing friction in customer
EXPERIENCE struggling to set up a router opens their interactions.

142 143
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

This isn’t just a technical leap - it’s a suggesting alternate payment plans, and energy optimization. this year will be squeezing every ounce robotics on factory floors demand driven by AI-enabled devices, will offer
financial one. Industry data suggests all without customer intervention. The of value from 5G. ultra-reliable connectivity. some financial relief, but this is not a
multimodal AI can slash support costs synergy between agentic and multi- Here’s the game-changer: larger RIC • Healthcare: Telemedicine and AI- renaissance year. It’s a year of recali-
by up to 40% while enhancing custom- modal AI is poised to redefine telecom and SMO providers, like Nokia and Like a well-tended garden, opera- assisted diagnostics require secure, bration - learning to do more with less
er satisfaction. Operators that invest in customer service, moving it from reac- Ericsson, will open their platforms to tors will focus on nurturing what has high-performance networks. and rebuilding resilience.
these systems won’t just meet rising tive support to predictive, personalized smaller innovators. This collaboration already been planted. Network slicing,
consumer expectations - they’ll define care. will create a vibrant ecosystem, reduce which allows the segmentation of net- Most enterprises, however, will choose Make no mistake, this is not a boom
them. In an era where consumers de- costs, improve quality of service (QoS), work resources to cater to specific use shared 5G models offered by operators, year for telecom. The recovery of 2025
mand instant, omnichannel solutions, Operators embracing agentic AI will and unlock new revenue streams. cases, will move from pilot projects to benefiting from network slicing and is about survival and rebuilding from
multimodal AI isn’t optional; it’s table see real returns, with operational costs Open RAN isn’t just about technology widespread implementation, enabling managed services without the opera- the depths of contraction. The industry
stakes. projected to drop by 20–25%. These - it’s about breaking down barriers and industries like healthcare, manufac- tional headaches of standalone deploy- is learning to operate leaner and more
systems will also pave the way for scal- inviting innovation from all corners of turing, and smart cities to tap into ments. In 2025, the narrative will shift efficiently, prioritizing sustainability over
But here’s the catch: operators can’t ing complex networks like 5G and edge the industry. customized connectivity solutions. Sim- from scaling private networks broadly rapid expansion. While some green
afford to deploy mediocre systems that computing. Combined with multimodal ilarly, edge computing will grow more to refining them for where they’re truly shoots of growth may appear, the scars
frustrate users instead of delighting AI, agentic AI will move telecom from a The financial impact will be significant, sophisticated, bringing data processing needed. of recent years will remain visible. The
them. Half-measures will lead to churn, service provider to a service anticipator with AI-driven automation projected closer to users and devices to support recovery is real, but it is the recovery
and churn is a death sentence in this - setting a new bar for reliability and to reduce operational expenses by latency-sensitive applications such as While private 5G networks will remain of an industry that has weathered its
competitive market. In an era where innovation. 20-25%. Additionally, these tools will augmented reality (AR), autonomous niche, their impact in specialized use darkest days, not one riding a wave of
consumers demand instant, omni- enhance service reliability, improve vehicles, and real-time analytics. cases cannot be overlooked. In 2025, innovation. For telecom, 2025 is not
channel solutions, multimodal AI isn’t By next year, telecom operators will customer satisfaction, and support the focus will be on refining and scaling about reaching new heights - it’s about
optional; it’s table stakes. deploy these AI systems at scale to the scaling of 5G and edge networks, This doesn’t mean innovation is stalled. shared 5G solutions for industries like finding solid ground.
automate diagnostics, predict outages, laying the groundwork for future inno- Network slicing, edge computing, and oil and gas, manufacturing, and health-
#2: AGENTIC AI: AUTOMATING THE and dynamically allocate resources. vations. AI-driven management will drive adop- care, with telecom operators position- #7: APIS: THE FUTURE OF TELECOM
BACKBONE OF TELECOM tion in industries like healthcare, man- ing themselves as strategic partners for INNOVATION
But innovation without trust is mean- But let’s call out the elephant in the ufacturing, and smart cities. But make enterprises seeking reliable and secure
Agentic AI will become the silent oper- ingless. Operators must address ethical room: Open RAN’s success depends no mistake - 2025 is about delivering connectivity. APIs will drive cross-industry partner-
ator, automating workflows, optimizing concerns head-on. Transparency in on a level playing field, and the big on 5G, not dreaming about 6G. ships and open new revenue streams,
networks, and delivering predictive AI decisions and robust data privacy players need to play fair. Smaller in- #6: RECOVERY FROM THE INDUSTRY’S making telecom more than just a con-
service reliability. measures are non-negotiable. Any novators can’t compete if incumbents The hard truth: the industry doesn’t LOWEST POINT nectivity provider.
misstep here could shatter consumer manipulate standards to protect their need a new G - it needs a new plan to
2025 will be the year telecoms lean confidence and derail adoption. turf. Industry regulators and alliances make 5G profitable. Telecom’s recovery in 2025 will come In 2025, telecom’s evolution into an
heavily on agentic AI - autonomous must step up to ensure true openness, not from growth but from necessity, enabler of digital ecosystems will
systems that think, decide, and act #3: OPEN RAN WITH RIC: BREAKING or this “revolution” will become just #5: PRIVATE 5G: A NICHE BUT NECESSARY as the industry rebounds from rock take off, thanks to APIs. Standardized
without waiting for human instructions. BARRIERS AND DRIVING INNOVATION another marketing ploy. TOOL bottom. frameworks, like those pioneered by
Tools like IBM Granite 3.0, Microsoft AT&T, Deutsche Telekom, and Ericsson,
Agents, and Claude 3.5 Sonnet are Open RAN adoption will gain mo- #4: 6G TAKES A BACKSEAT Private 5G networks will remain niche, Let’s call it what it is: 2024 was a disas- will allow industries from healthcare to
already proving their worth, and their mentum as operators integrate Radio thriving only in high-value industries ter. Declining ARPU, layoffs, and ove- retail to seamlessly integrate telecom
deployment is only accelerating. Imag- Intelligent Controllers (RIC) to enhance :6G will remain a distant concept, as where they solve specific challenges. rhyped 5G promises left the industry capabilities into their applications.
ine a system that identifies a cluster of efficiency and foster collaboration. operators focus on making 5G deliver battered. In 2025, recovery won’t feel
dropped calls in real-time and adjusts on its billion-dollar promises. Despite the hype, private 5G won’t like a victory lap - it’ll feel like survival. Location tracking, network slicing, and
network parameters autonomously to Open RAN isn’t just a buzzword take over the enterprise world in 2025. Operators are cutting costs, consolidat- billing APIs will generate billions in new
fix the issue. anymore; it’s becoming a lifeline for Let’s be honest - 6G isn’t happening in High costs and competition from Wi-Fi ing, and finding stability after years of revenue while fostering innovation far
operators needing flexibility and cost 2025, no matter how glossy the R&D 6 and 7 have kept its adoption limited. overreach. beyond traditional telecom. APIs won’t
Consider an agentic AI system mon- control. The addition of RIC capabilities headlines sound. There’s no spectrum, But for industries like oil and gas, man- just connect devices - they’ll connect
itoring customer data across chan- in 2025 will push this evolution further, no regulatory framework, and, frankly, ufacturing, and healthcare, private 5G AI will play a role, but not as a miracle industries, making telecom operators
nels - email, app interactions, and call enabling real-time network optimiza- no money. Operators have poured over will remain indispensable: worker. Predictive maintenance, auto- critical players in the broader digital
center records. If it detects a pattern tion and dynamic spectrum sharing. $1 trillion into 5G, and the returns ha- • Oil and Gas: Reliable, low-latency mated support, and dynamic resource economy.
of dropped calls or failed payments, it Startups will thrive in this space, ven’t exactly been blockbuster. Private networks for remote operations and allocation will help trim expenses and
could autonomously resolve the issue developing niche xApps and rApps to networks? Still niche. Industrial IoT? harsh environments. squeeze value from existing infrastruc- Here’s the bold opportunity: APIs
by adjusting network parameters or tackle challenges like traffic steering Not scaling fast enough. The priority • Manufacturing: Industrial IoT and ture. Modest growth in data usage, shouldn’t just be a revenue stream -

144 145
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

they should be a strategy. Operators

must use APIs to insert themselves into
2025 isn’t the year telecom soars to
new heights; it’s the year it regains
will define the future. The question isn’t
whether telecom will adapt - it’s how STANDARDISING OPEN RAN,
OPEN APIS AND NETWORK
the core operations of other industries. its footing. With tools like multimodal bold it can be in shaping what comes
Be indispensable, or risk being by- and agentic AI, collaborations in Open next.
passed. RAN, and API-driven ecosystems, the

AUTOMATION
industry will start rebuilding from its
Conclusion: a year of resilience and lowest point. This is not about meteoric
recalibration growth but about preparing for the long
game. Recovery is here, but resilience
Written by Gonzalo Mallea
Solutions Delivery Manager, Future Connections

BY EUGINA JORDAN

Eugina, a female executive and an immigrant, Eugina

has over 20+ years of strategic marketing leadership
experience, leading marketing and communications
for small and Fortune 500 global technology com-
panies like Starent and Cisco. Previously, she served
as the VP of Marketing of the major telecom industry
disruptor Parallel Wireless and was instrumental in
creating the Open RAN market category.

For a few years now shift changes have A few trends and predictions can be The future rApps (but also the xApps)
been expected in the telco world, driv- identified: should be capable to implement
en mainly by innovative technology be- sophisticated, high-value use cases by
ing made available, fierce competition #1: RAPPS VS XAPPS FOR THE RIC OPENRAN using Artificial Intelligence/Machine
and savvier end customers wanting Learning (AI/ML) to add pattern fore-
and expecting more for their money. rApps will consolidate as the preferred casting, hence issuing fast close loop
However, operators have been some- application type for the RIC OpenRAN correcting actions in the provisioning
what cautious on their investments, architecture, since the vast majority of systems of Communications Service
expecting to see more evidence that use cases are not needed to meet the Providers.
would reinforce their thinking before strict time-bound requirements typical
deciding to commit to specific innova- of the near real time response of the Finally, from a business perspective
tive approaches and technologies. xApps. RIC-hosting providers may consoli-
date as needed middleman players
2025, though, is expected to be the In addition, rApps allow for easier and in the OpenRAN applications arena,
year where such changes will start tak- quicker use case implementations. This along with the figure of the xApp/rApp
ing place, innovative technologies will is because they use a wider variety vendor specialists, as a way to reach
be supported by more clearly defined of simpler high-level programming the CSPs.
standards and examples of implemen- languages such as Python instead of
tations in the real world will increase C/C++ or Golang, which are more #2: TELCO APIS STANDARDISATION WITH
the operators’ confidence. typically used for xApps. CAMARA APIS

146 147
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

After the meta release in Autumn 2024 Along with adoption of 5G-SA and the mation and critical omissions, which failure prediction. failures and reduce service downtime action is taken by the network itself.
of 25 APIs, the CAMARA Telecom so-called paradigm shift expected in are often difficult to detect. On the oth- to achieve optimal user experience and The network is expected to be capable
consortium, backed up by major telcos the form of new benefits, intent-based er hand, the progress made on several In general, multiple ML models will operational expense. of not just identifying and reporting
and hyperscalers along with many network automation and large lan- AI fields such as using divide-and-con- coexist since the data to process and potential points of failure, but of also
other members (+1100 contributors guage models (LLMs) also bring their quer modular network management monitor comes from several sources The AI-based preventive maintenance implementing a self-healing or failure
and 396 organisations), will ensure inherent complexity with them. This is agents or integrated general LLMs and (i.e. sensors, weather records, network is likely to evolve and standardise to mitigation action plan.
that the new developed applications in addition to the compounded compli- human-expert grade models, show that PM counters, etc.), but the goal re- support the autonomous network
operate seamlessly and securely in any cation derived from their coexistence those obstacles may be addressed. mains the same: anticipate equipment journey, where an intelligent course of
network. Such APIs have the power to with older RAN technologies such as Further progress will be made in 2025
provide opportunities to monetise the 4G or 3G. In a not too distant future, but we will know only later.
investments in 5G, by offering powerful this will push many CSPs to embrace
BY GONZALO MALLEA
network capabilities to applications im- a journey towards zero-touch autono- #4: PREDICTIVE MAINTENANCE
plemented in a wide range of use cases mous network, as defined by the TM Gonzalo Mallea is a Solutions Delivery Manager
such as credit card fraud prevention, Forum. Intent-based automation will Telecommunications industry is one of at Future Connections. He is responsible for client
low latency video streaming, semi-au- be an important piece of work as early the most complex and rapidly growing liaison and multidisciplinary team coordination, for
tonomous vehicles and augmented as in Level 3 ‘conditional autonomous industries where key players such as full end-to-end product and service delivery. Gonzalo
reality. network’ of the autonomous journey, CSPs attend millions of users consum- has matured more than 20 years of international
and it will definitely be a prerequisite ing a wide range of digital services. experience in mobile telecommunications and, prior
Analysts indicate thatthe API market for those few CSPs working towards The complexity of the underlying to that, in facial recognition.
shows a potential of $100 to $300 their Level 4 ‘highly autonomous net- technologies (multiple RANs such as
billion in the next five to seven years. work’, where human supervision is still 5G, 4G, 3G, CORE, Transport, …), and
However, in the past up to two thirds required. infrastructure that make such services
of the value was lost by CSPs to other possible is paramount, and ensuring
players. For example, the far more flex- Intent-based automation refers to the the ‘promised’ service levels to the
ible WhatsApp and iMessage applica- fact that specific goals at the right level end customer means that nowadays
tions killed the expensive and outdated are formulated, leaving to the network AI-based predictive maintenance is
SMS service. an autonomous (and intelligent) course a must in the near future, if not in the
of action. As a consequence, the use of present time.Analysts have estimated
Now, CAMARA open-source project AI/ML technology becomes a must for that the predictive maintenance market
has an opportunity to change that pic- intent automation. Specifically, the use willreach USD 9 billion with and ex-
ture by allowing developers to deliver of LLMs as a subset of Machine Learn- pected growth of CAGR 17% until 2028.
new applications capable to exploit ing (ML) is an interesting proposition,
the unique 5G network benefits across bearing in mind its considerable capac- AI models applied to network main-
any network. However, agreement ities in natural language processing as tenance make possible for CSPs to
on key APIs, timelines and coordinat- well as the significant progress made in evolve from traditionally expensive and
ed efforts among telco players must reasoning. labour-intensive reactive maintenance
happen to succeed. For example, APIs to a preventive approach, where the
addressing location and fraud preven- However, it is also fair to say that LLMs interventions needed are predicted
tion may be relevant for banking or have encountered difficulties when try- and, if in close loop mode, self-healing
fintech and should not face regulatory ing to be applied on the very technical actions are carried out. Already several
issues. Also, other APIs looking at and highly specialised world of telecom operators such as AT&T, Telefonica and
quality on demand (i.e. applicable to network management. Examples of Vodafone have applied AI/ML models
bandwidth-sensitive or delay-sensitive difficulties could be the comprehension to implement predictive maintenance.
applications) may be of interest for of full context on data manipulation The range of applications goes from
certain customers too. and highly complex planning and deci- identifying signs of early equipment
sion-making, domain-specific challeng- degradation (signal degradation, ab-
#3: INTENT-BASED NETWORK AUTOMATION es such as applying complex technical normal temperature, etc.) to monitoring
AND LARGE LANGUAGE MODELS mobile standards, and hallucinations of power usage or real-time tracking
coming from apparently correct infor- of environmental factors to anticipate

148 149
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

RAMPANT EVOLUTION WILL AI regulations and governance Further evolution in 5G standardization

defines how satellites can be used to

TRANSFORM THE TELECOMS

What we should expect to see in 2025 extend a 5G service and enable true
is a rise in AI regulations on all possible connectivity in the middle of nowhere.
levels. AI is a genie that is trying to get

INDUSTRY IN 2025
out of the bottle, and as with any genie, 5G standards also define viable
we don’t know if it will be a kind one solutions to enable 5G devices to talk
until it is out, so we need to do our best directly to each other, and moreover, to
to manage it before it is completely out relay messages from isolated devices
of our control. all the way to the 5G network and back.
Written by Anatoli Levine Both satellite and device-to-device
Director of Products and Standards, Softil AI regulators should setup boundar- communications are not just useful,
ies – where AI can be used, where AI they are often critical, and we can
cannot be used, how to deal with the expect further advancements to be
consequences of using AI and how made in this area in 2025, moving from
It’s hard to believe, but another year is technology would go? to ensure that humankind will benefit standards to implementations.
about to become history. Time is relent- and not suffer from this most-capable
less and offers no favors to anyone, Technology keeps moving forward, technology. You should expect a rise #3: 6G
going only in one direction. Hence, let’s but not at such a neck-breaking speed in AI regulations on all governmental
make the best out of it and engage in anymore. The introduction of the com- levels worldwide in 2025. The most We are still implementing 5G – so
our favorite fun exercise of crystal-ball- puter as we know was revolutionary responsible companies might even try how come there is so much talk about
ing the future. – but the increase in speed, no matter to self-regulate, but here we might be 6G? Simple. If we look at the basics
how incredible, is not – it is simply an stepping into fantasy land. of technology (well, there is nothing
Predicting the future in technology is evolution. Even the beloved darling of basic about these basics, but let’s finish
not only fun, but daunting. Everyone everyone, Artificial Intelligence, is not #2: 5G ADVANCEMENTS – SATELLITES AND this chain of thought), we have three
wants to predict revolutions. Everyone a revolution – the concepts of AI were DEVICE TO DEVICE elements that are the cornerstone to all
wants to be the first to spot the ulti- developed back in 1970s – we just now other developments – computational
mate trend that nobody noticed before, reached the computational power and 5G is old news, of course. However, speed, memory/storage, and commu-
to be proud of the real achievement. storage availability which are allowing it is evolution that we are looking for nication speed/availability/reliability.
The only problem, however, is the sheer us to make usage of AI practical. here. There are a few important points
pace of technological advancements. to consider. First, 5G deployments are You can imagine 6G technology as an
But, be under no illusion – this is not showing good growth – it is not expo- ever hungry beast – the more it is fed,
Think about it. IBM’s 360 Model 30 - revolution. We live in an age of tech- nential, but it is significant. According the more it needs to eat – we always
the state-of-the-art computer - was nological evolution, so let’s talk about to the GSA (Global mobile Supplier As- need more computational speed, more
introduced to the telecoms world 60 evolutionary trends for 2025. And sociation), there were 348 commercial memory, and faster communications.
years ago in 1964. This was a so called remember, evolution simply means that 5G networks launched in 129 countries 6G is about quenching that thirsty
mainframe computer, occupying the whatever it is, it is the same thing, only and territories worldwide as of July of basic element – making communica-
space of the size of the average super- better. 2024. Also, the number of commer- tions faster, more reliable, and more
market, and boasting 64K of memory. cially available 5G devices increased available.
#1: AI 2.5 times over the course of the year.
Today, 60 years later, an average These numbers show wide adoption of Many of the previous “Gs” were de-
desktop computer, using the Intel i9 Yes, of course AI. How can we not start 5G technologies and offer a foundation signed for the benefit of technological
CPU and occupying a tiny portion of our annual conversation without the for further growth. advancement. But today networking
your desk, has 250,000 times more all-mighty AI? The prediction is simple is literally the “fabric of life” – com-
memory at modest 16G speeds and is – AI will become better, and it will solve And here is the next important point: munications need to be available for
about 5.5 million times faster at 191,000 all of our problems. Right? Well, con- by design, 5G includes elements all, everywhere, all the time. Up and
million instructions per second. The sidering that everyone is talking about which are critical to achieving ubiqui- coming 6G implementations have to
difference is astounding – but do you AI, but very few people actually know tous broadband (more on this later), be designed to accommodate not only
think anyone back when the IBM 360 what it is, our prediction should be a lot such as satellite communications and pure technological needs, but societal
was introduced could predict how far more focused. Let’s try again. device-to-device communications. needs, and the only way to do it is by

150 151
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

designing it accordingly from the start. voice, video, and data). The evolution of learning. Analyze what happened in of our lives, literally next to bread and solved. Establishing such ubiquitous connec-
Expect more work to be done in 2025 MCX is ongoing, and while the adop- the past and learn how to prevent it water. tivity is an ongoing trend, and we are
in defining 6G use cases before a sin- tion might not be moving at the speed in the future; find loopholes and weak Today, we are fully reliant on mobile slowly developing all the necessary
gle architecture diagram is drawn up in we all desire, progress is steady. points and patch them before they can Heck, we can even boldly proclaim that networks to communicate with one an- technologies such as satellites, for
the international and standards bodies. be used. Machines are much faster the ability to connect to the network other, and more importantly, to be able example, and mobile networks capa-
There are more MCX devices on the than humans, so AI gives us great hope is a “fundamental human right.” Yet to ask for help when necessary. And ble of using satellites to provide this
#4: DRONES AND ROBOTICS market, more service providers em- not only in defending against, but also network connectivity is not universal – these networks are not just for humans ubiquitous connectivity. The need for
bracing MCX, and those who do not preventing cyber threats before they there are many places on Earth where anymore – machines also need to be ubiquitous broadband is dire, and
Both had been around for a while, but yet deliver the services have concrete have a chance to become impactful. network connectivity is not available, able to talk to one another, exchange whilst evolution towards it is slow, it will
both drones and robotics are attracting plans and timelines to do so. We should expect a lot of innovation in and this is a problem that needs to be data and make decisions. flourish albeit at a steady pace.
more and more interest, and thus the the AI+Cyber space next year.
speed of innovation in this sector is The same MCX technology is also
accelerating. making its way into future railway com- #7: INCREASING PITFALLS OF
BY ANATOLI LEVINE
munications, not only as the underlying TECHNOLOGICAL ADVANCEMENTS AND
And, of course, one of the most im- technology for FRMCS (Future Railway EVOLVING HUMANS Anatoli Levine is Director of Products and Standards
portant accelerators here is AI. As AI Mobile Communications System), but for Softil, Ltd., responsible for developing strategy and
algorithms and capabilities develop, also on its own. The story is as old as the world itself. product roadmap for the Softil’s portfolio of enabling
AI will enable drones to be more and Each and every technology and ad- products for developers, including technologies
more “independent” – set the task and We expect steady progress to continue vancement can be used for good and such as Mission Critical Communications (MCC) over
let the drone deal with it. Same goes next year not only with more advanced can be used for evil. Gun powder was LTE and 5G, WebRTC, VoLTE/ViLTE/RCS, SIP, IMS and
for “robots”, no matter what shape and solutions on the market, but also with invented for entertainment (fireworks), many others. Mr. Levine actively participates in the
form they are taking. some important breakthroughs such as hundreds and hundreds of years before development of open international communication
implementations of MCX device-to-de- anyone decided to put it inside bullets. standards at industry bodies such as 3GPP, ETSI, IETF
Products in both categories are get- vice (direct mode) communications. and other SDOs.
ting better, more independent, and The Age of AI we are entering right
advancements will continue at an #6: CYBERSECURITY now has amazing potential for great
increased pace. That actually begs deeds – but it has an equal potential for
another consideration – remember we Of course, cybersecurity has been on the bad. Think about deep fakes, for ex-
said “AI Regulations” are critical? Same top of everyone’s mind for a long time. ample, where AI can be used to create
applies here – while the drone might As cybersecurity defense mechanisms realistically looking pictures and videos
be fully autonomous and goal oriented, improve, so do the cyber-attacks – a of people and events that never took
it is important to set the boundaries never-ending arms race. It is hard to place. AI is here to stay, so humans
and have the way to enforce them. But believe that some companies still ig- need to evolve and live with it.
yes, we can expect to be amazed at the nore the importance of proper security
pace of innovation in 2025. implementation, so this should be one There will be a learning process, and
of the simplest trends – implement your there will be no instruction manual, but
#5: MCX, A.K.A. BROADBAND GROUP cybersecurity solutions, using state-of- we must start this journey now. This
COMMUNICATIONS FOR PROFESSIONAL the-art techniques such as Zero Trust “learn to live in the world with AI” will
USERS and Multi Factor Authentication. But be an ongoing trend for the foreseeable
the most important trend in cyberse- future and has already started.
With ubiquitous mobile broadband be- curity relates to what we already talked
coming truly ubiquitous, so raises the about – AI. #8: UBIQUITOUS BROADBAND FOR EVERYONE
opportunity to make broadband group AND EVERYWHERE
communications truly ubiquitous too. Cybersecurity + AI/ML
Food. Water. Network connectivity: as
Let’s make sure we are on the same AI and ML (machine learning) might odd that it might sound, these are the
page here – we are only talking about be the best tools to defend against three most basic needs of humans. For
standard-based, interoperable broad- cyberattacks. In cyber, above all, it is good or for bad, but the ability to call,
band group communications, com- critical to learn from past mistakes exchange messages and obtain infor-
monly known as MCX (mission-critical – and this is where AI excels – in mation became a fundamental element

152 153
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

GROWING INNOVATION AND ADAPTABILITY IN

ing the integration of CBRS spectrum solutions without overhauling their organizations to align infrastructure
with distributed systems, are lowering existing set-ups. These systems sim- growth with long-term goals and
the costs and complexities associated plify upgrades and allow businesses emerging trends.

THE CELLULAR CONNECTIVITY INDUSTRY

with private 5G, making it more acces- to scale incrementally in response to
sible for a broader range of organiza- changing needs. The road ahead
tions.
#5: SHIFTING TOWARD PROACTIVE 2025 marks a turning point in the evo-
#4: ADAPTING TO CHANGE WITH SCALABLE CONNECTIVITY PLANNING lution of cellular connectivity. The ex-
Written by Bruce Lancaster INFRASTRUCTURE pansion of spectrum availability, greater
Chief Executive Officer, Wilson Connectivity Historically, connectivity challenges technological convergence, practical
The rapid pace of technological evo- were often addressed reactively, deal- private 5G deployment, and a shift
lution presents a unique challenge for ing with issues as they arose. However, toward proactive planning all signal an
organizations. Nearly half of leaders there is a growing shift toward proac- industry rising to meet the challenges
The cellular connectivity and 5G #2: THE CONVERGENCE OF TECHNOLOGIES diverse range of applications. in IT and operations cite the speed of tive planning, with 27% of organiza- of a connected future.
landscape are undergoing a signifi- change and complexity of integrating tions adopting this approach and 35%
cant transformation. With 2025 poised Technological boundaries in connec- #3: THE PRACTICAL DEPLOYMENT OF new solutions as significant obstacles. using a mix of proactive and reactive Connectivity is no longer just a utility
to bring remarkable advancements, tivity are blending as industry trends PRIVATE 5G However, the availability of modular, strategies. but a strategic enabler supporting in-
the focus is shifting toward scalable, move toward seamless integration. scalable connectivity solutions helps novation and growth across industries.
flexible solutions that address present Modern users care less about whether Private 5G networks, once considered businesses integrate emerging technol- Treating connectivity as a core element By focusing on fundamental connectiv-
challenges while setting the stage for their devices operate on Wi-Fi, LTE, or futuristic, are now transitioning from ogies more seamlessly. in operational strategy, rather than ity infrastructure and scalable solu-
future innovation. This evolution under- 5G networks. Instead, the emphasis lies concept to reality. Recent research an afterthought, ensures businesses tions, enterprises can confidently meet
scores the critical role of connectivity in ensuring uninterrupted, high-quality suggests 41% of businesses view Infrastructure innovations, such as mitigate risks, optimize budgets, and today’s complexities while preparing for
as the foundation for modern business performance regardless of the system. private 5G as a key investment, with distributed systems that can scale remain agile in the face of technolog- tomorrow’s opportunities.
operations, requiring practical function- adoption rates in sectors like manufac- across wide frequency ranges, enable ical change. Proactively addressing
ality and long-term adaptability. This convergence of technologies is turing exceeding 70%. These networks organizations to adopt future-ready connectivity requirements enables
essential for enabling smart cities, offer tailored performance and security,
#1: EXPANDING SPECTRUM AVAILABILITY industrial IoT, and autonomous sys- making them ideal for environments
tems, where uninterrupted transitions with specialized operational demands BY BRUCE LANCASTER
In recent years, spectrum availability between networks are critical. A unified like factories, campuses, and ware-
Bruce Lancaster is the Chief Executive Officer at
has dramatically increased, unlock- approach to connectivity ensures houses.
Wilson Connectivity. Prior to joining Wilson Con-
ing new opportunities for enhanced mobility, low latency, security, and high
nectivity (previously Wilson Electronics) in late 2015
connectivity. The emergence of private throughput across an increasingly Simplified deployment models, includ-
Mr. Lancaster spent 15 years at Logitech, where he
LTE and CBRS (Citizens Broadband
most recently served as Vice President and General
Radio Service) has added bandwidth
Manager of its Harmony Home Control and New
equivalent to all legacy 3G spectrum in
Ventures groups. While there, he led the charge in
North America. Similarly, the rollout of
developing enterprise-level products across all areas
C-band frequencies has doubled the
of the company.
spectrum available in the region within
just four years.

This expanded spectrum availability

creates new pathways to deploy private
5G networks and other tailored con-
nectivity systems. Organizations now
have more options to meet immediate
and evolving operational needs, while
reducing dependency on traditional
network infrastructure.

154 155
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

HOW MMWAVE This is where the 60 GHz frequency

band – known as the “V” band – comes
data centres. This enables faster and
more localised processing – a winning
scenarios.

TECHNOLOGY
into play. As a licensed-exempt spec- combination of speed and security. The unique properties of 60 GHz meet
trum with 14 GHz of unutilised capacity, By allowing AI workloads to operate all the rigorous requirements of modern
60 GHz is an accessible, scalable, and closer to the user, mmWave reduces applications. Limited wall penetration

WILL SUPPORT
cost-effective solution for high-band- latency further, ensuring that real-time minimises signal propagation beyond
width applications like GenAI, free from GenAI applications remain smooth and intended spaces, effectively reducing
the typical challenges of spectrum efficient. interference levels to near zero. This

NETWORK
allocation and interference. makes it ideal for dense settings such
#3: PRACTICAL USE CASES WILL SHOWCASE as hospital, data centres, and event
For GenAI, this capability is transforma- MMWAVE’S VERSATILITY venues, where high-speed, interfer-

EVOLUTION TO
tive. LLMs, real-time video generation, ence-free connectivity is crucial.
and AI-powered analytics all need Practical use cases demonstrating
reliable and scalable infrastructure. mmWave’s transformative impact #4: INNOVATIONS IN MMWAVE WILL DRIVE
COST-EFFECTIVE ADOPTION

GENAI
The ability of mmWave to handle these include fixed wireless access (FWA),
workloads seamlessly means that where mmWave networks power AI-
bandwidth limitations will no longer based monitoring for smart grids (as These collective capabilities are
constrain GenAI outputs. one example) and analytics. GenAI can spurred on by mmWave innovations,
take this to the next level by generating namely advanced antenna technology
Written by Wim Van Thillo #2: LOW LATENCY NETWORKS WILL synthetic datasets for modelling or sim- and beamforming. Rather than broad-
CEO & Co-Founder, Pharrowtech TRANSFORM REAL-TIME AI APPLICATIONS ulation scenarios. Elsewhere, AI-pow- casting widely, beamforming focuses
ered AR/VR applications are augment- signals in specific directions, ensuring
Real-time responsiveness is another ing gaming, professional training, and reliable connectivity even in challeng-
critical function for GenAI. Applications education by delivering immersive ing environments. By tracking devices
Across the world, generative AI (GenAI) such as virtual assistants and immer- experiences that depend on low-laten- and minimising interference, beam-
– the artificial intelligence technology sive augmented and virtual reality (AR/ cy and high-bandwidth networks. Here, forming supports the high bandwidth
that produces text, images, audio, and VR) systems depend on a near-instan- GenAI capabilities include real-time and low latency requirements of GenAI
even synthetic data – is transforming taneous exchange of data. Delays of content creation. applications.
industries. Last year, according to a even a few milliseconds can disrupt the
Riverbed survey, over a third (34%) of user experience or, in the case of au- Autonomous systems like vehicles, Moreover, adopting CMOS-based man-
organisations had adopted Gen AI for tonomous systems, compromise safety. drones, and robotics rest on real-time ufacturing processes– the world’s first
IT operations or prototyping; this year, AI-powered decision-making. Har- CMOs 60Ghz transceiver exists, cover-
that will nearly double to 67%. Still, this MmWave technology ensures ultra-low nessing GenAI can create simulations ing the full 60 Ghz band – has further
revolution relies on an unsung hero: become a cornerstone of next-genera- critical for training large language mod- latency, enabling real-time interaction for training these systems or produc- elevated mmWave technology by re-
the network infrastructure capable of tion networks, enabling GenAI applica- els (LLMs) and the high-throughput and decision-making for AI systems. ing synthetic environments to test ducing costs and power consumption.
meeting its immense demands. As tions to thrive. requirements of real-time AI inference. For example, an AI-powered virtual decision-making models. Industrial This makes it more accessible across
GenAI becomes mainstream, the need assistant can respond to a user’s query automation has levelled up factories industries like consumer electronics.
for networks that can handle intensive #1: ULTRA-HIGH BANDWIDTH WILL POWER Beyond its sheer capacity, mmWave instantaneously, while an autonomous through AI-enabled predictive main-
computational workloads has never DATA-INTENSIVE AI WORKLOADS offers practical deployment advantag- vehicle can process sensor data and tenance and autonomous machinery. Additionally, the integration of mmWave
been greater. es. Its ability to enable gigabit speeds make critical decisions in real-time. GenAI can help here by creating virtual into Wi-Fi mainstream standards,
Known for its ultra-high bandwidth and wirelessly in hard-to-reach areas Older network architectures based on replicas – known as digital twins – of like the newly created IEEE 802.11bq,
Millimetre wave (mmWave) radio low latency, mmWave refers to 24-100 makes it a valuable alternative to fibre, sub-6 GHz frequencies, which prior- machinery for testing. showcases its growing adoption.
frequency (RF) technology can meet GHz – a frequency range that can carry which often entails expensive and itise broad coverage over speed and Standardisation ensures interopera-
these demands; it can support the next an extraordinary amount of data. Unlike time-consuming groundwork. From ru- responsiveness, cannot match this In healthcare, AI-assisted diagnostics bility and accelerates its use in GenAI
generation of AI-driven applications traditional wireless technologies that ral areas to dense urban environments, performance. and remote surgeries require high data applications, cementing mmWave as
and act as a key enabler for this next operate on sub-6 GHz frequencies, mmWave provides high-speed, de- transfer rates. Potential GenAI appli- a cost-effective, reliable, and versatile
stage of network evolution. With its paired with progressions in signal cod- pendable connectivity where traditional Moreover, mmWave’s scalability sup- cations include generating synthetic solution that can support the demands
ability to deliver ultra-high bandwidth ing, mmWave can transfer thousands networks fall short. ports edge AI and distributed comput- medical images for training AI models of our digital future.
and low latency, mmWave is poised to of times more data. This capacity is ing, reducing the load on centralised or creating real-time surgical simulation

156 157
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

To reap the most rewards, GenAI’s

complex demands on network infra-
demands head-on, making it a transfor-
mative enabler for the next generation
ments, enterprises, and technology
providers must collaborate to acceler- CLOUD, AI
& 5G SA:
structure must be managed effectively. of GenAI applications. ate adoption and build the networks
No longer is reliable and scalable con- that will nourish GenAI-powered prog-
nectivity – which means high band- It’s at the very heart of the networks ress. Prioritising mmWave technology –

DRIVING THE
width and low latency – an optional needed for GenAI applications in 2025 specifically the 60Ghz frequency band
extra. Now, it’s essential for unlock- and beyond. Embracing this requires a – creates a solid foundation, ensuring
ing GenAI’s full potential. Uniquely, commitment to investment in mmWave countless industries can seize tomor-

EVOLUTION OF
mmWave technology meets these research and infrastructure. Govern- row’s opportunities today.

BY WIM VAN THILLO

Wim founded Pharrowtech in 2019 and has led it

since then. He has more than 15 years of experience
focused on mmWave R&D, first as a researcher and
MOBILE CORE
later as a program director for imec’s activities in
cellular, WiFi and mmWave transceivers. In that role,
he was essential in shaping the R&D roadmap and
NETWORKS
business opportunities for emerging wireless tech-
nologies. Wim was also a visiting researcher at UC Written by Steven Van Zanen
Berkeley, focused on researching 60GHz high speed Chief Marketing Officer and Head of Product Management, BroadForward
wireless communications. He holds a PhD in Electri-
cal Engineering and a degree in Business Economics
from the Katholieke Universiteit Leuven.
The mobile industry is influenced (SCP), required for control of signaling Routing Agents (DRA) and Signal
by many factors, but increasingly by and the Security Edge Protection Proxy Transfer Point (STP) with systems that
geopolitical tensions, the accelerat- (SEPP) necessary for roaming inter- can run in more modern data centers,
ing adoption of 5G standalone (SA), connect. The SCP provides centralized e.g. the cloud. The acceleration of this
and advancements in AI. We observe traffic management and optimization, transition in the coming years will be
a trend in the Mobile Core Network monitoring/tracing capabilities, and determined by cost considerations, fur-
market towards consolidating signal- seamless scalability, which are import- ther systems going end-of-life, and the
ing products across 2G to 5G, often ant for 5G’s dynamic environments. necessity to mitigate resulting security
replacing outdated legacy systems. As SEPP, mandated by 3GPP for securing risks.
operators transition to modern soft- 5G inter-operator signaling, becomes
ware-based architectures, priorities indispensable as – according to Juniper Although incumbent vendors no longer
include finding new revenue streams, - connections will increase 900% from provide a roadmap, development,
cutting costs, enhancing flexibility and 53 million in 2023 to 526 million by or support for older systems, 2G-4G
security, while future-proofing their 2027. With these technical demands, services will remain operational for
networks for long-term success. SCP and SEPP will be priorities in 2025 years, driven by roaming commitments,
for operators transitioning to stand- legacy device usage, rural 2G connec-
#1: BROAD ADOPTION OF 5G STANDALONE alone architectures. tivity needs, and regulatory obligations.
(SA) AND INCREASED INTEREST IN SCP & As operators look to secure-proof their
SEPP FUNCTIONALITY #2: CONSOLIDATION OF SIGNALING networks, they recognize the inefficien-
FUNCTIONS ACROSS 2G-5G NETWORKS cies of maintaining unsupported, costly,
In Q4 2024 GSA reported that 151 vendor-dependent, appliances for 2G-
operators in 63 countries are invest- Another trend is operators consolidat- 4G alongside their 5G infrastructure.
ing in 5G SA networks. Indeed, we ing various existing signaling functions Consolidation through cloud-native
are already seeing increased interest as they adopt cloud-native SEPP and solutions offers service providers scal-
in various 5G core network functions SCP platforms for their 5G networks. ability, flexibility, and unified operational
such as the Communication Proxy This entails replacement of Diameter management of services. Both public

158 159
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

and on-premise cloud migrations sup-

port this transformation, with operators
viewing this as a path to lower total
particularly in signaling, are still in their
infancy.” MNOs and vendors will ex-
periment with AI, but tangible deploy-
multi-protocol firewalls, and real-time
(anomaly) threat intelligence. UK TELECOMS: KEY TRENDS TO WATCH IN 2025
cost of ownership (TCO) while ensur- ments for signaling will largely focus In 2025, signaling security will remain
ing long-term operational resilience. on narrow use cases, such as anomaly a top operational priority for operators Written by Luke Kehoe
detection, rather than automated deci- and governments, as they invest in Industry Analyst, Ookla
#3: AI AND SIGNALING—THE BUZZWORD sion-making. strategies and technologies to mitigate
WITH LIMITED IMMEDIATE IMPACT these evolving threats and safeguard
#4: CLOSING THE SIGNALING SECURITY GAP critical communications infrastructure.
Artificial Intelligence (AI) will continue ACROSS TECHNOLOGIES AMID STATE-LEVEL The past year has seen many UK oper- #1: MARKET CERTAINTY AND NETWORK short-term, the tie-up paves the way for
to dominate as a buzzword in telecom- THREATS #5: CLOUD TRANSFORMATION ACCELERATES ators lead the way in 3G retirement and CAPACITY WILL IMPROVE THANKS TO THE the merged entity to facilitate deeper
munications. However, the adoption 5G Standalone (SA) deployment, but THREE-VODAFONE MERGER sharing between the separate Three
of real-time signaling optimization and Rising geopolitical tensions and By the end of 2025, we foresee cloud the country still lags on overall mobile and Vodafone networks (e.g., MOCN
decision-making applications in core sophisticated cyberattacks, extending adoption for signaling functions will be network performance and consumer The successful approval of the merger sharing) to enable subscribers of both
networks will remain a promise in 2025. beyond traditional criminal activities, a trend among operators of all sizes. satisfaction compared to its European between Three and Vodafone by the to benefit from different site footprints
are heightening signaling security Both public and private cloud de- counterparts. Competition and Markets Authority and spectrum (e.g., 1800MHz band).
Nevertheless, there are several prom- concerns significantly. This trend is ployments will dominate in delivering (CMA) brings much-needed certainty
ising signaling use cases for AI and compelling operators to prioritize scalability, flexibility, and operational Over the past eight years, the UK has to the market in 2025. This decision is #2: MOBILE DATA TRAFFIC GROWTH SET TO
Machine Learning in mobile networks, security-by-design solutions capable of efficiency. shifted from being a G7 leader to a expected to encourage long-term com- CONTINUE MODERATING
such as: real-time anomaly detection and seam- laggard in mobile download speeds at mitments to capital spending on net-
• Network security monitoring and less cross-network intervention. Cloud-based architectures are in- the 10th percentile—a key indicator of work upgrades across all operators in In 2024, Ofcom reported an 18% in-
anomaly detection. creasingly popular because of the baseline network performance, as it re- the UK market, not just those required crease in total monthly mobile network
• Network capacity requirement State actors and organized cybercrim- ongoing shift to 5G SA, consolidation flects the experience of users with the by the merged entity for integrating traffic in the UK, marking a slowdown
predictions based on signaling inal groups are increasingly exploiting of legacy systems, and the increas- slowest connections. These speeds are its networks and complying with the from the 25% growth observed in
traffic trends. vulnerabilities in signaling systems. ing need to integrate with evolving now lower in the UK than in all but one CMA’s behavioural remedies. both 2022 and 2023. It is likely that
• Proactive signaling resource Their soft targets are often off-the-ra- enterprise services. Container-based other G7 country (Japan), with the gap this trend of moderating mobile data
allocation based on signaling traffic dar legacy systems, End-Of-Life, and deployments will take center stage, to the leader (France) now widening With the consolidation and redistribu- traffic growth will continue into 2025,
patterns. no longer receiving security patches. enabling rapid deployment, seamless rapidly on a year-on-year basis. tion of spectrum benefiting both the reflecting an S-curve profile increasing-
• Optimization of platform service High-profile incidents, such as Chi- scaling, and efficient maintenance or merged entity and Virgin Media O2, ly evident in developed markets across
logic configuration. na-linked hackers accessing surveil- updates of signaling functions. Public Despite the challenges, there is a two of the three in-market operators Europe and North America with high
lance data from telecom providers, cloud partnerships (e.g., AWS, Azure, silver lining: the UK’s mobile networks will have more headroom to expand mobile penetration.
Despite growing interest in AI-driven underscore the urgency for stronger Google Cloud) and private on-premise demonstrated improved performance network capacity where needed. In the
automation, core network signaling defenses across the technology board. cloud deployments provide flexibility to over the year, as reflected in several
remains too complex and high-stakes The complexity of multi-protocol envi- address regulatory and performance key indicators. Speedtest Intelligence
for immediate AI adoption. Analysts like ronments spanning 2G to 5G amplifies concerns. This transition aligns with data revealed an improvement in
Heavy Reading note “AI in telecoms is these risks, necessitating in-depth operator goals of reducing hardware median market-wide latency to 51.8 ms,
advancing, but practical, carrier-grade measures like advanced Security Edge dependency and aligning network op- reflecting progress across three of the
solutions for network automation, Protection Proxy (SEPP) capabilities, erations with modern IT practices. four operators. Moreover, the observed
quality of experience (QoE) for bread-
and-butter activities such as gaming
BY STEVEN VAN ZANEN
and video streaming moved in the right
Steven van Zanen is the Chief Marketing Officer direction again after a decline last year.
(CMO) and Head of Product Management at
BroadForward. He has extensive experience in the 2025 is a new year as the UK continues
telecommunications industry, having led multiple its attempt to rejoin the global telecoms
innovations in areas such as signaling interworking, race, and with the UK’s largest tele-
mobile broadband, messaging, and mobile advertis- coms merger in decades, significant
ing. Before joining BroadForward, Steven held several developments are up ahead.
senior management positions, including Head of
Mobile Broadband and SVP Strategy and Marketing
at Acision.

160 161
The Fast Mode Trends and Outlook 2025

This could alter the structure of the It looks likely that 2025 will herald the with numerous 5G SA deployments in
industry in the long term and change arrival of D2D services in the UK. In dense urban areas in cities and towns.
the cadence of equipment upgrades fact, Ofcom is developing a framework
as mobile operators calibrate their to authorise and facilitate D2D services, EE has led the way in restricting access
capital cycles, spectrum demand, and with a consultation scheduled for the to its 5G SA network to more expensive
technology choices to reflect a marked first half of 2025. It will also review its tariffs in an effort to upsell its base and
slowdown in marginal capacity needs. approach to the mobile satellite service drive ARPU growth, providing addition-
(MSS) licences in the 2 GHz band, as al services such as Google’s Gemini
#3: THE UK’S PRIVATE NETWORK ECOSYSTEM the current EU-wide licences are set to Advanced AI model and a ‘Network
WILL CONTINUE TO GROW expire in 2027, allowing the UK to inde- Boost’ feature for priority access during
pendently determine their future use. congestion. As 5G SA rollouts mature,
Further growth in the UK’s private net- the emphasis on bundling is expected

FIXED NETWORKS
work ecosystem is expected in 2025 as Regarding potential partnerships, EE is to grow as operators face challenges
Ofcom continues to support the sector. the most likely potential candidate for in convincing consumers to opt for
The removal of the requirement to a Direct-to-Cell (D2C) partnership with higher-priced tariffs solely for access to
individually register end-user devices SpaceX’s Starlink, while Vodafone has 5G SA. Many operators will shift their
for low-power use in shared bands, been flirting with AST SpaceMobile. focus to the enterprise segment, where
increased availability of medium-power The initial wave of D2C solutions in the 5G SA features like network slicing may
licenses, and the launch of a compre- UK is expected to mirror the approach offer comparatively more potential for
hensive spectrum mapping tool for the of One New Zealand, the first global long-term monetisation. 1. How Fiber Innovation is Key to Unlocking AI Data Center Network Growth������������������ .164
3.8-4.2 GHz band were key milestones operator to launch nationwide services
in this support last year. at the end of 2024, initially focusing on Conclusion 2. The Future of AI and Fiber: Five Key Trends Shaping Telecommunications�������������������� .166
basic text messaging.
These measures contributed to the The creation of three fully converged
3. Connectivity and Telecom Trends to Watch for in 2025����������������������������������������������������������������������� .169
provision of 113 new shared access #5: OPERATORS SEEK MONETISATION operators in the UK through the Voda-
licenses between July and November, THROUGH BUNDLING AS CONSUMER fone/Three merger provides an exciting
4. An Evolution to a Revolution – How AI is Redefining Ethernet’s Role in the Digital
including 45 in the 1800 MHz band and DEMAND FOR 5G SA PRESENTS A CHALLENGE market backdrop in 2025, setting the
68 in the 3.8-4.2 GHz band. Looking stage for a renewed focus on conver- Age�������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������.172
ahead, Ofcom will enable low-power While the UK lags behind its developed gence and network quality. Private
indoor access in the 2.3 GHz band, peers in mobile network performance, network growth, preparations for the
signalling an intention to grow the UK’s it has emerged as a global leader in the arrival of D2D services, and continued
private network ecosystem in 2025. commercial rollout of the standalone progress in the 3G sunset indicate an-
5G architecture. It remains the only other year of significant change in store
#4: BURGEONING SATELLITE OPERATOR European country with three com- for UK telecoms.
PARTNERSHIPS WILL BRING DIRECT-TO- mercially available 5G SA networks as
DEVICE (D2D) SERVICES VMO2, EE, and Vodafone end the year

BY LUKE KEHOE

Luke Kehoe is an Industry Analyst at Ookla. Luke

leads Ookla’s research and thought leadership efforts
in Europe. An electronic engineering alumnus of
University College Dublin, Luke has extensive experi-
ence collaborating with mobile operators, telecoms
vendors, and government agencies in research and
advisory roles across Europe.

162 163
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

HOW FIBER INNOVATION IS KEY TO UNLOCKING AI AI enabled data centers require over
10x more optical fiber than traditional
My customers are constantly asking
how to adapt their networks for long-
#3: INTERCONNECTED DATA CENTERS AND
LONG-HAUL CONNECTIVITY WILL FUEL THE
FUTURE OF AI

DATA CENTER NETWORK GROWTH

data center networks – often, more term growth. To meet new computing
fiber is incorporated into pre-existing needs, customers are creating multi-
spaces. As the back-end networks year plans to stay ahead of the curve, Speed and scalability aren’t the only
that enable generative AI continue to and often approach us looking for challenges brought on by the data
evolve, they will require even more solutions to help adapt their networks demands of AI. In 2025, I expect to see
Written by Michael Crook graphical processing unit (GPUs) for growth. Compared to traditional operators engage in different connec-
Data Center Market Development Manager, Corning Optical Communications clusters and fiber rich interconnects cabling methods, preconnectorized tivity methods to expand network ca-
to accommodate more complex and or “plug and play” solutions including pacity, including long-haul connectivity
data intense computations – and space MMC connectors and distribution sys- between data center campuses.
optimization is key for keeping up with tems can drastically increase deploy-
growing demands. ment speeds. Long-haul connectivity will prove
itself to be an alternative solution to
To support the volume required to Data center operators all have their overcome access to power and land
power AI, data centers will need to own unique plans for how they’ll constraints.
leverage fibers with low attenuation scale over time. For those that plan to
and exceptional bend performance. increase data processing capabilities Interconnected data centers allow
With these features, operators will within their data center, having a struc- operators to increase network capacity
more easily manage cable congestion, tured cabling system that can support and transmit large amounts of data
particularly where they’re retrofitting multiple generations of optical trans- over extended distances – eliminating
an existing facility. As networks evolve, ceivers is critical. geographical constraints. By using
operators will need to leverage smaller, long-haul fiber solutions, data cen-
bendable fiber to optimize the optical Cable and fiber flexibility is also crucial ter operators can build campuses in
footprint within server rack space. for maximizing space and computa- unconventional locations and access
tions for a considerable amount of data. the valuable resources they need while
#2: DEPLOYMENT SPEED AND SCALABILITY It offers improved control over signal increasing network density. By utilizing
WILL REMAIN CHALLENGES loss and facilitates connections over long-haul fiber and cable solutions,
large networks. operators have the ability to establish
As bandwidth needs grow, two chal- campuses in non-traditional places.
lenges that will continue to persist As operators find methods to build and
are deployment speeds and network activate AI networks to accommodate In 2025, I expect to see new fiber prod-
scalability. Network needs and technol- computing needs, bendable high fiber ucts, practices, and network providers
ogy changes are happening at a rapid count cables and preconnectoized make further strides to accommodate
pace, and organizations will need to solutions will continue to be sought-af- the increasing demands of AI.
quickly adapt network componentry to ter optic tools.
keep up.

BY MICHAEL CROOK
The rise of artificial intelligence (AI) of AI, data centers will need to design #1: SPACE OPTIMIZATION WILL REMAIN
and machine learning (ML) have put a and implement dense, efficient, scal- CRITICAL FOR HYPERSCALERS Michael Crook is a Data Center Market Development
spotlight on data centers and the im- able, and savvy networks. As we head Manager at Corning Optical Communications. He
portant role they play in our daily lives. into 2025, fiber innovation, scalability, While operators mapped out plans and supports our hyperscale, multi-tenant, and enterprise
Now, more than ever, data centers are and density will be key to accelerat- design configurations in 2024, 2025 customers with new fiber optic innovations and com-
critical to hold a vast amount of digital ing growth as data centers adapt to will be all about action. Next year, we’ll mercial solutions. With over 15 years of experience,
information to enable web content, support AI. see hyperscalers move from planning Michael has amassed a great deal of knowledge
streaming services, and cutting-edge and design phases to implementa- related to designing and building fiber optic network
AI use cases. Here are three key predictions on what tion phases as they activate their AI infrastructures for data center and carrier environ-
data centers will need in order to thrive networks. ments.
To maintain the growth and evolution in the year ahead:

164 165
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

THE FUTURE OF AI AND FIBER: FIVE KEY TRENDS unpredictability will lead to winners
and losers in the telecommunications
landscape of AI. The efficiencies AI can
bring—streamlining operations and re-
en workloads. The bottom line: AI is
not just a software revolution; it is a

SHAPING TELECOMMUNICATIONS
market. Fiber-to-the-home (FTTH) ducing costs—are immense. But that’s physical infrastructure revolution, and
providers, who are heavily dependent not why China is racing to dominate fiber networks must rise to meet the
on their working capital being sourced AI. They see it as a tool for power, one challenge.
through the federal government, will that can be used to disrupt the critical
struggle. The winners in this environ- infrastructure of its “so-called” com- #4: FIBER’S ROLE IN AI SCALABILITY
Written by Bill Major ment are going to be organizations that petitors. The cybersecurity implications
CEO, FiberLight can execute and have more financial are massive, which is why the United As AI workloads increase, fiber net-
flexibility and access to working capital. States must continue investing heavily works must keep pace. The question
in AI to maintain its leadership position. is not whether fiber can handle AI de-
In the mid-20th century, the space race There’s a lot of talk about funding Fiber networks will be critical in sup- mands, but how networks will need to
was more than a battle for dominance being redirected towards satellite and porting this technological race, ensur- evolve. History has shown that connec-
over Earth’s atmosphere—it was a wireless solutions. That’s a win for fiber ing secure, high-capacity connectivity tivity technologies advance, but fiber
geopolitical contest that captivated companies since these technologies to power AI’s future. has remained a constant.
imaginations and fueled awe-inspiring still depend on high-capacity fiber net-
technological advancement. As the U.S. works. Enterprise fiber is now needed OpenAI has also made headlines with From early T1 connections in the 1990s
and the Soviet Union competed to see more than ever. Regardless of where its groundbreaking advancements, to the massive data throughput de-
who would be the first to put a man the funds ultimately go, fiber remains but let’s be clear: they’re still reliant mands of today, fiber has consistently
on the moon, they inadvertently laid the backbone of connectivity, ensuring on Nvidia chips and are. Microsoft, adapted. AI and quantum computing
the groundwork for innovations that that even next-generation wireless and Google, and Amazon have no intention will push fiber infrastructure to new
would define our modern digital age. satellite solutions rely on strong fiber of slowing down either—they will con- limits, requiring expanded network
From satellite communications to GPS, infrastructure. Investments in fiber not tinue doubling down on AI investment topologies, increased redundancy, and
the technological leaps driven by that only support AI-driven advancements this year. This relentless momentum faster data transmission speeds. The
era still shape our world today. Popular but also lay the foundation for long- underscores the need for a robust fiber current boom in data center con-
culture at the time reflected this ex- term economic resilience and digital infrastructure to support AI’s exponen- struction is a direct response to these
citement, with shows like The Jetsons transformation. tial growth. demands, with hyperscalers investing
envisioning futuristic concepts such in fiber routes designed to bring com-
as smart homes, 3D printed food, and #2: THE GEOPOLITICAL RACE TO AI #3: SAUDI ARABIA’S $600 BILLION BET ON puting power closer to the edge.
video calling that were once a fantasy SUPREMACY U.S. AI
and now an everyday reality. Today, While some argue that satellites will re-
a similar dynamic is unfolding with Talk to any web scaler or hyperscal- AI investment isn’t just coming from place terrestrial fiber, the reality is that
artificial intelligence (AI), where global er and there’s no doubt about it: the governments like China or the usual satellites cannot provide the same level
competition drives rapid innovation, global AI arms race is well underway. commercial players like Microsoft, of security, reliability, or bandwidth that
reshaping how we connect, communi- With China ramping up AI investment Google, and Amazon. Recently, Saudi fiber can. Just as The Jetsons gave me
cate, and build the future. unprecedented ways. Here are five key areas, BEAD was created with the and companies like DeepSeek making Arabia committed $600 billion to the hope that one day I’d grow up to drive a
trends that will shape the future of AI promise of driving economic growth waves, the race is still in its early stag- United States, partly to fund AI devel- flying car—an idea that remains largely
In today’s telecommunications industry, and fiber in the coming year and into – enabling businesses to expand, es—but the stakes have never been opment. Crown Prince Mohammed bin out of reach—some view satellites as
we find ourselves in a new race—not the future. communities to thrive, and innovation higher. DeepSeek’s launch signals a Salman’s pledge signals a significant the inevitable future of connectivity. But
to space but to AI supremacy. Nations to accelerate. Since it was signed into challenge to rival U.S. dominance, with shift in global investment, with Saudi the truth is, just as roads remain essen-
and corporations alike are investing #1: THE UNCERTAINTY OF BEAD FUNDING law, its rollout has been plagued by a promise of offering cutting-edge Arabia choosing to develop AI in the tial for transportation, fiber remains the
billions in AI research, infrastructure, AND ITS IMPACT ON INFRASTRUCTURE bureaucratic hurdles. capabilities at a fraction of the cost. U.S. rather than at home. backbone of digital infrastructure. As AI
and applications, recognizing that continues to develop, fiber will be the
leadership in AI is not just an economic Hailed as a transformative initiative to With a new administration in Wash- AI is no longer just a commercial This influx of capital is a game-chang- most critical component of next-gen-
advantage but a strategic necessity. bridge the digital divide, the Broad- ington, DC, the future of BEAD funding endeavor; it is a geopolitical priority. er. As AI infrastructure expands, eration connectivity, ensuring the high-
Just as the space race reshaped global band Equity, Access, and Deployment remains uncertain. Potential freezes The wars of the future won’t take place the demand for high-capacity fiber speed, low-latency networks needed to
power structures and spurred techno- (BEAD) program was established on federal grants and loans have cast on traditional battlefields—winning will networks will skyrocket. Data centers, power the future.
logical progress, the AI revolution is in 2021. With more than $42 billion doubt on state broadband plans, rais- come down to who gets to market first edge computing, and fiber-optic routes
redefining the landscape of connectiv- earmarked for providing reliable high- ing concerns about the long-term sta- and who controls the cybersecurity will need to scale to support AI-driv-
ity, requiring fiber networks to scale in speed internet access to underserved bility of infrastructure investment. This

166 167
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

#5: THE FUTURE OF FIBER DEPLOYMENT:

LONG-HAUL EXPANSION AND METRO
DENSITY
hyperscalers are driving this demand,
investing heavily in long-haul fiber
The bottom line: execution will
define success CONNECTIVITY
AND TELECOM
routes that provide connectivity to un-
derserved regions. These deployments The future of AI and fiber is not about
When was the last time a fiber com- create new opportunities for business- speculation—it is about execution.

TRENDS TO
pany truly built long haul fiber? The es, allowing them to reach previously Those with access to capital, strategic
answer: it’s been a long time because, inaccessible markets and build out vision, and the ability to deploy fiber
for years, fiber expansion has been fo- metro fiber networks that support local efficiently will thrive. The telecommu-

WATCH FOR
cused on urban and suburban deploy- AI applications. nications industry is at a turning point,
ments, with little investment in long- and the next few years will determine
haul routes. That’s now changing as At the same time, metro fiber deploy- who leads and who gets left behind.

IN 2025
companies like Zayo have announced ments are becoming increasingly
plans for new long-haul fiber builds, dense. The race to the edge is real, and As AI reshapes the world, fiber remains
signaling a shift toward more diverse those who can build the most efficient, the foundation of digital progress. The
and resilient network infrastructure. high-capacity fiber networks will be the companies that recognize this reality
winners. This shift will define the next and invest in building the networks of
Long-haul fiber is not just about decade of fiber infrastructure develop- the future will not just survive—they will Written by Purva Rajkotia
connecting cities—it is about enabling ment, and the companies that execute define the next era of connectivity. Director, GBSI and the Connectivity and Telecom Practice Lead, IEEE SA
AI, cloud computing, and next-gen- effectively will emerge as industry
eration applications. Web scalers and leaders.

Technology innovators continue to feed As an example, we could see increased smart devices and applications. To pre-
BY BILL MAJOR the world’s craving for connectivity, de- use of holographic telepresence. The vent Wi-Fi from becoming a bottleneck

Bill Major is the CEO of FiberLight, a leading livering more connected devices, more more robust 6G should be powerful in the end-to-end user experience,

provider of mission-critical, high-bandwidth fiber users, richer content, and new appli- enough to enable these digital twins service providers can adopt the latest

infrastructure. With over 25 years in the telecom- cations. At IEEE Standards Association and the related audio to be project- standards, such as IEEE 802.11be™

munications industry, Bill has been at the forefront (IEEE SA), we’re continuously work- ed across a 6G network in real-time. (also known as Wi-Fi 7), which improve

of fiber network expansion, helping businesses and ing to understand and support these The use cases leveraging holographic throughput and latency. IEEE 802.11be

communities stay connected in an ever-evolving emerging technologies. Here are a few telepresence are endless, including also supports connectivity using 6GHz

digital landscape. of the trends we’re tracking: cases for good. Imagine a surgeon in license-exempt spectrum, which
a remote area leveraging the guidance enables higher speeds in high-density
#1: DEEPER, RICHER USER EXPERIENCES of a lifelike hologram surgeon, who is environments.
actually thousands of miles away.
6G technology has already demonstrat- #2: MORE POWERFUL DEVICES ARE NEEDED
ed the ability in tests to transmit data 6G can also create new virtual oppor-
at rates up to 1 Tb/sec compared to tunities in e-commerce, events, and With higher data rates, however, comes
5G’s 1 Gb/sec. top speed. The signifi- many other areas and facilitate the use the need for more powerful devices
cant increase in potential data speed of AI in networked applications. with substantially more computing
is expected to foster compelling new power to process much higher data
use cases, something that 5G failed to To use these robust applications, fast flows without overheating or running
deliver. connectivity must be provided to the out of battery power.
masses. The rapid adoption of ultra-
6G technology has the potential to high-speed broadband technologies Existing devices have only a limited
enable new applications and ways such as FTTH can enable reliable ability to provide these types of expe-
of doing things, providing users with multi-gigabit speeds to homes and riences and to work with AI technolo-
richer, fulfilling experiences. The ability businesses. Wi-Fi is essential for ex- gies. More powerful processors, faster
to deliver these experiences will thus tending these benefits throughout the and larger memories, improvements
open doors for new use cases for both premises of homes, offices, and facto- in RF components, and stronger,
consumers and enterprise. ries, to ensure seamless, high-speed longer-lasting batteries are needed,
connectivity and enable a vast range of along with effective ways to keep these

168 169
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

devices from overheating. #4: SUSTAINABLE INFRASTRUCTURE IS A KEY Until now, nearly all AI processing has home gateways. Cybersecurity is increasingly more nection program. The program’s goal
GOAL been done at data centers. The data important across all of the functions is to build a community to discuss cy-
IEEE SA has numerous standards-re- center industry is racing to not only This trend can be seen in two key and use cases for connectivity and bersecurity issues and rethink architec-
lated initiatives and activities under- The United Nations has stated that keep up with demand but also to host application areas: telecom systems. tures to address critical market needs.
way to understand and define these access to the internet is a basic hu- systems capable of supporting the • AI for Network Optimization and
requirements within the 6G Technology man right, and one of the sustainable stringent demands of AI. There are Enhanced User Experience: AI Unfortunately, the reality is that most of Cybersecurity is likely to remain a ma-
for Connectivity & Telecom Systems development goals in its 2030 Agenda many advantages to centralized AI systems can enable superior user our identity is already on the internet, jor trend going forward because new
Industry Connections program. IEEE for Sustainable Development is that processing, but there are also disad- experiences by anticipating more requiring vigilant monitoring at the technologies present new challenges.
SA’s Industry Connections programs universal and affordable access to the vantages. Data transmissions always efficient traffic prioritization and enterprise level but also the personal And as cyber criminals get better at
are collaborative frameworks where internet be available everywhere. run the risk of bottlenecks. There is also management as well as enabling level. But with the advent of the emerg- what they do, including their use of
interested stakeholders collaborate on the threat of a cyber hack to consider, proactive network troubleshooting. ing technologies we’ve discussed, AI to hack into systems, cybersecuri-
rapidly changing technologies. IEEE SA launched an initiative called which would result in the loss or cor- • Edge AI Applications: AI our personal information could soon ty technology must stay a few steps
IEEE Planet Positive 2030, a collabora- ruption of valuable information. Distrib- capabilities at the edge, enabled expand to include holographic imag- ahead.
#3: EXPANDING CONNECTIVITY TO MORE tive, multidisciplinary effort to develop uted processing reduces these risks at by advancements in mobile es, voice, etc. Unless and until better
PEOPLE pragmatic, sustainable technology the data center, but are we moving the devices and home gateways, can cybersecurity measures are developed Get engaged with IEEE SA
solutions to lessen humanity’s im- risk from one location to another? present new use cases. Edge AI and put in place to ensure that our data
Although the world is more intercon- pact on the planet. It builds on IEEE’s applications entail combining edge won’t be hacked by third-party vendors Through our Connectivity & Telecom
nected than it was just a few short experience in identifying and judging Each approach will make use of AI and cloud computing to deliver or misused, people will not trust or use Practice, IEEE SA is building a global
years ago, 32 percent of the world’s the potential positive and negative im- functionality in different ways. Data low-latency, private, seamless, these systems. ecosystem of stakeholders to address
population still doesn’t have internet pacts of how technologies are applied. centers will use AI to learn, anticipate, and personalized AI-powered the need for robust, responsible, and
access, according to the ITU. Telecom You can download a newly published and direct the behavior of devices applications for smart homes and Cybersecurity with regard to data affordable wired/wireless platforms
infrastructure is expensive to build and guidebook with more information. It is and networks, while edge-connected enterprises. processing will be a critical part of that can provide improved and reliable
maintain, requiring users to pay for it. part of the IEEE’s overall Sustainable devices will use it locally for real-time the IEEE Connectivity and Telecom connectivity.For more information or to
Lack of connectivity in many remote Infrastructures and Community Devel- processing and inferencing. #6: CYBERSECURITY: A CRITICAL practice in 2025, with initiatives such as join the standards activity, please visit
and/or sparsely populated rural areas opment program, which brings togeth- REQUIREMENT the Cybersecurity for Next Generation the Connectivity & Telecom Practice
is simply because telecom companies er multidisciplinary teams to identify AI is rapidly evolving to include the Connectivity Systems Industry Con- webpage.
don’t forecast an adequate return on sustainable and practical technology emergence of agentic AI, which will
investment. solutions. undoubtedly have a role in the future
of connectivity and telecommunica-
However, internet connectivity would #5: AI FOR BETTER, FASTER, MORE EFFICIENT tions. Agentic AI is a software system
provide these populations with more DATA PROCESSING combining automation with large
options to achieve economic self-suffi- language model (LLM) capabilities to
BY PURVA RAJKOTIA
ciency and better quality of life, among AI approaches are likely to be used for create AI agents that can perform tasks
other benefits. As an example, connec- virtually all functions within connec- independently, including the ability to Purva Rajkotia is the Director of Global Business
tivity can open up households to use tivity and telecom systems to achieve adapt in real-time and solve complex Strategy & Intelligence (GBSI) and the Connectivity
home healthcare devices for monitor- greater efficiencies, speed and reduced problems based on context. Agentic and Telecom Practice Lead at IEEE SA. Prior to IEEE,
ing and diagnosis, a tremendous asset costs. But it remains unclear how AI AI is like having digital assistants on Purva held leadership positions with Qualcomm,
for rural communities lacking adequate will impact data processing architec- steroids that can take initiatives, solve Samsung, and Disney in various capacities. Purva
healthcare facilities and providers. tures. complex problems, and adapt their also held leadership positions in various standards
approaches based on new or changing organizations such as ITU, 3GPP, 3GPP2, CENELEC, etc.
6G technology could be one way to For example, where will the data circumstances.
help bridge this digital divide, which is processing actually happen? Will it be
why IEEE SA formed the 6G Rural Con- primarily at centralized data centers, Finally, AI is also beginning to have
nectivity and Intelligent Village Industry or will AI processing be more of a dis- a growing influence in shaping the
Connections program. Its goal is to tributed function on the client (device) user experience for IEEE 802.11™ Wi-Fi,
investigate, understand, and define the side as well? In other words, if you substantially broadening the array of
challenges, opportunities, and poten- engage your handheld mobile device use cases and consumer applications.
tial solutions that may help provide (a more powerful one than currently on This is driven by the advancement of AI
sustainable, reliable, and affordable the market), will the device crunch the capabilities in edge devices, especially
connectivity to underserved areas. data, or will it happen via the cloud? mobile devices, access points, and

170 171
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

AN EVOLUTION TO A REVOLUTION – HOW Consider NVIDIA, whose stock price

nearly tripled in 2024, propelled by
sortium and UALink Consortium, new
industry groups are emerging to tackle
ployment needs. With soaring demand
and increasing projected volumes,

AI IS REDEFINING ETHERNET’S ROLE IN

a relentless and seemingly insatia- these hurdles. the industry must swiftly adopt more
ble appetite for AI-focused products. efficient methods to satisfy divergent
Hyperscalers are projected to spend #2: WIDE OPEN THROTTLE: ACCELERATING requirements.
ETHERNET’S EVOLUTION

THE DIGITAL AGE

some $180 billion on data center capital
expenses, predominantly for acceler- #4: MAPPING OUT ETHERNET’S BRIGHT
ated GenAI-accelerated servers. The While the IEEE P802.3dj 200 Gb/s, 400 FUTURE
surge is transforming resource priori- Gb/s, 800 Gb/s, and 1.6 Tb/s Ethernet
ties, for example, the ability to access, Task Force is targeting 200G per lane AI networks are moving to Ethernet
Written by Peter Jones, reserve, and deliver power. It’s one of Ethernet by late 2026, key stakehold- faster than ever, with hyperscalers
Chair, Ethernet Alliance the driving factors behind Microsoft’s ers are already clamoring for 400G leading the charge. Their need for
20-year contract to help restart the per lane. The need for faster Ethernet technology and supplier diversity is
Three Mile Island nuclear plant, as well speeds is intensifying, even as mani- terraforming the Ethernet landscape,
as Amazon’s and Google’s investments fold technology complexities continue giving rise to new technical and busi-
in small modular reactors (SMRs) to to rise. In addition to more speed, ness challenges. But as the industry
feed their power-hungry data centers. customers are desperate for significant charts Ethernet’s bright future, one
Water, another critical consideration, reductions in relative network energy thing remains abundantly clear: failure
adds further complexity as outlined in consumption. Why? Because every is not an option.
How data centers use water – and why watt saved on a networking device
kicking the habit is nearly impossible. is one that can be spent on compute #5: ETHERNET: ON THE FAST TRACK TO 400G,
workloads instead. 800G, AND BEYOND
#1: THE NEXT BIG THING: NAVIGATING
TOMORROW’S AI CHALLENGES TODAY #3: POWERING PROGRESS: DIVERSE The race to 400G per lane is well
APPROACHES FOR ETHERNET’S NEXT ERA underway, but the journey is far from
So, just how will all this AI-induced over. Industry players are under im-
disruption affect Ethernet? During the Energy efficiency and innovation are mense pressure to deliver, but success
Ethernet Alliance’s Technology Explo- at the forefront of Ethernet’s continued demands more than just higher speed
ration Forum: Ethernet in the Age of evolution. Breakthroughs like Meta and – it also requires measurable power
AI (TEF 2024)), industry leaders from Microsoft’s collaboration on the Mt. efficiency and reliability improvements.
Google, Meta, Microsoft, and beyond Diablo disaggregated power platform, Scale-out networks are gearing up
addressed challenges ranging from which leverages 400V DC to provide for 800G, and eyeing 1.6T and 3.2T
optical module design issues to hyper- upwards of 100kW per rack, are revolu- technologies to follow. Investing billions
scalers grappling with deployment ob- tionizing power delivery. This paradigm in data center infrastructure and GenAI
stacles. For example, link and module shift is, in turn, expediting the adoption silicon demands a frictionless network
failures in training clusters resulting in of “direct-to-chip” (DLC) liquid cooling, that keeps servers operating at peak
lost operational and opportunity costs significantly improving thermal efficien- performance round the clock. Higher
underscore hyperscalers’ preference cy for high-power racks. link speeds and lower link numbers
for DAC and AEC copper intercon- are critical to achieving this seamless
nects, valued for their reliability, power At the same time, hyperscalers are efficiency.
efficiency, and overall cost-effective- taking on shared challenges using a di-
ness. versity of approaches. Google’s custom #6: OVERCOMING LAYER 0 ROADBLOCKS
optical switches and Meta’s rail-only
Currently, NVLink™ and Infiniband™ design underscore the principle that Physics is tough, especially the foun-
dominate scale-up and scale-out net- “one size never fits all”, as RFC 1925 – dational Layer 0 challenges of power,
The AI era – and Generative AI (GenAI) GenAI’s emergence, hyperscalers were heights, fueled by a mind-boggling works, however, hyperscalers are eager The Twelve Networking Truths’ Rule 10 cooling, and water. And no matter how
in particular – is upon us, and its arrival already pushing for an acceleration in volume of new investments. for Ethernet-based solutions facilitating so aptly states. These varied strategies many AI chips you deploy, without reli-
is scrambling the Ethernet landscape the pace of technology development. greater technology and supplier diver- are impelling the development of spe- able power delivery and effective heat
at an unprecedented rate. Even before Now, this demand has reached new sity. Led by the Ultra Ethernet Con- cialized solutions tailored to unique de- management, even the most advanced

172 173
The Fast Mode Trends and Outlook 2025

chips will fall short. Unlocking GenAI’s Rule 12 that states, “In protocol design, in an ever-evolving landscape.
full potential means tapping into inno- perfection has been reached not when
vative solutions like DLC liquid cooling, there is nothing left to add, but when Peter Jones is a Distinguished Engi-
geothermal energy, and SMRs. there is nothing left to take away”, Eth- neer in Cisco’s Networking Hardware
ernet is poised to gracefully navigate team, chairs the Ethernet Alliance, and
Always bet on Ethernet the complexities of the AI era, continu- is active in IEEE 802.3. Peter works
ing its legacy of reliable, scalable, and on system architecture and standards
Ethernet has consistently outperformed efficient connectivity. strategy across Cisco’s networking
countless competitors to become the portfolio and was deeply involved
world’s dominant wired networking As the demand for faster speeds and in development of the Catalyst 9000
technology. Its simple yet powerful val- greater efficiency intensifies, Ethernet switches and the switching ASICs they
ue proposition – “It just works” – has all stands as the safest and smartest bet use. Peter is passionate about Network

QUANTUM
but guaranteed its enduring success. for the future of networking, offering Evolution and Adoptable Technology.
Guided by the wisdom of RFC 1925, unmatched stability and dependability

BY ETHERNET ALLIANCE

The Ethernet Alliance is a global consortium that

includes system and component vendors, industry
1. Quantum Computing’s Next Frontier: 5 Networking Trends to Watch in 2025��������������.176
experts, and university and government profession-
als who are committed to the continued success
2. Trailblazers in Quantum-Safe Networks: Which Industries Are Paving the Way?�������.178
and expansion of Ethernet technology. The Ethernet
Alliance takes Ethernet standards to market by sup-
porting activities that span from incubation of new
Ethernet technologies to interoperability demonstra-
tions and education.

174 175
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

QUANTUM COMPUTING’S NEXT #3: CONSOLIDATION AND INTEGRATION OF

QUANTUM NETWORK DEVICES
QUANTUM NETWORKS
•
quantum communication.
Spin-Photon Transducers: These

FRONTIER: 5 NETWORKING
2025 will see the operationalization of allow for the conversion between
The physical architecture of quantum the first hybrid homogeneous quan- spin states of electrons and
networking is undergoing significant tum networks, offering a multi-modal photons, essential for integrating

TRENDS TO WATCH IN 2025

transformation. The push towards more approach to quantum computing: different quantum systems.
compact, rack-mountable quantum • Interoperability: Different • Ion-Photon Transducers:
network devices is notable: quantum computing technologies By managing the interface
• Smaller Power and Space like Superconducting Qubits, between ions and photons, these
Footprints: By minimizing the Photonic Qubits, Trapped Ions, transducers facilitate the transfer
Written by Michael Wood size and energy consumption of Neutral Atoms, and Silicon Spin of quantum information across
Chief Marketing Officer, Aliro Quantum quantum networking devices, more Qubits can now work in concert, different physical systems.
of these can be integrated into data enhancing the flexibility and scope
centers or cloud facilities. This not of quantum applications. Conclusion
only reduces the physical footprint • Scalability: By networking various
Quantum computing has leaped reliability of quantum algorithms. but also allows for a denser quantum computing paradigms, In 2025 the synergy between quantum
forward in the last decade, setting the • Higher Entanglement Success: deployment of quantum network these hybrid systems can scale computing and quantum networking
stage for monumental breakthroughs Entanglement is the cornerstone infrastructure. more effectively, potentially leading will become increasingly evident. The
in 2025, particularly through the advent of quantum computing power. • Enhanced Networking to quantum computers that are trends outlined above not only promise
of quantum networking. Here’s a Improved error correction will Capabilities: With these devices both more powerful and versatile. to refine existing quantum technolo-
detailed look at the top five trends that lead to more stable entanglement, optimized for integration, the gies but also to expand the horizons
are steering the quantum computing enabling quantum networks to connectivity and interoperability #5: DEVELOPMENT OF QUANTUM of what’s computationally possible.
landscape towards these significant handle more sophisticated quantum between various quantum network TRANSDUCERS Quantum networking is not just about
advancements: states. components are significantly enhancing current quantum computers
enhanced. This development The challenge of maintaining qu- but is fundamentally about creating
#1: QUANTUM NETWORKING ERROR #2: ADVANCEMENTS IN SIMULATION is crucial for achieving robust bit states across different quantum a more interconnected and efficient
CORRECTION CAPABILITIES quantum networks that can support architectures is being addressed with quantum ecosystem. This year will like-
complex quantum communication innovative transducer technology: ly be remembered as a pivotal moment
Quantum networking error correction The ability to simulate large-scale and computation tasks, paving the • Microwave-to-Optical where quantum computing steps out of
is pivotal in pushing the boundaries of quantum networks has been a bottle- way for true quantum advantage in Transducers: These devices the lab and into real-world applications,
what quantum computers can achieve. neck in quantum research. However, network applications. convert quantum information from thanks to these networking innova-
This trend focuses on reducing qubit 2025 marks a year where simulation microwave to optical domains, tions.
error rates, enhancing the fidelity of capabilities will significantly expand: #4: EMERGENCE OF HYBRID HOMOGENEOUS which is crucial for long-distance
quantum operations, and increasing • Increased Number of Component
the success rate of entanglement be- Models: More detailed and varied
BY MICHAEL WOOD
tween qubits. By improving the quality models of quantum components
of connections between quantum allow for simulations that more A Silicon Valley luminary, Michael Wood has
computers, we can expect: accurately reflect real-world spearheaded marketing and product innovation at
• Lower Qubit Error Rates: Through quantum network behaviors. tech giants like Cisco, VMware, and Aliro. His tenure
advanced error correction codes, • Improved Hierarchical Modeling: at Aliro as CMO has him at the cutting edge of ultra
quantum networks can maintain This allows for simulations at secure quantum networking. With a knack for scaling
the integrity of quantum information different scales, enhancing our startups to multi-million dollar successes, Michael’s
over longer distances and times, understanding and optimization of strategic vision has redefined cybersecurity, SaaS,
which is crucial for scaling up quantum systems. and cloud solutions.
quantum computing systems. • Leveraging Cloud Compute
• Increased Fidelity: Higher fidelity Systems: The integration of cloud
in quantum operations means that computing resources means that
quantum networks can perform simulations can be run faster and
more complex computations with at a larger scale, reducing time-to-
fewer errors, directly impacting the discovery in quantum research.

176 177
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

TRAILBLAZERS IN QUANTUM-SAFE NETWORKS: asymmetric cryptography and recom-

mended that government agencies use
works may be furthest ahead in their
awareness and adoption of QSN,
possibly leading to more QSN trials.
Big banks and large financial services

WHICH INDUSTRIES ARE PAVING THE WAY?

symmetrically encrypted networks to mission-critical infrastructure operators institutions, trading organizations and
transmit sensitive data. Other agencies in other industries, like finance, utilities, many more are particularly vulnerable
around the world have since published transportation, and more are likely to to “harvest now, decrypt later” (HNDL)
their papers and reports on quantum make significant strides in quantum attacks, in which a threat actor har-
security, including France’s National security in 2025. vests sensitive information today, hold-
Written by Martin Charbonneau, Head of Quantum-Safe Networks, Nokia and Cybersecurity Agency (ANSSI). ing it until it can be decrypted later by
James Knight, Director and Principal Architect, Kyndryl The reality is that many mission-critical a cryptographically relevant quantum
For reasons of scale and protection infrastructure operators can be under- computer (CRQC). These organizations
for temporary connectivity, such as at standably conservative when it comes need to act now to protect customer
the application layer, a quantum-safe to change. However, we are seeing information and vital economic data.
public key infrastructure will be re- an increase in government standards Investing in quantum-safe networking
quired. This triggered the development and directives around cybersecurity, to capabilities not only addresses current
of post-quantum cryptography (PQC), which these mission-critical infrastruc- regulatory requirements like the Digital
using mathematics-based asymmetric ture operators must respond. Operational Resilience Act (DORA) and
algorithms. In August 2024, NIST re- the Health Insurance Portability and
leased its first three finalized standards We’ve seen this with power utilities, Accountability Act (HIPAA), and effec-
for PQC, with Federal Information where some directives have emerged tively addresses critical aspects such
Processing Standard 203 being es- requiring data confidentiality, integri- as confidentiality, integrity and authen-
pecially relevant to applications. NIST ty and authenticity. For example, the ticity, but also prepares organizations
is encouraging the Information and North American Electricity Reliability for future regulatory developments.
Communication Technology industry to Corporation Critical Infrastructure Pro-
begin transitioning to the new stan- tection Standards (NERC CIP) includes The list does not stop there. Private
dards as soon as possible. That said, requirements for data security in US industries, manufacturers, research
it will take some time for PQC to be and Canadian power utility networks. organizations and more – anyone who
integrated into commercial application In 2021, NIST and NERC published uses and relies on data for their oper-
products, so full-scale adoption won’t guidelines on how utilities can map the ation will need enhanced data security
happen in the near-term. NIST Cybersecurity Framework to the and privacy protections that can be de-
NERC CIP regulations. ployed working with trusted partners.
PQC at the application layer is just
one of the tools that can help achieve Similarly, the European Union’s Net- #3: TELECOM SERVICE PROVIDERS ARE ALSO
a quantum-safe outcome. Ideally, work and Information Security (NIS) INNOVATING
physics-based symmetric pre-shared 2 Directive, which came into force in
key (PSK) cryptography, PQC and, 2023, sets out mandatory cybersecu- Some CSPs are recognizing the quan-
possibly, quantum key distribution rity risk management measures and tum threat and seeing QSN as a way
(QKD) are used together. These are reporting obligations for railways, en- to differentiate their service offerings in
applied across different application ergy providers, CSPs and other critical a highly competitive market. In March
Quantum computing will bring both use today. This includes common se- #1: LEADING THE WAY: DEFENSE and network layers to achieve robust, infrastructure providers. The require- 2024, Korea’s SK Telecom joined Nokia
challenges and opportunities, es- curity and privacy protections used for crypto-resilient defense-in-depth pro- ments of NIS2 have prompted many and six other companies to create a
pecially for industries that manage health records, financial transactions The defense and intelligence commu- tection. It’s like having a fence, a door organizations to embark on ambitious “quantum alliance” dedicated to explor-
mission-critical infrastructure. Yes, it and sensitive government information. nity recognized the threat posed by deadbolt lock and bright lights to keep modernization projects to enhance ing joint investment and deployment
will unlock new capabilities in com- While some industries have recognized quantum computers well before it was intruders out of your home. their cybersecurity. opportunities. In Korea, Nokia working
puting, networking and sensing. But the quantum threat and are already even on the radar of the commercial with SK Broadband enabled a quan-
when quantum computers advance far implementing quantum-safe network industry. In February 2016, the U.S. #2: INDUSTRIES POISED FOR A QUANTUM- The healthcare and finance sectors, tum-safe network deployed on behalf
enough (and when, not if, threat actors (QSN) solutions, others lag behind — National Institute of Standards and SAFE BREAKTHROUGH while already highly regulated in terms of the Korea Hydro and Nuclear Power
get access to them), if protections but even those behind may be poised Technology (NIST) released the first of data privacy and security, will likely (KHNG), while in Poland, Netia and
aren’t put in place now, they’ll also to make big leaps in their quantum draft of Interagency Report 8105, Re- While the defense and intelligence become increasingly aware of the Nokia successfully completed a pilot to
be able to crack data confidentiality, readiness in 2025. port on Post-Quantum Cryptography, sector and other government net- quantum threat in the years to come, test QKD for real-world data trans-
integrity and authenticity methods in which discussed the vulnerabilities of

178 179
The Fast Mode Trends and Outlook 2025

mission and quantum key generation retail services, bringing quantum-safe It’s vital to stay one step ahead. With
between two of the CSP’s live data network outcomes to even more enter- 2025 being the United Nations Year
centers. A similar trial was conducted prise segments. of Quantum Science and Technology,
in Belgium with Proximus, with Nokia and as the world’s industries continue
being the first in that country to pilot a #4: THE QSN OPPORTUNITY IS NOW with digital transformation and mod-
hybrid quantum encryption key in a live ernization projects as we progress into
optical network. The rapid development of quantum a Quantum Secure Economy, we have

INTERNET OF
computing, AI and other technology an opportunity to put in place quan-
These and other providers are working concurrently in a tense global en- tum-safe networks now that protect
to innovate. By combining leading net- vironment suggests a danger that against tomorrow’s threats.At this mo-
working technologies with expert part- mission-critical infrastructure operators ment, we have the advantage and the
ners who possess deep network and and service providers of all types need opportunity, so now is the time to act.

THINGS
IT knowledge across various industries, to protect themselves against.
CSPs can offer managed, wholesale, or

BY MARTIN CHARBONNEAU

With a career immersed in the evolution of emerging

1. Use of AI in Healthcare in 2025: Medical Miracle, or Medical Malpractice������������������������ .182
technologies, Martin Charbonneau pioneers
Quantum-Safe networking solutions at Nokia. His
2. The State of IoT: Transformative Momentum and What Lies Ahead���������������������������������������� .185
focus on Quantum-era cybersecurity fosters global
collaborations with research and technology leaders 3. The AI Revolution of 2025: Transforming Smart Cities with Predictive Power and
in the quantum cryptography sphere, aiming to craft Ethical Innovation����������������������������������������������������������������������������������������������������������������������������������������������������������������������� .188
solutions for a dynamic security landscape.

BY JAMES KNIGHTS

With over 30 years of telecom expertise spanning

across Kyndryl, IBM, and Bell Canada, James has
designed and built networks for high-profile sporting
events, world-class sports venues, hospitals, finan-
cial institutions, connected factories, and govern-
ment organizations.

180 181
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

USE OF AI IN HEALTHCARE IN 2025: MEDICAL prevention to recognize abnormal

patterns in network traffic and data •
treatments, pricing plans, etc.
Violations of security and privacy
the hospital CEO to direct staff to
send all patient data to a specific

MIRACLE, OR MEDICAL MALPRACTICE

flows, in addition to identifying legal requirements address, website, fax number, etc.
specific types of data within the for a valid-sounding reason (e.g.,
network that could indicate an These actions can help prevent medi- a merger with another hospital
intrusion. cal identity fraud, IP violations, non- system) that, in actuality, is a
• Data Encryption and Privacy. compliance with legal requirements, criminal, competitor, or other type
Written by Rebecca Herold GenAI-driven encryption systems and a wide range of other harms to not of threat actor.
are in the early stages of use for a only the associated individuals, but also • Bad actors use GenAI to find the
variety of purposes. For example, to the healthcare organizations. open digital windows and unlocked
The healthcare industry has whole- #1: USING GENAI TO DETECT HEALTH THREATS supporting life. Accurate GenAI tools using such encryption to ensure digital doors in organizations’
heartedly embraced AI, and the can be used in a wide range of ways patient data is encrypted if there #3: USING GENAI TO HARM HEALTHCARE networks, and they can do this from
beneficial possibilities for using AI are GenAI tools that have been validated as to identify such threats, and then block is an indication that a network ACTIVITIES the other side of the world. GenAI
exciting to imagine. It is important, being accurate are particularly effective them from disrupting digital-based intruder may be targeting PHI has now made it much easier for
though, for healthcare professionals when used to evaluate patterns in large healthcare treatments and connections based on real-time risk assessment. Bad actors love healthcare data, and crooks to find even more such
and business leaders to understand patient databases to identify possible to life-supporting devices. A few exam- • Detection of Data Access Pattern how they can utilize that data to do a vulnerabilities than ever before,
that there are literally thousands of abnormalities which could represent a ples include the following: Anomalies. GenAI is being used much wider range of crimes than they after which the crooks can then
AI tools, some of which are good and wide range of types of threats to health. • Intrusion and Data Breach to monitor and analyze the types can accomplish with only basic, more easily exploit the vulnerabilities
some that are buggy and inaccurate. For example, exploited threats to the Detection and Prevention. GenAI of access, and access patterns, widely collected, personal data. Bad to load ransomware, steal patient
medical devices that patients depend tools are being used in intrusion in patient health databases, and actors can also sell that data at a much health databases, inject malware
There are many different ways in upon could change drug dosages, detection systems (IDS), intrusion then to send alerts when detecting higher price than other types of per- into medical devices to cause
which AI can be used to support better settings for life-preserving actions, and prevention systems (IPS), and unusual activities. sonal data. GenAI provides health-da- disfunction during surgeries, and
healthcare. GenAI tools and applica- even digitally shutting off the device for PHI breach detection and ta-loving bad actors have another type more.
tions in particular are being increas- #2: USING GENAI TO DETECT UNAUTHORIZED of tool they love almost as much as the • AI tools can be used by bad actors
ingly used. In fact, National Institutes of ACCESS TO PATIENT DATA data. to cause a wide range of harms. For
Health research finds that healthcare example, change the patient health
pros of all types are using GenAI tools In addition to those benefits described Healthcare leaders, as well as cyber- data that could result in physical
in a wide variety of ways, for all types previously, additional possibilities for security and privacy pros, need to harms to the associated patients
of treatment, payment and operations beneficially using accurate GenAI tools understand how these capabilities and, make apps and websites that
activities. include identifying and preventing ac- can impact the security and integrity are mistaken for valid healthcare
tivities that could result in a wide range of their associated healthcare digital software, and then do an unlimited
To be beneficial for the wide range of of fraud: ecosystems. Healthcare business as- range of harmful activities.
healthcare activities, such as detecting • Unauthorized access to protected sociates (BAs) also need to stay on top
threats to health and detecting unau- health information (PHI) of the AI driven threats and supporting #4: THOUGHTFUL, RISK-AWARE USE OF
thorized access to patient data. GenAI • Unexpected/unusual use of PHI tools and to not use them with their GENAI CAN IMPROVE THREAT DETECTION
must be accurate, secure and protect • Attempts to exfiltrate PHI and other CEs’ data that have been entrusted to AND FRAUD MITIGATION
privacy. If not, it could cause much types of sensitive data, such as them.
more harm than good. It is important medical intellectual property (IP) Ultimately, every healthcare organiza-
to ensure before choosing an AI tool • Cybersecurity incidents resulting GenAI tools are being used by those tion must establish rules and policies
that it is thoroughly vetted to ensure from leaked IT specifications, health-data-loving bad actors to trick for the use of AI within their organiza-
it is accurate, and protects privacy. administrative settings, etc. victims through the use of new and tion that cover both the risks and the
It is also important for all healthcare • The creation of additional attack more effective social engineering benefits, provide training for GenAI
organizations to understand that GenAI vectors that would allow hackers to (phishing) tactics added to their land- issues, and include AI tools and use
can also be used to harm healthcare enter the healthcare organization’s scape of attack tools. within their risk management program
organizations, as used by cybercrimi- digital ecosystem • AI tools can impersonate quite scope. Security leaders play a pivotal
nals, and take actions to mitigate such • Leaking network and system convincingly the images and role in ensuring such actions are taken.
risks. parameters, access points, etc. voices of healthcare leaders, such
• Commercial losses from stolen, as hospital CEOs and Medical A final warning: Always test any AI tool
unreleased products and Directors, such as to impersonate that claims to be providing benefits to

182 183
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

ensure that it provides accurate results, does not put PHI at risk by exposing requirements for patient data.
will not negatively impact the per- or inappropriately sharing PHI, and
formance of the associated network, does not violate the organization’s legal

BY REBECCA HEROLD

Rebecca Herold is an IEEE Member and has 35+

years of IT, security and privacy experience, founded
Privacy Professor Consultancy (2004) and Privacy
& Security Brainiacs SaaS services (2021). Rebecca
has authored 22 published books, co-authored
many NIST cybersecurity and privacy standards, and
serves as an expert witness for many cases involving
cybersecurity, privacy issues.
THE STATE OF IOT: TRANSFORMATIVE
MOMENTUM AND WHAT LIES AHEAD
Written by Ron Totton
President and CEO, KORE Wireless

As 2024 comes to a close, the Internet AND ENHANCED EDGE COMPUTING faster responses. This is especially
of Things (IoT) has firmly established valuable in industries like agriculture,
itself as a force revolutionizing indus- We’ve been hearing about AI and IoT manufacturing and logistics, where low
tries across the board. From enabling individually as separate game-chang- latency and operational cost efficiency
smarter healthcare and transportation ers for years, but 2025 is when their are critical. In 2025, edge computing
systems to transforming agriculture convergence will truly take center will continue to mature, with increased
and retail operations, IoT is driving stage. The focus will shift from using AI focus on integrating edge devices
efficiencies, unlocking new revenue as just an analytics tool to embedding with cloud platforms and optimizing
opportunities, and even tackling global AI throughout the IoT ecosystem – at energy efficiency. Beyond the technical
challenges like climate change. the edge, in connectivity management, aspects, the edge’s role in addressing
within security layers, etc. AIoT trans- data sovereignty concerns will be a
Yet, as we step into 2025, some critical forms IoT devices into intelligent deci- major driver for adoption, as business-
questions emerge: What innovations sion-makers that drive efficiency gains es strive to comply with regional reg-
will shape the next wave of IoT’s im- across industries where the costs of ulations while maintaining operational
pact? How do we ensure always-on, unexpected failures are extremely high, efficiency.
reliable connectivity? And how can such as predictive maintenance for
organizations overcome the lingering machinery, real-time route optimization #2: IOT STEPPING UP WITH VERTICAL-
security concerns that continue to hold for fleets and better patient outcomes SPECIFIC SOLUTIONS (VERTICALIZATION OF
them back? Let’s explore the trends, in connected health. IOT) AND PRIORITIZING SUSTAINABILITY
challenges and opportunities that will
define IoT in the year ahead. Edge computing enables real-time IoT is moving away from one-size-fits-
processing and decision-making closer all solutions toward highly specialized,
#1: INTEGRATION OF AI-POWERED IOT (AIOT) to where data is generated, enabling vertical-specific applications. This

184 185
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

trend is driven by the unique needs of #3: ADDRESSING INTEROPERABILITY, cellular, satellite (NTNs) and LPWAN tection and frequent firmware updates, solutions and actively demonstrate how to move forward.
industries such as connected health, STANDARDIZATION AND CONNECTIVITY will cater to diverse needs, while NTNs fostering confidence rather than fear. security protocols mitigate real-world
agriculture, fleet management and DISRUPTION FOR IOT will expand IoT’s reach into remote At the same time, the industry must threats. This transparency will help shift 2025 is shaping up to be a pivotal
smart cities. In 2025, businesses will regions, unlocking transformative use move away from using security as a the narrative from fear to confidence. year for IoT. A year that cements IoT
demand IoT solutions that are tailored As a long-time barrier to widespread cases. eSIM technology, paired with marketing tactic and instead prioritize as a must-have for organizations. In
to their operational challenges, whether adoption, the fragmented nature of the robust connectivity management education and collaboration to address Again – this needs to be shouted from the coming year, we should be excit-
that means remote patient monitoring IoT ecosystem has caused hesitation to platforms, will be essential in simpli- shared challenges. the rooftop – partnerships will play ed about what’s to come but also be
in healthcare, precision farming in ag- invest in IoT solutions. As businesses fying network transitions, optimizing a significant role in building trust. By intention about empowering business-
riculture, or real-time asset tracking in demand solutions that integrate effort- performance and scaling Massive IoT The path to widespread IoT adoption collaborating with well-established IoT es to harness the full potential of IoT
logistics. The next phase of verticaliza- lessly with existing systems and devic- deployments. These advancements will lies in education, collaboration and providers who specialize in end-to-end without getting bogged down by its
tion will focus on leveraging cross-ver- es, 2025 marks a turning point where ensure seamless, always-on connec- simplification. For organizations who solutions, organizations can tap into a challenges. It’s time to fully embrace
tical learnings, applying insights from interoperability and standardization tivity across multiple networks, forming are hesitant to adopt IoT, fostering a wealth of expertise and resources de- the innovations, tackle the uncertain-
one industry to drive innovation in to achieve seamless communication the backbone of future-proof IoT eco- deeper understanding of the security signed to make deployment seamless ties and make this the year IoT truly
another. across devices, platforms and networks systems and enabling global deploy- measures already available is critical. and secure. These providers simplify transforms the world.
becomes a necessity. Open APIs, in- ments with minimal complexity. Platforms and providers need to better complex implementations, offering
Along the same lines, the push for dustry collaboration and adherence to communicate the robustness of their businesses the reassurance they need Here’s to a connected future!
sustainability is no longer optional, and global standards will be key to breaking #4: BUILDING CONFIDENCE IN IOT AND
IoT is stepping up to the challenge. down silos and ensuring scalability. Be- OVERCOMING CONCERNS: WHAT’S NEXT?
Smart agriculture solutions are already yond technical compatibility, the focus
conserving water and chemical usage, will also extend to operational simplic- This year will be the time to tackle the
while smart energy grids and connect- ity making IoT easier to manage for elephant in the room: security. Ev-
BY RON TOTTON
ed building technologies are optimizing end-users and reducing the complexity ery year, surveys show that security
energy consumption. As organizations of multi-vendor ecosystems. concerns remain a top barrier to IoT Ron Totton is President and CEO of KORE, a global
increasingly prioritize ESG (Envi- adoption. And while IoT security re- leader in Internet of Things Solutions and provider of
ronmental, Social and Governance) The IoT industry is at an inflection mains a critical concern, the fear-based IoT Connectivity, Solutions and Analytics. Totton has
commitments, 2025 will see a broader point where technological innovation narratives surrounding it are becoming nearly three decades of strategy, finance, sales, and
adoption of IoT solutions designed to is addressing longstanding barriers like counterproductive. In 2025, there will leadership experience, including many years in the
measure, monitor and reduce carbon fragmented connectivity, high costs be greater focus on proactive, built-in technology, media, and telecoms sectors.
footprints across industries. However, and limited coverage. In 2025, connec- security measures that are seamlessly
the industry must also address the tivity will be reshaped by 5G expansion, integrated into IoT ecosystems. Orga-
sustainability of IoT itself, including the the rise of Cat 1bis, NB-IoT reconsider- nizations will focus on solutions that
lifecycle of devices, the energy con- ation and the continued importance of combine ease of implementation with
sumption of networks and the recy- LPWAN in sectors like agriculture and robust protection against threats, such
clability of hardware. utilities. Hybrid solutions integrating as end-to-end encryption, threat-de-

186 187
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

THE AI REVOLUTION OF 2025: TRANSFORMING Another example is energy manage-

ment within buildings. Invisible AI sys-
#3: LOW-BANDWIDTH, HIGH-IMPACT
SOLUTIONS
for decision-making. Cities and large
enterprises can simulate urban policies,

SMART CITIES WITH PREDICTIVE POWER AND

tems can analyze patterns in electricity test disaster responses, and even plan
usage across offices, apartments, and Sustainability will also play a central large-scale projects––like integrated
public spaces, dynamically adjusting role in AI’s urban evolution, as many communities and amusement parks––

ETHICAL INNOVATION
power allocation to reduce waste and smaller cities and rural areas lack the all within a risk-free virtual environ-
optimize energy efficiency without the high-bandwidth networks required ment.
need for manual intervention or new for traditional smart technologies.
hardware installations. Enter: low-bandwidth, edge-based This predictive modeling capability
AI solutions. These systems operate will revolutionize urban planning by al-
Written by David Ly #2: ETHICAL INNOVATION: BALANCING locally––reducing the need for data-in- lowing things like traffic management,
Founder, Iveda PROGRESS AND CAUTION tensive transfers––and enable smaller disaster readiness, and infrastructure
municipalities to reap the benefits of AI development to be optimized with a
As we step into 2025, the role of arti- #1: BEYOND CAMERAS: THE RISE OF overhauling existing infrastructure. While the rise of autonomous AI sys- without costly infrastructure upgrades. level of precision that was previously
ficial intelligence (AI) in urban envi- “INVISIBLE” AI tems brings endless potential, it will unimaginable.
ronments is poised for a revolutionary One standout application of this is in inevitably stir political and social de- For instance, localized AI applications
shift. Gone are the days when AI was For years, urban infrastructure––like emergency services. AI won’t just man- bates as well. Concerns about privacy, could monitor energy usage or manage A future within reach
merely a passive assistant; now, it’s cameras and IoT sensors––has been age alerts; it will predict emergencies. accountability, and potential biases will traffic flow in real-time with minimal
evolving into a decision-maker. Across the backbone of smart city technology. By analyzing abnormal traffic patterns, lead some municipalities to adopt a data exchange. These solutions dra- The AI-driven smart city of 2025 isn’t a
several cities, AI systems will autono- But in 2025, these visible systems will weather conditions, and contextu- “smart pause” approach. This cautious matically democratize access to cut- distant dream—it’s unfolding before us.
mously manage real-time adjustments start to take a backseat. The future lies al cues, AI can anticipate potential rollout strategy involves implementing ting-edge technology, all while aligning From leveraging invisible technologies
to things like traffic flow, energy con- in “invisible” AI-driven solutions that incidents and preemptively coordinate higher standards for AI auditing, trans- with broader global sustainability goals. to pioneering ethical innovation and
sumption, and public safety, heralding leverage existing data streams without resources. Imagine a city where emer- parency, and bias reduction. democratizing AI access, cities are on
a new era of predictive and preventive requiring extensive physical upgrades. gency response teams are dispatched #4: DIGITAL TWINS: A VISIONARY LEAP IN the brink of a transformative era. While
urban management. This leap forward By tapping into the vast amount of to likely hotspots before an accident Far from stalling progress, these ethical URBAN PLANNING challenges remain, the potential for
promises not just cost savings but information already generated by cities, even occurs—saving time, resources, considerations could position such smarter, more sustainable urban living
also an unprecedented boost in urban these systems will enable transfor- and most importantly, lives. cities as leaders in responsible innova- Digital twins—virtual replicas of has never been greater. By embracing
efficiency. mative changes without the need for tion. By addressing societal concerns physical environments—will be- both the possibilities and responsi-
upfront, these cities may end up setting come indispensable tools in 2025. By bilities of this AI revolution, cities can
global benchmarks for ethical smart integrating real-time data and predic- chart a path toward a future where
city practices that blend technological tive algorithms, these digital models technology truly serves the people.
prowess with human-centered values. provide an operational dashboard

BY DAVID LY

David Ly is the visionary founder of Iveda, having

served as CEO and Chairman of the Board of Direc-
tors since the company’s inception in 2003. With over
20 years’ experience in wireless data, cellular, IT, and
cloud video surveillance, David has built a pioneering
cloud video hosting and real-time surveillance
infrastructure with use cases across the globe.

188 189
 eBook: The GenAI leap: how networks are reinventing themselves

REVOLUTIONIZING PUBLIC SAFETY: THE CRITICAL

NEXT STEPS FOR NG911 OPTIMIZATION IN 2025
Written by Liz Nguyen
CTO, Intrado

PUBLIC SAFETY
In 2024, a multitude of challenges outcomes, as well as support equitable for public safety operations. Regulatory
emerged as states and 911 jurisdictions and effective emergency responses requirements for Next Generation 911
embraced emerging Next Generation nationwide. (NG911) technology deployment are
9-1-1 (NG9-1-1) technologies that will pushing public safety answering points
play a significant role in shaping the #1: BUSINESSES & SCHOOLS WILL SEIZE (PSAPs) and 911 authorities to adopt
future of public safety. In 2025, it will UNPRECEDENTED OPPORTUNITY FOR DEEPER new infrastructures and protocols.
become increasingly critical for public ANALYSIS OF PUBLIC SAFETY DATA THROUGH Simultaneously, legislation at the local
1. Revolutionizing Public Safety: The Critical Next Steps for NG911 Optimization safety answering points (PSAPs), STANDARDIZATION level that requires 911-connected panic
public safety authorities and critical buttons for schools and retail environ-
in 2025���������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������� .191
infrastructure providers to implement New legislation and technological ments is creating new participants in
and deploy NG911 technology. This will capabilities are pushing the boundaries public safety networks and infrastruc-
2. The Importance of Connectivity for AI Applications in Public Safety and enhance synergy and public safety of what has been historically possible ture. However, seizing opportunities for
Critical Response������������������������������������������������������������������������������������������������������������������������������������������������������������������������ .194

3. The Ghost Escapes the Machine: GenAI Brilliance for Networks������������������������������������������������.197

190 191
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

public safety improvement depends on OF NG911 INFRASTRUCTURE IP-based NG911 technology, a unified COLLABORATION TO MAXIMIZE THE IMPACT Aside from enabling connectivity, Next age the range of available data to im-
accurate and effective data collection, standard for NG911 infrastructure will OF PUBLIC SAFETY TECHNOLOGY Generation 911 technology is trans- prove public safety, the companies who
analysis and sharing, which will be- States across the U.S., as well as the need to be established. Standardization forming public safety by enhancing the provide those products will need to
come a top priority for both public safe- FCC, have recognized the benefits of across states’ 911 infrastructure will In 2024, the public safety industry ex- ability to deliver and receive supple- ensure their data is accessible by pub-
ty and private organizations in 2025. Next Generation 911 (NG911) technol- help to protect investments in NG911 perienced unique and unprecedented mental data that can provide invaluable lic safety systems and professionals.
ogy and have pushed public safety deployment and significantly reduce opportunities to improve communica- context and information during a crisis. We’ll likely see increased collaboration
There is a plethora of public safety data authorities to deploy IP-based 911 in- the potential for geopolitical rivals to tion with, and the safety of, citizens and Vehicular data from car crashes, home and partnerships between public safety
available that can be generated and frastructure to ensure the safety of cit- negatively impact the safety of U.S. businesses. Satellite technology is now alarm data from break-ins, mobile de- organizations and private companies,
shared from enterprises, schools and izens and businesses. Currently, there citizens and businesses. being used to supplement wireless vice data that provides health informa- with the aim of improving the sharing
other facility security systems. Addi- is no unified standard for NG911, and coverage, especially in remote loca- tion and location details, as well as live of safety data, in 2025 and beyond.
tional supplementary data can also be many states are attempting to deploy #4: PUBLIC SAFETY AUTHORITIES tions with little or no access to cellular camera feeds can all be used to better
provided by other systems and public bespoke NG911 infrastructure without WILL ENSURE EQUITABLE PROTECTION service. This supplemental coverage inform 911 telecommunicators and Prioritizing the future of public
safety touchpoints, including: vehicles, thinking about the implications for NATIONWIDE BY REMOVING NG911 from space is also essential for the first responders. Proper data sharing safety
which can provide crash data and synergy and collaboration across state DEPLOYMENT DISPARITIES handling of natural disasters like Hurri- reduces public safety response times,
visual evidence; home alarm systems, lines. To truly leverage the full benefits cane Helene, when satellites provided enhances first responder prepared- As we look ahead to 2025, the potential
which can provide statistics on break-in of NG911 technology, 911 jurisdictions Everyone in the U.S. deserves the same supplemental coverage after traditional ness, and improves outcomes. for NG911 technology to revolution-
methods and frequency; mobile devic- (at the state, county and city levels) level of protection from public safety wireless networks failed. The FCC ize public safety is both exciting and
es and health-focused tech, which can must be able to seamlessly interact organizations and first responders. recognizes this advantage and has In 2025, we’ll see more and more urgent. By prioritizing data standard-
provide information on locations, heart and share information across public Many states across the country have introduced requirements that include non-public safety organizations – such ization, infrastructure uniformity, and
rates and biological signatures during safety systems. In 2025, public safety already begun exploring or deploying the delivery of 911 services via satellite, as satellite providers, vehicle manufac- cross-industry collaboration, public
medical emergencies; as well as other authorities and policymakers will need Next Generation 911 (NG911) infrastruc- highlighting the growing role of the turers, mobile device companies, and safety authorities and private enterpris-
sources. That said, the utilization of to ensure the interconnectivity of na- ture and technologies to ensure the convergence of satellite and wireless others – begin to proactively consider es can unlock transformative benefits
this data is hindered by fragmentation, tionwide 911 networks and systems by safety of their citizens and business- in 2024, that will become increasingly the implications of their data collection for emergency response nationwide.
silos, and incompatible data standards. creating a uniform standard for NG911 es. Currently, each state is exploring important in 2025. and sharing capabilities. To fully lever-
infrastructure, regardless of where it is different deployment strategies and
In 2025, data centralization and stan- deployed. customization of NG911 solutions,
BY LIZ NGUYEN
dardization will become a key priority which creates a disparity in the speed
on multiple fronts, including for PSAPs, #3: 911 NETWORKS’ CYBERSECURITY of NG911 deployment and the level of Liz Nguyen, Intrado’s CTO, brings 20+ years of
wireless carriers, public-facing insti- DEFENSES WILL BE BOLSTERED BY NG911 protection provided to citizens across experience in software development and technology
tutions such as schools, and private UNIFORMITY the U.S., dependent on where they strategy. Formerly CTO at Vertafore, she led digital
enterprises. To truly maximize the value are located. It’s unfair for U.S. citizens transformation in insurtech. At Intrado, Liz focuses
of all available public safety data, these Attacks from various geopolitical ac- located in rural areas and underfund- on innovative solutions, fostering collaboration, and
parties will need to ensure data sharing tors, who are trying to gain access to ed regions to accept a lower standard addressing evolving customer needs while driving a
and analysis challenges don’t impede 911 infrastructure, are becoming more of safety compared to their peers in culture of continuous learning within her teams.
emergency responses or outcomes. frequent. As states across the U.S. more populated areas. In 2025, public
Thankfully, these organizations have deploy IP-based, Next Generation 911 safety authorities and policymakers
been presented with the opportunity to technology (NG911) – which is essential will need to address this challenge by
effectively standardize data with NE- to the efficient protection of citizens establishing a uniform standard for
NA’s i3 data standards. Cross-industry and businesses – there will need to be NG911 infrastructure and technology
standardization of safety and security established and enforced cybersecurity deployment. Not only will this speed up
data will be essential to the success standards to prevent attackers from ac- deployment and enablement of NG911
of public safety moving forward, and cessing critical information and disrupt- capabilities, but it will also ensure every
no organization should let poor data ing public safety operations. In 2025, citizen receives the same level of safety
hygiene cause an issue when their public safety authorities and policy- from the first responders and public
workers’ and customers’ safety is at makers will need to address this issue. safety institutions they depend on for
stake. We cannot be dependent on outdated, protection.
legacy infrastructure to execute public
#2: PUBLIC SAFETY SYNERGY ACROSS STATE safety operations in the information #5: PUBLIC SAFETY ORGANIZATIONS &
LINES WILL DEPEND ON STANDARDIZATION age. However, to protect the security of PRIVATE ENTERPRISES WILL INCREASE

192 193
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

THE IMPORTANCE OF CONNECTIVITY FOR AI awareness through real-time data

interpretation and use intelligent
agencies.

#2: THERE IS NO AI OR EDGE COMPUTE

satellite (GEO, MEO and LEO) and
cellular (3G, 4G, 5G, LTE, etc.). By com-

APPLICATIONS IN PUBLIC SAFETY AND CRITICAL

navigation and path planning, bining these diverse networks, from di-
avoiding obstacles and optimizing WITHOUT RELIABLE CONNECTIVITY verse carriers, we can provide a reliable
routes. and high-bandwidth connection.

RESPONSE
• AI-driven surveillance and Bridging communication between
monitoring systems: These physical sensors, the digital edge and #4: DEVELOPMENTS IN SATELLITE
systems use advanced algorithms, private or public cloud services, reliable CONNECTIVITY
powered by computer vision, to connectivity is critical to enable edge
analyze live video feeds, identifying compute so users can anticipate and New technology is being developed
threats like suspicious activities, respond, in real-time, to boost effi- on the satellite side that is blurring the
Written by Kevin Fernandes unknown objects, or unusual ciency and improve mission-critical lines between terrestrial and non-ter-
Chief Revenue Officer, Dejero behavior. Examples include facial systems. restrial telecoms. Satellite connectivity
recognition in the community and is becoming more prevalent for phone
object detection systems in airports. As we continue to enable cloud com- communications - which was tradition-
• AI-powered predictive analytics: puting and AI to leverage the power of ally dominated by wireless connectivity
In a range of public and private edge By analyzing historical data and computing at the edge, we need con- i.e. terrestrial but, satellite to phone is
environments, innovative AI/machine cross-referencing it with social, nectivity to move that data back and low bandwidth at 1-5 Mbps, the capa-
learning (ML), analytics, internet of geographic, and demographic forth from AI-powered edge devices to bilities of which just aren’t not enough
things (IoT), and container solutions variables, AI models can forecast the vast databases and large language for AI processing.
have enabled new applications, deploy- extreme-weather or crime-prone models at the core.
ment options, and use cases, and most areas, enabling first responders Even with the vast and rapid deploy-
crucially in the realm of public safety. and law enforcement to proactively Many advancements have been made ment of LEO satellites, bandwidth will
addresses. sensors assist with thermal imaging, allocate resources more effectively. with regards to processing data at the still be challenging. Cloud-Based AI
While AI’s potential in public safety is fire perimeters, preliminary damage • AI-driven mapping: AI analyzes edge – but limitations on edge device Applications (Data Analytics, Machine
widely recognized, its adoption has We’ve seen the chaos and destruction assessment, including preliminary satellite imagery in real-time to processing power, local storage and Learning Models, Video Conferencing
been relatively slow. A Verizon Frontline caused by the recent wildfires across aerial and satellite assessment, to map the impact zones of natural compute means that almost every with AI) all need at least a bandwidth
survey of 1,700 first responders in 2024 Los Angeles. At one point, 200,000 direct damage inspection crews to high disasters such as earthquakes application needs reliable and secure of 10-50 Mbps.
revealed that only 14% currently utilize people were under evacuation, 24 peo- priority locations. The agency’s fire and floods. It offers real-time data connectivity to access larger AI data-
AI or smart solutions. However, the fu- ple have died, thousands of structures behavior analysts also use the remote analysis for accurate damage bases. If you’re running complex AI tools that
ture of AI in public safety is encourag- have been devastated resulting in tens sensors and AI-powered cameras to assessment, evacuation routes require cloud processing (e.g., AI mod-
ing, with 75% of respondents believing of $billions worth of damage, and more receive real-time updates and to moni- and predictive analytics to forecast #3: NEVER RELY ON A SINGLE NETWORK els for analytics, data processing, or
it will be either crucial or a top priority than 30,000 acres of land have been tor weather. affected areas. AI-driven maps collaborative tools), you’ll need a more
for their agencies. This suggests a sig- lost. support better coordination among So what are the best methods of con- stable and higher-speed connection.
nificant shift is underway, and we can #1: PUBLIC SAFETY AI TRENDS TO WATCH IN emergency responders and necting the edge to the cloud to power For intensive tasks such as running
expect to see a surge in AI adoption According to StateScoop, AI cameras, 2025 facilitate quicker recovery efforts AI models and databases? - and how large-scale AI models (e.g., natural
within public safety in 2025. sensors, and drones have been aiding by identifying critical infrastructure do we make that connectivity reliable? language processing, computer vision),
Cal Fire to battle the L.A. wildfires. The The application of AI in public safety needs. especially when training models or
AI in the fight against the Los organization uses AI-powered cameras is certainly evolving fast, and its future Transmitting real-time video and data transferring large data files to/from the
Angeles wildfires placed on mountain peaks and hilltops offers even more potential. However, The effectiveness of all this AI-pow- from the field requires a huge amount cloud, a much higher bandwidth may
that perform 360-degree sweeps every realizing this potential depends heavily ered technology depends on reliable of bandwidth, and relying on a single be needed of 50-100 Mbps or more.
Crisis communication deals with two minutes and can see as far as 60 on robust and secure communication communication networks for real-time connection leaves organizations vulner-
managing unexpected events that pose miles on clear days, and up to 120 miles infrastructure. data transmission and control. These able. Having the ability to work across This is why hybrid networks become
significant risks to the public requir- on clear nights. applications require high-bandwidth all the different modes of connectivity, essential to ensure reliable connectivity.
ing immediate action, such as natural New trends to watch include: and low-latency connectivity to ensure both terrestrial and non terrestrial,
disasters - hurricanes, wildfires, major In addition to the AI-powered camer- • AI-driven Unmanned Aerial the seamless flow of information and provides a more robust foundation for Key Considerations:
incidents or top-level security events as, Cal Fire’s remote sensing program Systems: AI-powered UAS or facilitate rapid decision-making, with AI to work. • Latency: AI applications,
such as presidential inaugurations, the has been working 24/7 to provide drones offer enhanced autonomy the capability to access vast amounts especially those relying on cloud
Super Bowl and State of the Union intelligence to first responders. The and improved situational of data and reliable data sharing mech- It is now possible to blend all available infrastructure, benefit from low
anisms across different public safety IP networks together including wired, latency to ensure smooth real-time

194 195
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

•
interactions.
Upload vs. Download Speeds:
destination and then return, incurring
significant latency—5G aims to locate
All the ingredients for reliable connec-
tivity in AI applications THE GHOST ESCAPES THE MACHINE:
GENAI BRILLIANCE FOR NETWORKS
Upload speeds become more applications and the cloud compute,
important when you’re actively storage, and network functions needed For us, new satellite networks, 5G
sending data (e.g., training a model to run them, close to the access edge. and even 6G (an AI-native environ-
or streaming AI-based video). By doing so, 5G promises to drop ment expected to launch in 2029 5G
• Network Stability: A stable and latency significantly (the aim is for Americas Forum: 5G Advanced, 6G
consistent connection is more <1ms), which is expected to enable and and a looming spectrum crisis | S&P Written by Cemal Dikmen
important than sheer bandwidth, improve applications including AI use Global) are simply new ingredients in Chief Technology & Security Officer, SS8
especially for real-time AI tasks. cases. our blending recipe, joining 3G and 4G
cellular, satellite, Wi-Fi, and broadband
#5: 5G DEVELOPMENTS However, 5G services are currently lim- connections. We continue to embrace
ited to densely populated urban areas the transition as 5G, GEO, MEO and For years, network operators have been implementations. spin instances up and down at will, for
5G has the potential not only to enable or municipalities which are friendly to LEO satellites, and emerging connec- awash in data that is easier to collect seamless failover and elastic capacity,
transmission of more and higher-qual- the zoning requirements associated tivity options come online. than it is to derive value from. Sources #1: MORE-CAPABLE SELF-HEALING without keeping equipment in reserve.
ity video, but to allow public safety with small cell deployments. According ranging from system telemetry to user As the orchestration engines for those
organizations to leverage AI and edge to Kagan’s 5G Tracker (June 2024), out AI is becoming indispensable for effec- sessions funnel into network opera- Not long ago, best practices were to workloads and resources improve, one
computing resources to pre-process of the total 290 telecoms operators that tive first response and disaster man- tions centers, flooding and overwhelm- support failover with redundant physi- effect is to make the environment more
feeds to prioritize immediately action- have launched 5G in 106 markets, at agement, but its effectiveness across ing response teams. cal systems on standby. Together with resilient by accelerating recovery from
able or especially valuable information, least 69 operators in 40 markets world- public safety hinges on high-band- buying and maintaining hardware for outages.
while storing secondary feeds for later wide have activated 5G Stand Alone width and low-latency connectivity. Lawful and location intelligence ex- capacity headroom, this approach was
transmission. (SA) networks. Operators in major 5G tends the scope of data collection even both inefficient and expensive. AI-enabled orchestration adds predic-
markets such as the US and mainland It’s crucial that public service agencies further, drawing information from inside tive capabilities and workflow auto-
Essentially, 5G is the fifth generation of China have transitioned to 5G SA, but implement more advanced, enabling and outside mobile networks—call data Now, software-defined networking and mation that can reduce the burden
cellular network technology providing other markets have yet to follow the technology, whether it’s telemedicine records, location data, open-source cloud computing make it possible to on human operators while improving
20 times the peak data rate, 10 times trend. and ambulances where we’re transmit- intelligence, and much more. Unifying
lower edge latency, and 3 times more ting EKGs or having video conferences and drawing insight from that data in
spectral efficiency than 4G LTE. More 5G isn’t replacing 4G (which has with doctors during surgery, or having all its scale and diversity has been a
and more devices are supporting 5G. replaced 2G and is replacing 3G); remote access to building plans and central focus of lawful intelligence, to
instead, 5G and 4G will work in a com- sending drones video back to head- better enable investigations and pro-
Unlike previous generations of mobile plementary fashion to handle different quarters and command posts. tect society.
network technology—which required types of traffic and to address different
traffic to traverse the network to some use cases. AI expands the ability to analyze
streaming and static data at any scale,
with algorithmic interpretation that far
surpasses human abilities alone. The
BY KEVIN FERNANDES
requirements and capabilities that flow
Kevin Fernandes joined Dejero in 2017 and leads the from that evolution are changing how
global sales team as Chief Revenue Officer. He brings networks are being architected and
15 years of experience working with audio and video used. With the ability to understand
customers to develop and support equipment for and synthesize language, generative
broadcast, medical, and industrial markets. Prior to AI (GenAI) and large language mod-
Dejero, Kevin held various senior level B2B sales roles els (LLMs) give machines human-like
with a focus on connecting clients with the right communication that is redefining
solutions to fulfil their business needs. Kevin holds a network operations.
Bachelor’s in Computer Engineering and a Masters
of Applied Science in Electrical Engineering from This article discusses the resulting
McMaster University, located in Hamilton, Ontario. state of change in communication
service provider (CSP) networks, lawful
intelligence solutions, and other large

196 197
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

uptime, including with self-healing a GenAI agent might detect potential the continued growth of other usages Conclusion and risk. Looking ahead, as GenAI becomes
for autonomous recovery. GenAI uses response-time improvement by in- that demand low latency for immedi- a fixture in every environment, hu-
LLMs to deduce logical sequences of creasing the capacity of a constrained ate response, from self-driving cars GenAI’s broad-based network evolu- Better user experiences will set the man-machine collaboration will enter a
steps to resolve issues, either to be cluster, then look at best practices, to smart cities. More-advanced edge tion will bring improvements across stage for competitive advantage and new phase where technicians, analysts,
completed autonomously or explained historical data, and predictive analysis resources will set the stage for further operations, functionality, and secu- increased revenue from new services and investigators will leave the heavy
narratively to operations teams. Natural to propose specific measures. development of GenAI solutions, and rity. New GenAI-driven insights and and networks. Automating workflows lifting to models and algorithms. For
language processing makes it possi- those solutions will in turn advance network capabilities will set the stage with chatbots and other AI agents lawful intelligence and all network
ble for chatbots and other automated The agent might contact a human an- development of edge infrastructure, in for more robust lawful intelligence as will drive efficiencies that help lower usages, turning that page will unleash
attendants to feed network incident alyst and say, “The CRM is low on com- a virtuous cycle. well as advances for enterprises, CSPs, costs, to do more with less. Enhanced new abilities that will advance every
management with input from customer pute resources. Based on the demand and other environments. GenAI-based cyber-protection will reduce the risk of organization and mission.
reports, social media, and other lan- profile, I recommend adding six vCPUs. #4: MORE-ROBUST CYBER PROTECTION technologies are poised to deliver ben- doing business.
guage-based data sources. Would you like me to proceed?” The efits related to business revenue, cost,
analyst could ask for more information Improving network security is a natural
These expansions to what is possible or an alternative resolution, then direct fit for GenAI agents. In fact, adversaries BY CEMAL DIKMEN
with autonomous and semi-autono- next steps, enhancing both operating have already adopted the technology to
Cemal has been with SS8 since 1995. As SS8’s
mous network recovery are emerging efficiency and control over the network. improve the stealth and power of their
Chief Technology & Security Officer, he plays an
as an early usage model for GenAI attacks, and applying GenAI to defens-
integral role in the company’s strategic direction,
that effectively multiplies IT operations #3: MORE-ADVANCED EDGE COMPUTING es is a necessary step to keep pace.
development, and future growth in regards to its
resources. They are particularly valu-
security, personnel, physical assets, and information.
able to cost-effectively increase high GenAI inferencing draws on large For example, GenAI can synthesize
A renowned expert and thought leader in the legal
availability of mission critical systems amounts of data that grow as imple- malware samples and analyze them to
compliance and communications analysis domain,
such as lawful intelligence. mentations become more widespread understand how they operate and how
he has been a frequent speaker at various industry
and sophisticated. Sending that data to mitigate them, helping guard against
conferences over the past 10 years. Prior to joining
#2: MORE-EFFECTIVE NETWORK OPERATIONS over the wire for centralized inference emerging threats. Email scanning and
SS8, Cemal held software engineering and business
raises bandwidth expense and incurs self-learning continuous analysis can
development positions at NewNet, ADC Telecom-
In addition to incident management transfer latency that interferes with help detect sophisticated phishing at-
munications and Telenity. He holds BS, MS, and PhD
and self-healing, AI-based tools are real-time and near-real-time usages. tempts, warn users, and offer to contain
degrees in Electrical Engineering.
quickly becoming more prominent It may also create complex data paths, malicious URLs or files. Monitoring
in running networks under normal with long-haul, multi-hop transmissions threat intelligence feeds can draw in-
operating conditions. As network increasing the potential for failures and sights from massive amounts of textual
management tools evolve to draw on creating security exposure with a larger information and reveal how it pertains
GenAI, they will become more capa- potential attack surface. to internal networks.
ble of human-like interactions based
on language, which could make them As GenAI becomes more enmeshed As with other usages, GenAI agents
dramatically more effective. in both network operations and us- can suggest prevention or remedia-
er-plane workloads, these forces are tion steps using natural language in
Collections of GenAI agents may soon likely to accelerate the existing industry conversation with human analysts or
traverse networks, identifying tuning trend of pushing compute resources to even users. They also show promise for
opportunities and then relaying them the network edge. In many cases, the producing realistic training scenarios
back to the network operations center. added demands of large-scale Ge- for security teams to hone their skills.
They could create action plans for per- nAI inference will be too great for the Similarly, they could respond to the
formance and stability improvements, abilities of current edge servers, and presence of phishing or other attacks
recommend them to human agents, those resources will need to be rear- by automatically creating a training
and then respond appropriately, both chitected or expanded. Measures such module that explains the threat to rele-
verbally and with actions. as deploying more robust servers and vant users and shows them how to be
adding discrete GPUs or other acceler- more aware of future threats.
These agents resemble junior staff ators will change the asset mix in these
members that extend teams but whose locations.
authority to act on their own can be
limited to mitigate risk. For example, This trend will be compounded by

198 199
 NOTABLE TRENDS IN THE MOBILE WI-FI SPACE
FOR 2025: AI-DRIVEN CLOUD CONNECTIVITY
Written by Eric Plam
President, SIMO

As 2025 approaches, the mobile Wi- overall user experience. eliminating the need for traditional SIM

WIFI
Fi landscape is undergoing a rapid cards, vSIM technology is transform-
transformation, propelled by advances In today’s hyper-connected world, ing mobile connectivity, delivering a
in artificial intelligence (AI) and cloud reliable mobile internet has become a seamless, adaptive internet experience
computing. AI-powered cloud connec- necessity, not a luxury. As the demand worldwide.
tivity is poised to revolutionize network for fast, secure, and stable access
operations, providing smarter, more continues to grow, AI-driven cloud The future of mobile Wi-Fi in 2025 will
efficient, and highly adaptive solutions solutions are redefining how we stay be heavily influenced by the integration
1. Notable Trends in the Mobile Wi-Fi Space for 2025: AI-Driven Cloud for both individuals and businesses. connected. Virtual SIM (vSIM) technol- of AI with cloud connectivity. As cloud
This new wave of connectivity will en- ogy, powered by AI-enhanced cloud computing and artificial intelligence
Connectivity������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������� .201
able seamless, real-time data process- systems, dynamically optimizes net- technologies continue to advance,
ing and delivery, optimizing everything work performance, ensuring consistent, their fusion will redefine how mobile
2. Ten Wi-Fi Predictions for 2025�������������������������������������������������������������������������������������������������������������������������������������� .204 from network performance to the high-speed connectivity globally. By networks operate, manage traffic, and

200
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

provide services to users. latency more accessible than ever. AI connection, are gaining popularity as Conclusion ty, delivering personalized, seamless future is clear—faster speeds, greater
will play an essential role in managing mobile users seek enhanced protection experiences to users. From smarter reliability, enhanced security, and a
Here’s a closer look at the key predic- these networks, ensuring optimal per- against public Wi-Fi vulnerabilities. As AI and cloud technologies con- edge computing to green connectivity, more sustainable mobile ecosystem are
tions and trends shaping AI-driven formance across devices and applica- With devices like these, users can cre- verge, 2025 will mark a new era for the fusion of AI and cloud technology on the horizon.
cloud connectivity in the mobile Wi-Fi tions. Cloud-enabled services such as ate private, mobile Wi-Fi networks and mobile Wi-Fi connectivity. AI-powered will drive the next phase of the mobile
space for 2025: immersive AR/VR, ultra-high-definition protect themselves against tactics like cloud platforms will enhance network Wi-Fi revolution, transforming how we
streaming, and remote health care will Evil Twin attacks. performance, security, and scalabili- stay connected in the digital world. The
#1: AI-POWERED CLOUD-BASED NETWORK benefit from AI’s ability to optimize re-
MANAGEMENT al-time data flows, making mobile Wi-Fi #6: AI-OPTIMIZED CLOUD RESOURCE
more responsive and reliable. ALLOCATION
BY ERIC PLAM
AI will transform mobile Wi-Fi manage-
ment by enabling cloud-based sys- #4: EDGE COMPUTING AND AI-DRIVEN CLOUD AI-powered cloud platforms will enable Eric Plam is the President at SIMO, the company
tems to monitor, analyze, and optimize ENHANCEMENTS dynamic allocation of computing and behind Solis, which provides advanced mobile Wi-Fi
network traffic in real time. By 2025, network resources based on real-time hotspots designed for travelers and remote workers.
mobile networks will increasingly rely The rise of edge computing will bring demand and system requirements. With a career deeply rooted in the telecommunica-
on AI-driven solutions to predict and AI capabilities closer to the user, As mobile Wi-Fi networks connect tions and mobile technology sectors, Plam has been
resolve congestion issues, allocate enhancing cloud connectivity. By an ever-growing number of devices, instrumental in launching innovative connectivity
bandwidth dynamically, and adjust processing data at the network’s edge, the ability to scale efficiently will be solutions that address common travel and connec-
network parameters based on user AI can reduce latency and make faster critical. By 2025, AI systems will predict tivity challenges, such as inconsistent Wi-Fi, need
behavior and demand. This intelligent decisions for mobile Wi-Fi users. This network loads and automatically adjust for SIM cards, international data costs and more. His
management will lead to faster, more will be critical for applications like aug- cloud infrastructure to handle peak us- leadership continues to shape the future of global
reliable internet connections, especially mented reality (AR), virtual reality (VR), age periods without sacrificing perfor- connectivity, ensuring seamless, reliable access for
in densely populated areas or during and the Internet of Things (IoT), where mance. This will allow service providers users around the world.
high-traffic periods. real-time data processing is essential. to maintain high-quality connections,
In 2025, AI-driven cloud systems will reduce costs, and improve energy
#2: SEAMLESS INTEGRATION OF WI-FI AND intelligently manage data distribution, efficiency while ensuring a seamless
5G THROUGH THE CLOUD optimizing which data gets processed experience for users.
at the edge to improve performance.
Wi-Fi and cellular networks, partic- #7: GREEN CONNECTIVITY AND
ularly 5G, will become increasingly #5: AI-ENHANCED SECURITY AND PRIVACY SUSTAINABILITY
integrated by 2025, with AI-powered
cloud platforms playing a pivotal role. As mobile devices rely more on cloud Sustainability will be a major focus in
Mobile devices will use AI algorithms services, security and privacy become 2025, and AI-driven cloud connectivity
to determine the best network for each increasingly critical. AI-driven cloud will help reduce the environmental
task, whether it’s streaming, video connectivity will elevate security by impact of mobile networks. AI will
calling, or file transfers. AI will enable using machine learning models to optimize network usage by minimizing
dynamic switching between Wi-Fi and detect and block potential threats in energy consumption and dynamically
5G, ensuring uninterrupted connectiv- real-time. AI will identify anomalies in adjusting resource allocation based on
ity as users move between networks. user behavior, network traffic, and data demand. Additionally, cloud-based AI
This seamless handoff will enhance requests, offering robust protection systems will allow for remote mainte-
the overall user experience by making against cyberattacks. Additionally, AI nance and updates, reducing the need
transitions between networks invisible. will enhance privacy by dynamically for physical infrastructure and lowering
encrypting sensitive data based on the carbon footprint of network oper-
#3: THE IMPACT OF 5G AND 6G CLOUD- user profiles, ensuring that mobile ations. As global data traffic increases,
ENABLED SERVICES Wi-Fi users maintain control over their these AI-driven advancements will
personal information. play a critical role in promoting greener
With the rollout of 5G and the eventual mobile connectivity.
rise of 6G, AI-driven cloud connectivity Solutions like mobile hotspots, which
will make ultra-fast speeds and low create a secure, encrypted Wi-Fi

202 203
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

TEN WI-FI #4: WI-FI AND 5G CONVERGENCE #5: OPENROAMING Fi HaLow standard will develop as a
disruptive connectivity technology for

PREDICTIONS
The convergence of 5G/6G and Wi-Fi 2025 will continue to see an accelera- IoT, with the potential to transform the
will ensure seamless, high-quality tion for OpenRoaming as global adop- IoT landscape. With its extensive range,
connectivity by dynamically switching tion continues to grow. Moving forward superior penetration, and enhanced

FOR 2025
between the best available networks. on the vision to transform public and battery life, Wi-Fi HaLow is poised
In a smart city, for example, a person guest Wi-Fi user experience, and to revolutionize industries, including
moving from a Wi-Fi-rich office to a changing the way we connect to Wi-Fi, agriculture, smart cities, and manufac-
5G-powered urban area will experi- from remote communities, to universi- turing, improving efficiency and data
ence uninterrupted service, thanks ties, stadiums, retail chains, large city collection. Wi-Fi HaLow is ready for
Written by Tiago Rodrigues to technologies like OpenRoaming deployments and more. OpenRoaming primetime in the IoT ecosystem and is
CEO, Wireless Broadband Alliance (WBA) and Passpoint, which enable secure, capabilities are extending into the IoT a natural fit, especially for long-range,
automatic connections to trusted Wi-Fi space, with zero touch provisioning intelligent applications.
networks. 5G’s network slicing further of IoT devices, emergency calling and
enhances this by dedicating specific response and private cellular networks. As I look ahead to 2025, both the Wi-Fi
network resources to applications like At same time OpenRoaming expands ecosystem and enterprises globally
AR/VR and real-time gaming, which the opportunity for MNOs and MVNOs have an opportunity to redefine con-
As we head into 2025, the Wi-Fi in- adopting Wi-Fi 7 to enable real-time systems. Following the U.S. model, can integrate smoothly with Wi-Fi. As to incorporate Wi-Fi as part of their nectivity for a digital-first world. Now is
dustry has reached a pivotal moment. data collection from IoT devices for traf- other countries are anticipated to adopt enterprises and industries drive this wireless solutions to expand capac- the time for industry leaders, innovators
Demand for enhanced Wi-Fi user fic management, public safety, and en- similar AFC frameworks, accelerating convergence, 6G will add features ity and/or coverage, in particular for and policymakers to collaborate and
experience with seamless, secure, and vironmental monitoring. Public venues deployment in enterprise and pub- like terahertz frequencies for nearly indoor scenarios. push the boundaries of what Wi-Fi can
high-performance connectivity contin- like stadiums, airports, and convention lic spaces by ensuring devices can instant communication over wide and achieve. Together, we must prioritize
ues to surge across homes, enterprises centers will adopt Wi-Fi 7 early to man- operate at standard power levels in local areas. Edge computing, which #6: INCREASED IOT DEVICE CONNECTIVITY seamless interoperability, security,
and cities making collaboration within age the connectivity needs of thou- outdoor or high-traffic areas without processes data closer to its source to and scalability to meet the demands
our ecosystem more critical than ever sands of simultaneous users, providing compromising incumbent communi- reduce latency, will leverage both Wi-Fi IoT Evolution with Wi-Fi Generations of an increasingly connected society.
before. seamless streaming and data access cations. In Canada, AFC approval by and 5G/6G to offload tasks to the best and Wi-Fi Halow – As the number of By embracing innovation and foster-
for fans, travelers, and attendees. regulatory bodies will drive adoption in network, optimizing real-time perfor- IoT devices continues to grow, Wi-Fi ing partnerships, we can ensure that
Here are my ten predictions for Wi-Fi Within the consumer sector, tech-savvy urban areas and suburban networks, mance. Wi-Fi will dominate high-den- networks will be optimized to handle Wi-Fi remains the cornerstone of digital
in 2025 and beyond, which reflect how users and smart home aficionados will supporting widespread use in places sity areas like offices, while 5G/6G will large-scale device connections. Wi-Fi transformation, delivering value to busi-
Wi-Fi is evolving to meet the demands upgrade to Wi-Fi 7 routers to maximize like shopping centers, airports, and enable broad IoT deployments, paving 6’s and Wi-Fi 7’s ability to manage more nesses and empowering communities
of increased connectivity, higher the performance of their growing array sports venues. In Europe, the Euro- the way for innovations like smart cities simultaneous devices will become worldwide.
speeds, better security, the growing of connected devices, such as smart pean Telecommunications Standards and autonomous vehicles reliant on crucial in supporting smart homes,
number of devices connected to the appliances, security systems, and en- Institute (ETSI) will likely coordinate robust, ubiquitous connectivity. IoT, and smart cities. Further, the Wi-
internet and a greater focus on energy tertainment systems for 8K streaming region-specific AFC guidelines, balanc-
efficiency. and gaming. ing connectivity needs with incumbent
protections.
#1: WI-FI 7 ADOPTION #2: 6 GHZ EXPANSION AND AFC
#3: AI-DRIVEN NETWORK OPTIMIZATION BY TIAGO RODRIGUES
Early adopters of Wi-Fi 7 will include AFC (Automated Frequency Coordina-
Tiago Rodrigues is the CEO of the Wireless Broad-
technology-driven industries, smart tion) systems will see phased rollouts AI-powered routers and cloud plat-
band Alliance (WBA), responsible for leading the
home enthusiasts, and enterprises across multiple regions, particularly in forms will analyze traffic patterns, ad-
overall strategy, together with the WBA Board, and
needing high-density, high-speed the United States, Canada, the Eu- just bandwidth allocation, and manage
the operational planning for the WBA. Previously,
connectivity. Large tech companies ropean Union, and parts of Asia, as devices for optimal performance, par-
Tiago spent 15 years at the Portugal Telecom Group.
and advanced enterprises will be regulators increasingly approve 6 GHz ticularly in smart homes and IoT-heavy
He holds an M.Sc. in Economics and Management of
among the first to implement Wi-Fi 7 use for unlicensed Wi-Fi under AFC environments. As hardware becomes
Science and Technology.
in their office infrastructures to sup- management. In the United States, the increasingly commoditized, infrastruc-
port increased demands from remote FCC will lead AFC implementation for ture vendors create their respective
collaboration tools, IoT sensors, and standard-power Wi-Fi operations in the secret sauces to innovate and differen-
high-definition video conferencing. 6 GHz band, with licensed database tiate.
Smart cities will also lead the charge, administrators managing these AFC

204 205
 eBook: The GenAI leap: how networks are reinventing themselves

THE AI
ADVANTAGE:
OPTIMIZING
VIDEO
STREAMING
VIDEO IN 2025
1. The AI Advantage: Optimizing Video Streaming in 2025��������������������������������������������������������������������.207 Written by Romary Dupuis
Chief Data Officer, Viaccess-Orca

2. What’s on the Scene for Video in 2025?������������������������������������������������������������������������������������������������������������� .209

3. The Next Wave: How Operators Will Adapt to the Evolving TV Landscape����������������������.212
Reducing the cost of operations has the world without the potential for By combining cloud TV platform ana-
become critically important in the video encountering quality issues such as lytics with viewer consumption trends,
streaming industry. A major expendi- buffering, loss of connection, and more. media companies can analyze viewers’
ture for media companies today is the daily and weekly behaviors, enabling
investments they make in cloud ca- Recent advancements in AI promise to more accurate demand forecasting. AI
pacity to meet peak demand. While the be a game changer for video streaming can anticipate both routine usage and
cloud is inherently flexible and allows by eliminating the uncertainty in de- unexpected spikes in viewing — for
media companies to spin up servers to termining the precise amount of cloud example when major sports events
meet increased numbers of viewers, it capacity needed for media processing. are streamed or a blockbuster movie
is not instantaneous. premieres — enabling media compa-
#1: AI IS TRANSFORMING CLOUD RESOURCE nies to optimize their cloud resources,
Why does this matter? Live sports MANAGEMENT reduce costs, and enhance environ-
streaming has grown significantly mental efficiency.Although AI inevitably
in popularity, with the likes of Net- Through state-of-the-art AI models, consumes cloud resources and energy,
flix streaming major live sports and media companies can accurately its ability to optimize cloud resource
Peacock shattering streaming records predict CPU loads and cloud capaci- usage for TV services contributes to
last year when it delivered the first ty requirements, optimizing resource reducing the overall energy footprint
NFL playoff game available solely on a allocation and minimizing guesswork. of video streaming operations in the
streaming service. To handle high-de- To achieve this, media companies must cloud.
mand sports events, media companies collect system metrics from a cloud
must either run at 100% capacity year- TV platform and analyze performance Machine learning (ML) and deep learn-
round or try to estimate future demand data from Kubernetes clusters. Tracking ing (DL) techniques are key to optimiz-
and then provision additional nodes, processing load, memory load, network ing cloud resource management and
as needed. Without sufficient cloud load, and other metrics every 30 sec- maintaining the best possible video
capacity, events like these wouldn’t be onds to one minute provides a detailed streaming quality. ML excels at detect-
able to reach millions of fans around understanding of system performance. ing daily and weekly patterns, such as

206 207
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

increased demand during weekends

and peak hours, as well as more subtle
trends that become apparent only
allows media companies to allocate the
right infrastructure to meet their future
programming needs, ensuring seam-
a reduced carbon footprint.But trade-
offs between accuracy and cost need
to be assessed and balanced carefully,
WHAT’S ON THE SCENE FOR VIDEO IN 2025?
when analyzed alongside consump- less performance and quality regard- especially when employing the latest
tion patterns. While traditional ML is less of peaks and troughs. GenAI techniques running on modern Written by Martin Prins
effective at forecasting platform load in GPUs. Head of Product, Media Distillery
advance, DL techniques — based on #2: THE EVOLUTION OF AI FOR VIDEO
large language models (LLMs) — offer STREAMING In the video streaming environment,
even greater accuracy. However, these three critical pillars lie at the core of
advanced methods do come with high- AI models will evolve over the next few viewer satisfaction: video experience,
er computational costs. years and manifest in various forms. content quality, and price. Efforts to
This includes agents that analyze TV optimize TCO directly impact these
DL also plays a crucial role in extract- platform performance through auto- areas. By leveraging AI to improve
ing and enriching metadata from TV matic root cause analysis of poor qual- resource allocation, media companies
programs. It utilizes techniques such as ity of experience, recommend cloud can enhance video quality and stream-
natural language processing and image architecture improvements, dynamical- ing reliability, while simultaneously
processing. It can also be applied ly scale cloud infrastructure, fine-tune achieving cost savings — which opens
to simple or multi-variate timeseries video parameters and throughput up opportunities to refine pricing mod-
prediction. Media companies that capacities, and act as security guards els, invest in superior content, and even
automate the scalability of resources against cybersecurity threats and reduce the impact on the environment.
also take advantage of LLMs’ chain of piracy. These agents will also enhance Ultimately, an AI-driven approach to
thoughts capacity, enabling them to customer service by providing 24/7 cloud resource management strength-
take appropriate actions such as scal- support through humanized interfac- ens subscriber loyalty and drives
ing Kubernetes clusters up or down. es, significantly improving customer sustainable growth in an increasingly
satisfaction. competitive streaming landscape.
Ultimately, AI provides a level of insight
that goes beyond just predicting pat- Conclusion
terns. It can also anticipate potential
outliers, such as disruptive scheduling The benefits of using AI for cloud
choices — news events, new movie resource management are significant.
releases, series finales, sports events, Early tests indicate that AI can help
and more — and accurately forecast operators use 30% less infrastructure,
how these will affect capacity. This leading to substantial cost savings and

BY ROMARY DUPUIS
In this article, Martin Prins of Media #1: A STRONGER FOCUS ON VIEWER the minds of customers; it’s particularly
Romary Dupuis is Chief Data Officer at Viaccess-Orca.
Distillery explores key trends shaping ENGAGEMENT important in today’s crowded content
With more than 20 years of experience in engineer-
the future of video services in 2025, market.
ing, management, and strategic innovation for the
from prioritizing viewer engagement Increasing viewer engagement will be
telecommunications industry, Romary Dupuis brings
and enriching sports viewing experi- the top priority for many video services We’ll be seeing new advances in strat-
a solid understanding of big data and AI to Viac-
ences to leveraging AI-powered fea- in 2025, and regarded as a business egies and tactics to enhance discov-
cess-Orca (VO). Prior to being the chief data officer
tures and catering to diverse audience imperative. This will lead to operators erability, in recognition of the fact that
at VO, Dupuis led state-of-the-art strategic programs
needs. He also discusses how striking expending more resources, including many consumers today actually face
for Orange in Europe and Asia, and founded several
a balance between unbundling and technology, dedicated to galvanizing too much choice. A viewer’s patience
startups in big data and AI in Japan.
rebundling strategies, as well as em- customer relationships, striving to make and acceptance can be pushed to the
bracing advertising innovations, will be them deeper and broader. I think we’ll limit if overwhelmed by too much con-
crucial for success. see unprecedented creativity as opera- tent to choose from. There is the same
tors employ novel strategies and tactics risk to customer happiness if finding
to create a feeling of a service’s value in something immediately appealing to

208 209
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

watch takes too long, or if the deci- the capabilities of AI will continue. It’s comes to pricing and bundle flexibility. for pay-TV operator, streamer, and most to the content the viewer is watch- consumers will make a more conscious
sion-making process is lengthier than key to remember, though, customers The services that prevail will be those importantly, consumer. ing, or what brands or locations are choice about how much advertising
expected - this ‘thinking time’ colorfully are not buying AI, they’re looking for that cater to different viewer needs. visible in the content. We’ll also see they will tolerate and what they’re will-
described by today’s researchers as solutions to their problems or unfulfilled #6: ADVERTISING ON THE RISE more shoppable ad experiences on ing to spend on specific video services.
‘dwell time.’ Making sure that viewers needs. In 2025 the difference between #5: FROM UN-BUNDLING TO REBUNDLING TV or companion app transactional The video service providers that offer
can have fast access to relevant con- AI hysteria and real-world application Advertising will play a more important purchasing. There will continue to be flexibility in that sense are most likely
tent will be very important for opera- will be stark. Everyone in our industry Not all that long ago there was a bla- role in 2025. I predict that there will be a widespread appeal for a free or very to keep viewers engaged. They can
tors, and they’ll harness technology in has been stretching their imagina- tantly adversarial relationship between many innovations in privacy-friendly low-cost ad-supported product tier for leverage new monetization options like
new ways to improve discovery. tions to explore where AI can make a operators and direct-to-consumer contextual advertising solutions. New streaming services. But in parallel there advertising, or creating premium prod-
difference both operationally and in streaming services. D2C streamers commercial paradigms will include will be a desire on the part of viewers ucts that offer ad-skipping for catchup/
#2: RICHER SPORTS VIEWING EXPERIENCES terms of user experience. I think that declared their independence from contextual ads, which closely relate to avoid ad fatigue. So I believe that replay content.
in 12 months’ time, there will be some traditional pay-TV operators, that they
Sports rights holders in particular are breathtaking AI-based innovations. would surely replace the incumbents in
determined to improve their return on We’ll see some surprises, and discover consumers’ hearts, and in many cases,
investment in content rights by provid- that AI can be harnessed to accomplish the appeal of the shiny upstarts would
ing fans with an array of additional con- things that today seem impossible. incite cord-cutting. However, as our in- BY MARTIN PRINS
tent that complements televised live dustry, and the role of streaming within
Martin Prins is Head of Product at Media Distill-
matches. Premium sports TV is a pow- #4: SUCCESS RELIES ON ADDRESSING it, have evolved, that power dynamic
ery, which helps video services improve viewer
erful driver of engagement, but it’s also DIFFERENT VIEWERS’ NEEDS has shifted dramatically; it’s now widely
engagement and increase monetization opportuni-
by far the priciest to get transmission recognised that a partnership between
ties, through automated AI-driven content analysis.
rights to, and also the most expensive Short form clips and video previews trusted incumbent providers and their
With over 16 years of experience in streaming and
to deliver in ultra-high-quality, espe- will increasingly prove their value as a ambitious streamer counterparts, is
broadcast media, Martin is responsible for Media
cially for major live remote broadcasts. discovery tool and indeed conversion symbiotic and powerful. Typically, the
Distillery’s product portfolio and strategy.
Operators are doubling down on their catalyst. YouTube, Instagram and others operator is a known and trusted brand
efforts to maximize viewers’ perceived have irreversibly moved the consumer within a household, providing essential
value, and optimizing the experience of behaviour needle, and redefined what services including fixed and mobile
watching sports on TV. Fans have an we mean by ’content‘ in our industry. telecoms and broadband. Streaming
almost insatiable hunger for informa- Short form video is now serious busi- providers, once considered disruptors,
tion about their sports, leagues, teams ness, and full-motion clips will increas- have since become part of the broad-
and players. In 2025 and beyond, they ingly figure in the UI of video services, casting establishment. They’re invest-
can look forward to being presented as they lead to more viewing compared ing heavily in content, including sports
with an unprecedented array of game to static images or text. It’s increasingly and film rights. Importantly, they’re
statistics, player data and profiles, and evident that short form content clips ploughing millions into original produc-
other features like Multiview designed rapidly help people decide if a show tion, and they’re being rewarded for it.
to enhance live match broadcasts. We is of interest. Research conducted by Headline grabbing, and increasingly
know sport fans want to see everything Google shows that for Gen Z viewers, award-winning, films and series are
related to their favorite clubs: wheth- a video snippet can lead to a high con- pulling in the viewers. But by forging
er it’s the interview with a player or version rate, with almost 60% of young partnerships with the longstanding
the pre/post-match discussion in talk people going on to watch the full- familiar players like telcos, today’s
shows. Currently viewers need to look length programme. Static text-heavy streamers can instantly expand their
in different locations. With the help of UIs may well seem quite old-school reach to millions of subscribers. As part
AI-based topic detection, operators can before long. of an operator bundle, the streamer is
finally also give viewers easily access somewhat shielded from the inherent
to all this shoulder content in their Audiences are changing, diverging unpredictability of their no-contract,
inventory. widely in their expectations, and fast. click-to-cancel existence. And for their
Gen Z’ers have totally different require- part, the operator also benefits from
#3: GROWTH IN ADOPTION OF AI-POWERED ments from those of baby boomers. the collaborative relationship as it can
FEATURES – WHERE THEY ADD VALUE They also have different attitudes deliver the hottest new content within
towards linear and on-demand, as imaginatively curated and attractively
The tsunami of ideas and hype about well as divergent expectations when it priced bundles. The net result is a win

210 211
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

THE NEXT WAVE: HOW OPERATORS WILL ADAPT can help operators see significant cost
savings in subscriber management and
long committed to promoting techno-
logical innovations that make major
RECOGNISE THAT COMMERCIAL BUNDLING
WITH THIRD PARTIES INCREASES
CUSTOMERS’ PERCEIVED SERVICE VALUE,

TO THE EVOLVING TV LANDSCAPE

customer support. TV services available to the widest
possible population. Voice technol- AND CUTS CHURN
Today’s set-top box remote control ogy has advanced, but not very fast,
units as well as the set-top boxes and certainly not everywhere. Now, The shift to tighter integration be-
themselves have a built-in microphone, user interface technology and AI are tween pay-TV operator and third-par-
Written by Kai-Christian Borchers and users can have natural conversa- combining to enable dramatic progress ty streaming provider will intensify.
Founder & Managing Director, 3 Screen Solutions tions with sound bars or boxes without in this field. Accessibility menus now Not that long ago the buzzword, and
even having to lift a finger. During the figure much higher in TV user interfac- hoped-for cure-all, was service un-
user’s interaction with the service, es. Technology is playing an important bundling. Direct to consumer stream-
they’ll be able to easily toggle between part in helping viewers with disabilities ing services were riding high while
speaking and text-based interaction. In get more from their TV services. incumbent pay-TV providers were
2025 I believe we’ll see huge advances fiercely trying to stop cord-cutting. But
in deep integration of voice into the #3: OPERATORS WILL TAKE SMART HOME intervening economic and commer-
broader user interface and experience. SERIOUSLY IN PURSUIT OF NEW REVENUE cial realities of the past few years, and
This is in large part thanks to the fact STREAMS relentless video entertainment market
that AI technology is fast becoming fragmentation, have led to the new in-
practical and easy to use. Finding new revenue streams to dustry-wide marketing pivot of re-bun-
complement their existing offerings will dling. It’s like content super-aggrega-
#2: CUIS WILL TRANSFORM PERCEIVED remain a high priority for operators in tion has been replaced by ‘commercial
VALUE FOR CUSTOMERS WITH DISABILITIES 2025. Products and services relating super-aggregation.’ Novel approaches
to smart home security, convenience to subscription service packaging by
Highly capable CUIs can transform the and energy conservation will increas- innovative operators like TELUS and
value of a pay-TV service for people ingly be rolled out. Operators have others have powerfully demonstrated
with disabilities. About 20% of the pop- been working on smart home services that user-centric and flexible market-
ulation has some form of impairment. for many years, and exploring ways to ing strategies are working; the proof
There are millions of blind and partially broaden their customer offerings to is in the rising subscriber numbers.
sighted people consuming TV services include bundled-in smart home tech Operators have begun to realise that
every day. A good CUI can transform and services. But many operators have including popular streaming services in
these viewers’ ability to quickly find faced challenges with platform matu- monthly pay-TV, telephony and broad-
their chosen entertainment. But it can rity, technology standards and finding band packages gives customers a level
also empower these customers to man- business models that work. As a result, of flexibility and convenience that they
age their subscriptions, take advantage operator-backed smart home capability appreciate. Moreover, the customer
of targeted promotions and offers, and has a long way to go to achieve its full saves money.
access other meaningful customer commercial potential. However recent
support more easily. Also, people with initiatives from companies like Apple, Studies for telcos have shown that
reduced or no hearing can benefit from Google, Amazon, smart TV manufac- customers churn less when third-party
In this article, Kai-Christian Borchers of #1: EVERYONE WILL BE TALKING ABOUT ing, as we all know, is far more natural AI-generated subtitles for the many TV turers, and others, raise the possibility subscription services are included on
3SS lays out his vision for the TV and CONVERSATIONAL USER INTERFACES and easier than browsing, typing and programmes that currently do not offer for operators of more streamlined the monthly telecom bill. People are
video entertainment industry in 2025. zapping, bringing more natural engage- this feature. technical incorporation of smart home increasingly aware that managing one
From the rise of conversational user Conversational User Interfaces, or ment to viewers of all ages and de- technology e.g. Matter smart home bill is easier than managing five, six
interfaces to the increasing importance CUIs, will be a big focus in 2025. mographics. We’re really talking about This is part of a larger prioritization of connectivity standard. This could lead or seven each month. It’s also cheap-
of smart home integration, Borchers Operators will step up their efforts to the democratisation of the TV UX. In accessibility in the television and tele- to significant new revenue streams- er. Creatively packaged offers and
delves into the strategies that operators simplify content discovery and custom- addition to customers finding their coms industries. For all regulated oper- beyond TV, unlocking new business promotions are giving operators new
will employ to enhance customer expe- er self-care using voice input. Em- desired content faster, they’ll be able to ators, particularly government-backed by making everyone’s life simpler and churn-busting tools, with overall cost
rience, drive growth, and stay ahead of powering viewers to simply talk aloud speak to the service’s voice assistant telcos, accessibility is a licence smarter. savings for consumers.
the curve. with their pay-TV service is without that can help them solve any problems condition, and core value. Broadcasting
doubt one of the more compelling and with the service. Simultaneously, this and communications regulators have #4: OPERATORS WILL INCREASINGLY OTT partners also benefit from this
value-adding use cases for AI. Speak- CUI-enabled self-care, fuelled with AI,

212 213
The Fast Mode Trends and Outlook 2025

symbiotic relationship with operators. #5: LOW-COST OTT DEVICES WILL BE AT THE available, ideally featuring a UI that
The streamers increase their reach by HEART OF ANTI-CHURN STRATEGIES AND mirrors the premium user experience,
partnering with incumbents. But these LOWERING CUSTOMER ACQUISITION COST telcos and other providers can phase
third-party services recognize that the out support for old equipment while
biggest advantages of being part of a Operators will accelerate the de- retaining the connection with the cus-
telco super-bundle are better sticki- ployment of strategies to expand tomer, and their recurring subscription
ness, and a steady income that offers their reach and grow the customer revenues.
some protection against the volatile di- base. We’re seeing a marked trend of
rect-to-consumer reality. New US reg- operators launching low-cost Android Operators further benefit from offering
ulation from the FCC mandating that devices for customers who have for simple plug-and-play OTT streaming
OTT services must enable one-click whatever reason resisted a full pay-TV devices like streaming sticks, due to
cancellation (‘click to cancel’) will add a subscription. their very low customer acquisition

SUSTAINABILITY
new layer of commercial unpredictabil- costs. A cable or satellite TV installa-
ity for streamers. Being in a pay-TV or Importantly, operators also see that in- tion is resource-hungry, in technical,
telco super-bundle, streamers can be expensive OTT devices are an opportu- logistical and human terms, and in
somewhat shielded from churn. nity to reduce churn. One particular tar- time. By comparison, an OTT streaming
get group are longstanding subscribers device can be ordered, delivered and
who have legacy pay-TV set-top boxes installed in a matter of days, with no
in their homes but who are reluctant to need for in-person installation, set-up
upgrade to a next-generation offering. and onboarding support. 1. Spotlight on Sustainability, Scalability, and Security������������������������������������������������������������������������������� .216
By making low-cost Android devices

2. The Age of AI Requires a Rethinking of Networks������������������������������������������������������������������������������������� .219

BY KAI-CHRISTIAN BORCHERS

Kai-Christian Borchers co-founded 3SS (3 Screen

Solutions) in Stuttgart in 2009. He created Germany’s
first online DVD rental company in 2003. From 1997
to 2000, he led sales and marketing at his first
co-founded enterprise, aeonware AG. He has helped
build numerous other tech companies and has
worked with Volkswagen.

214 215
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

SPOTLIGHT ON SUSTAINABILITY, indelible mark over the past year, trans-

forming how we work and boosting
transformation, helping organisations
manage growing data demands while
ational resilience and ensuring compli-
ance with regulatory standards.

SCALABILITY, AND SECURITY

productivity in ways we once only improving energy efficiency. We are
imagined. But there’s no denying its seeing many businesses incorporating To address these needs, organisations
immense energy demands present a these solutions into their procurement should focus on key infrastructure
challenge for organisations striving to processes as they are drawn by the elements that enable smarter data
meet sustainability goals. For example, model’s ability to scale up or down as management:
Written by Rodney Kinchington training large language models like needed without the hefty upfront costs • A dense multi- and hybrid-cloud
Managing Director, AMEA, BT Business ChatGPT consumes electricity is equiv- of building infrastructure. This flexibil- connectivity that ensures multiple
alent to powering hundreds of homes ity and cost-effectiveness make the paths for seamless workload shifts
annually [1]. As adoption of GenAI ac- model particularly appealing in today’s during cloud service disruptions to
celerates, the question isn’t just about dynamic business environment. sustain operations and comply with
its potential, but it’s also about how we regulatory requirements;
embrace it responsibly. Right-sizing is a key enabler of this ap- • An integrated safe and secured
proach; this means that organisations ecosystem that is deployable
Striking this balance requires a fresh only pay for the resources they actually instantly without the need for
approach to infrastructure and energy use – be it cloud storage or network technician visits or manual
management. AI workloads often be- capacity. This goes a long way in terms connections;
have differently from traditional traffic, of streamlining operations and foster- • An on-demand flexible connectivity
causing unexpected power surges that ing agility. that adapts and meets various
strain networks. This means that tech requirements and regulations, and
leaders must think creatively, finding As data demands continue to rise, • An easy-to-use application capable
ways to optimise efficiency and reduce combining right-sizing with as-a-ser- of running multiple services
environmental impact while staying vice models offers businesses an effi- simultaneously without needing
aligned with customers’ ESG commit- cient and adaptable pathway to digital dedicated ports or access for each
ments. After all, innovation shouldn’t transformation, all while keeping sus- service.
come at the planet’s expense. tainability at the forefront. According
to Forrester, 88% of businesses expect By putting these elements in place,
However, some solutions are already sustainability gains from adopting as- organisations can truly harness the
within reach. Scaling network switches a-service models, with 65% projecting power of their data. With greater
to match traffic demands and adopting cost savings of more than 10% [3]. By accessibility and smoother integra-
distributed architectures can reduce providing access to the latest hardware tion, teams can make faster, more
energy usage by up to 25% compared without overprovisioning, these models informed decisions, driving meaningful
to centralised systems [2]. Additionally, address mounting concerns about improvements across the board. The
partnering with sustainable cloud pro- environmental impact and resource result? Stronger business outcomes,
viders and employing real-time energy efficiency. enhanced operational efficiency, and
The past year has been a transforma- dynamic landscape, I often rely on three principles are shaping the way monitoring tools are other practical measurable progress on key perfor-
tive period for the tech industry, driven what I call the ‘six S’s’ framework: sus- businesses approach innovation. From steps that organisations can take to #3: FUTURE PROOFING OPERATIONS mance indicators, which will all help to
by rapid advancements in generative tainability, scalability, security, stability, enabling responsible adoption of gen- further minimise environmental impact. THROUGH BETTER DATA MANAGEMENT set the stage for sustained success.
AI and data-driven strategies. These in- sovereignty, and speed. This framework erative AI to enhancing scalability and When implemented effectively, busi- (SECURITY)
novations are reshaping how business- provides a practical lens through which strengthening security, these pillars are nesses can adopt GenAI responsibly, As the tech landscape continues to
es operate, creating unprecedented leaders can address emerging chal- critical for addressing today’s challeng- driving innovation while maintaining a The growing reliance on IoT applica- evolve, tech leaders are uniquely posi-
opportunities for growth and transfor- lenges while aligning innovation with es and seizing tomorrow’s opportuni- strong commitment to sustainability. tions and edge computing has made tioned to drive transformative change.
mation. Yet, this rapid progress comes accountability. By prioritising the first ties. seamless and secure data movement When sustainability, scalability, and
with its share of challenges, particularly three pillars, sustainability, scalability, #2: RIGHT-SIZING IS THE WAY TO GO a critical priority for organisations. security become your modus operandi,
in navigating complexity and fostering and security, leaders can position their #1: RESPONSIBLE USE OF GENERATIVE AI (SCALABILITY) Whether transferring data from edge the remaining S’s (stability, sovereignty,
responsible innovation. organisations for long-term success. (SUSTAINABILITY) computing or between different cloud and speed) tend to follow naturally.
The ‘as-a-service’ model is gaining mo- environments, robust data manage- This is because by embedding these
To help organisations thrive in this In this article, I’ll explore how these Generative AI (GenAI) has made an mentum as a go-to solution for digital ment is essential to maintaining oper- the first three S’s, organisations are

216 217
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

opening themselves to end-to-end

performance monitoring, enhanced
and swift issue resolution across the
network. With that in mind, I’m excited
ful progress for your businesses, the
industry, and the environment. THE AGE OF AI
REQUIRES A
application visibility, and intelligent to see how you tackle these challenges
control, enabling smarter decisions in the year ahead, driving meaning-

RETHINKING
References
[1] How much electricity do AI generators consume? The Verge: www.theverge.com/24066646/ai-electricity-energy-watts-generative-
consumption

OF NETWORKS
[2] Ehsan Ahvar, Anne-Cécile Orgerie and Adrien Lebre. ‘Estimating Energy Consumption of Cloud, Fog and Edge Computing Infrastructures’, IEEE
Transactions on Sustainable Computing, 2022, 7 (2), pp.277-288
[3] Forrester, ‘Sustainability Starts Here: Accelerating Sustainable IT Programs With As-a-Service Models’ (February, 2022)

Written by Charles Yeomans

BY RODNEY KINCHINGTON Co-Founder & CEO, Atombeam

Rodney Kinchington is the Managing Director for

BT Business in Asia-Pacific, Middle East and Africa.
Rodney is a dynamic and results-oriented leader
Artificial intelligence (AI) is revolu-
who has been at the forefront of BT’s growth in the
tionizing industries, unlocking un-
Asia-Pacific, Middle East, and Africa (AMEA) region
precedented potential for automation,
since assuming the role of Managing Director in
creativity, and efficiency. However, this
2021. His strategic acumen and relentless drive have
transformative power comes at a price:
played a pivotal role in positioning BT as a leader in
the vast amount of data AI needs to
the AMEA market.
process, move, and operate is straining
networks like never before. Traditional
networks simply aren’t built to handle
this. The Wall Street Journal recently
identified the capacity of data center
“plumbing,” or data pipes, as a key
issue facing AI.

To truly support AI’s growth without

hitting a technological and economic
wall, we need to reinvent the very idea
of networks. The future lies in creating
systems that move data faster, use less
power, and remain secure - and do so
without breaking the bank. Much easier
said than done.

#1: RECOGNIZING THE PROBLEM: DATA

OVERLOAD

AI thrives on data. Whether it’s training networks are already showing signs of upgrades are already major challenges.
a large language model (LLM), analyz- struggling to keep up with this growing This is particularly evident in data cen-
ing millions of images, or optimizing demand. ters, which for better or worse, serve as
supply chains, the process requires the nerve centers of the AI revolution.
moving immense volumes of informa- Bandwidth bottlenecks, latency issues, These facilities handle the heavy lifting
tion across networks. But traditional and soaring costs for infrastructure of storing, processing, and transmitting

218 219
The Fast Mode Trends and Outlook 2025 eBook: The GenAI leap: how networks are reinventing themselves

data. But their ever-growing energy randomly accessible while encoded, a based on traffic predictions, AI systems innovation while reducing barriers to tonomous systems to supporting global we live and work, the networks that
consumption - already accounting for data center would no longer need to can keep networks running smoothly, entry. collaborations in scientific research, support it must evolve just as dra-
about 1% of global electricity use - is compress and decompress, or encrypt even during peak usage. the possibilities are endless. Compact, matically. The challenge is clear, but
approaching unsustainable levels. and decrypt, data for it to be used, The reinvention of networks will also secure, and efficient data transmission so is the opportunity—one that, if
eliminating major computational over- AI algorithms can also be used to pro- redefine what’s possible with AI itself. will pave the way for breakthroughs approached correctly, could shape the
The environmental impact of these head typical for data centers. cess and compress data more intelli- If more data is available, then greater that were previously out of reach. future of AI and the digital world at
power demands is also significant. gently, eliminating redundancies and things can be accomplished. From en- large.
Data centers require constant cool- Moreover, integrating encryption into enhancing efficiency. For example, ma- abling real-time decision-making in au- As AI continues to transform the way
ing to function properly, contributing the same computing step as com- chine learning models could be trained
to both carbon emissions and high pacting enhances security without to create smart compression tech-
operational costs. The situation is adding complexity. Compacting and niques tailored to specific workloads.
exacerbated by the rapid growth of AI encrypting data at the source creates This would ensure that no bandwidth is
applications, which are increasingly a streamlined flow that is both faster wasted and that more data can be sent BY CHARLES YEOMANS
complex and data-hungry. Without new and safer, making it more difficult for and utilized.
Charles Yeomans is the co-founder and CEO of
approaches, these challenges will only attackers to obtain. The ability to se-
Atombeam. With 25+ years in executive roles and
continue to increase, making it even cure data while making it more efficient Beyond optimization, AI enables
investment banking, Charles has led multiple firms
more difficult to strive for sustainability. is critical as AI applications become entirely new architectures, such as
and founded successful companies, including
increasingly embedded in sensitive decentralized networks where deci-
major insurance brokerages. He’s a former U.S. Navy
More efficient networks mean lower sectors like healthcare, finance, and sion-making is distributed rather than
intelligence officer with an AB from Kenyon and an
energy consumption, which trans- national security. Stored data can be centralized. These architectures could
MBA from Stanford.
lates into reduced greenhouse gas secure while still being accessible and one day reduce latency, improve fault
emissions. For companies aiming to reduced in size. tolerance, and allow for seamless
meet sustainability goals, reinventing expansion. AI’s ability to automate
networks is not just a technical imper- This rethinking of how data is struc- and streamline network operations
ative but a moral one. Without a shift in tured and transmitted has ripple effects could fundamentally change how data
approach, AI’s energy demands could beyond efficiency as well. Improving moves, making networks more resilient
eventually outpace the gains it brings efficiency can empower the same and adaptable.
in efficiency. networks to handle more AI work-
loads, which in turn will support more The opportunity ahead
#2: TAKING AN ALTERNATE APPROACH TO real-time decision-making, and reduce
NETWORKING the dependency on building new net- The need to reinvent networks for AI
works or constructing additional data isn’t just a technical challenge; it’s an
To address these challenges, the focus centers. This is a necessary shift—not opportunity. By embracing new ap-
needs to shift from merely increasing just to accommodate AI’s rapid ex- proaches to data handling and lever-
hardware capacity to rethinking how pansion, but to ensure that its benefits aging AI to optimize networks, systems
data itself is handled. One promising don’t come at an unsustainable cost. can be created that meet the demands
approach is compacting and encrypt- of tomorrow without overwhelming the
ing data to reduce its size, without loss #3: AI’S ROLE IN THE SOLUTION resources of today.
of data. When data is compacted, it be-
comes much smaller in a few millionths Interestingly, AI itself can play a pivotal By focusing on the efficiency of the
of a second, reducing the bandwidth role in reinventing networks. AI-driven data itself - rather than just upgrading
needed to transmit it. This leads to systems can analyze traffic patterns hardware and facilities - significant
faster data transfer rates, typically in real time to optimize data routing, savings in time, energy, and cost can
by 4x, and significantly lower energy ensuring that information travels the be realized. Compacting, encrypting,
consumption. fastest and most efficient paths. and optimizing data flow allows for
faster processing, streamlined oper-
Compacting data can also eliminate AI also has the capability to predict ations, and a lighter environmental
unnecessary computational steps and prevent network congestion before footprint. These benefits ripple across
along the way. For example, if com- it occurs. By dynamically reallocating industries, from manufacturing to
pacted data could be searchable and bandwidth or rerouting data streams healthcare to finance, accelerating

220 221
ABOUT THE FAST MODE
The Fast Mode is a leading independent research and media brand,
providing breaking news, analysis and insights for the global IT/
telecommunications sector. Founded in 2013, The Fast Mode has
grown to a world class brand with a global reach spanning millions
of readers annually and is currently one of the top five global digital
publications in the industry.

The Fast Mode partners with leading technology companies ranging

from start-ups to large multi-nationals on the research, analysis
and publishing of breakthrough ideas and forward concepts across
key verticals in telecoms and IT, with focus on IP/optical connec-
tivity, network intelligence, security, content delivery, the Cloud,
Internet-of-Everything, customer experience and operator service
innovations.

www.thefastmode.com

andrea@thefastmode.com

Copyright 2025 The Fast Mode \ v1.1 \ 11.04.2025

222