AGILE TELCO ONTREND OCTOBER 2026
Will telcos allow network planning to automate to innovate?
COMMENT
Fail to plan, plan to fail
The old management cliché of ‘failing to plan is planning to fail’ has never been more apt than when applied to telcos’ network planning efforts. The ‘build-it-and-they-will-come’ era ended in about 2010 since when telcos have been scrabbling to maximise returns on every dollar they invest in network infrastructure. Network planning has meant that 5G investments have been reasonably tightly targeted to ensure capacity matches demand and revenue can be generated to justify network build.
Engineers and telco systems have developed finely-tuned capabilities to support both optical and wireless network construction. The need to optimise expenditure has become paramount and the greater complexity of the next generation of cellular infrastructure in the form of 6G coupled with changing traffic and usage profiles is putting pressure on operators to innovate their way out of a growing crisis.
AI is placing huge new demands on networks and it’s unlikely that human engineers will be able to manually design and deploy the networks of the future. What’s needed is greater reliance on automated network planning functions. This enables telcos to accelerate and streamline planning through automation, enabling lower cost, lower friction approaches to rolling-out the capacity that users need.
Of course, this critical telco function, can’t simply be handed over to an automated system and assumed to be working optimally. This is essential telco enablement so before rushing headlong into AI-enabled planning, telcos will iterate, experiment and trial in parallel environments using digital twins and virtualised models to check AI’s learning and to ensure the next level of network planning is as robust and telco-grade as earlier versions.
It’s all about creating comfort and then reaping the benefits.
INTERVIEW
“Telcos need time to experiment with AI and learn from both successes and failures.”
George Malim: How far will telcos go in allowing AI into the network planning function?
Benjamin Vigouroux: We see AI playing a vital role in managing bandwidth, ensuring network availability and enhancing threat detection. We already use AI and quantum computing to optimise network performance through digital twins, helping us build more resilient, efficient and secure networks. AI is also helping us reduce power consumption across our points of presence (PoPs) by up to 15%. These applications are giving us a clearer understanding of network performance and helping our networks respond more effectively to changing demands.
GM: What needs to happen to make telcos comfortable to do this?
BV: Telcos need time to experiment with AI and learn from both successes and failures. They need to be able to expand, adapt or stop proofs of concepts based on those findings. Crucially, this experimentation should be guided by customer needs, rather than the latest AI trends.
Experimentation must also take place within a framework of control and trust, supported by human oversight, robust governance and trusted infrastructure. As AI becomes more embedded in network planning, people have to continue to set the direction and remain accountable, with AI supporting the execution.
GM: What are the main benefits application of AI to networking planning can unlock?
BV: AI has the potential to address longstanding network planning challenges, including how to adjust capacity dynamically as demand changes. Today, network planning still relies heavily on predictable patterns, such as anticipating increased traffic during major events. During the Paris Olympics, for example, Orange prepared for bandwidth spikes as spectators streamed content during breaks.
In the future, AI could help networks adapt to changing conditions in real-time. Through intelligent traffic management, predictive maintenance and real-time monitoring, telcos could improve network resilience and service quality while making more efficient investment choices.
GM: What are the impacts of carrying customer AI traffic on network planning? Is enough understood about future traffic demands and how this should influence network planning?
BV: Presently, customer AI traffic remains limited and is largely text-based, so its impact on network infrastructure is relatively modest. However, as AI-generated video and audio content become more common, we will see bandwidth requirements rise significantly. Real-time content generation will also place greater pressure on latency. This means the infrastructure needed to support the next generation of AI will require high-performance data transport, low latency connectivity and more edge computing capacity.
Looking further ahead, telcos must also prepare for the emergence of the ‘Internet of Agents’, where billions of autonomous AI entities interact at scale. This could create less predictable and more volatile traffic patterns, which networks will need to anticipate and manage.
MARKET REPORT
Can AI build enough trust to support telcos’ network planning needs?
No one is suggesting that it’s time to execute mass-market 6G deployments yet but it’s certainly time to get ready for massive increases in network infrastructure to support AI and other use cases. New capacity, both fibre and cellular, will be needed in locations that match demand and traffic profiles as AI workloads proliferate. In fact, STL Partners has projected that telcos are set to deliver more than 3.5GW of AI factory and edge capacity by 2030. Investments in AI factories are likely to exceed those made in network edge infrastructure, the firm says.
Even so, telcos still need to invest heavily in network infrastructure and will work to ensure they plan their networks correctly so investment is directed to the areas of greatest profit potential. A lot of this can be modelled efficiently using customer data processed via AI tools and this, alongside telcos’ deep networking planning experience, helps to ensure optimised network construction. Combining the data insights with the planning expertise provides a firm basis upon which to automate perhaps even the majority of the network planning function. Automation is set to become increasingly attractive in any case because network engineers are expensive so systems that can automate planning are being assessed widely.
Network planning has been one of the crown jewels at the heart of a telco and, as such, has seen reluctance to automate. Now though, the scale of roll-outs and the need to account accurately for return on investments on a day-by-day basis means automation is the only way to handle complexity and scale. As these demands have become unavoidable, automation is now seen as good enough to rely on for this essential telco task and telcos are engaging to take the strain from engineers and optimise their planning and other functions through automation of network planning.
The tech that came in from the cold
Much here isn’t new. The telecoms industry in general has been talking overtly about utilising AI in its operations for more than a decade but it has taken the current mainstream technology hypecycle surrounding AI to sharpen the focus on how the technology might be used. 5G standalone (SA) and open radio access network (RAN) adoption are increasing the number of network settings telcos must manage and telcos are responding by moving AI from limited trials into the network, operations and customer-facing systems.
This is borne out by research from Mordor Intelligence that predicts AI in the telecoms market’s size is expected to increase from US$4.18 billion in 2025 to US$5.78 billion in 2026 and reach US$29.17 billion by 2031, expanding at a CAGR of 38.23% over the 2026-2031 period. Both Ericsson and Nokia have launched AI in RAN systems with Ericsson unveiling its AI in RAN software subscription offering in June 2026. This reports gains of up to 20% in downlink throughput and 10% in spectral efficiency, using existing hardware. The following month, Nokia introduced its AI-RAN platform, claiming 20% spectral efficiency gains now – with a target of more than 100% improvement by 2028.
Figure 1: AI in the telecoms market
These introductions show how telcos are poised to reap the rewards of applying AI to the network to enable automation and ultimately zero-touch operations. In addition, positive figures such as those in the fourth annual NVIDIA State of AI in Telecommunications report should aid momentum. The report uncovered that 90% of survey respondents said that AI is helping to increase annual revenue and reduce annual costs. In addition, 77% of respondents said they expect to see deployments of AI-native wireless networks before the launch of 6G in around 2030.
Agent running in the field
This year’s report found that network automation has overtaken customer experiences as the leading use case for investment, deployment and ROI impact as top-ranked use cases. This signals a move towards autonomous networks, where AI not only enhances productivity but drives the future of network operations. Of respondents who have deployed agentic AI, 41% said they did so for network automation, with 52% believing network automation AI agents will deliver significant ROI within the next two years.
Figure 2: Top AI use cases
Source:NVIDIA
Live AI deployments already are proving their value in telecoms operations. GSMA Intelligence reports that energy accounts for 15–20% of a typical telco’s operational spend. AI-driven optimisations in RAN sleep states and cooling can deliver significant gains as well, the firm says. A 20% reduction in energy use can provide up to 4% opex savings it points out, stating that these savings show how quickly AI can create value beyond customerfacing use cases. In fact, GSMA’s Use Case Library lists a series of examples including a GenAI drive test proof of concept at Azercell, a cross-domain anomaly detection and business impact analysis for network KPIs at Turkcell and O2 Telefónica’s traffic forecasting engine and demand predictions system.
The O2 Telefónica network planning team uses AI to predict mobile data usage for its 45 million customers, analysing current and historical traffic statistics at 28,000 sites. The system, based on Google Cloud’s Vertex AI platform, processes 250 gigabytes of data in minutes and delivers a detailed forecast of expected data traffic for each site for the next few years. This is achieved in a few hours following a half day of training the model.
Deployments such as these demonstrate that value AI systems can deliver to network planning. However, they also illustrate that telcos are following a learning curve and will continue to pick away at full automation and integration of AI into network planning on a step-by-step basis.
A Samsung spokesperson told Agile Telco: “We believe the integration of AI into network planning is poised to be transformative due to its ability to extract comprehensive, actionable insights from vast datasets and enable intent-driven automation across the core, RAN, and transport elements. While we expect that operator adoption will follow a phased approach, we believe the ultimate impact will be significant.”
“Initially, AI capabilities will aid planners in assessing scenarios and recommending actions,” they added. “Over time, agentic AI will evolve to interpret an operator’s business intent and coordinate different planning and optimisation applications. By combining telco knowledge with decision intelligence, these specialised AI agents will be able to transcend predefined rules, enabling collaboration across intelligent applications and telco-trained models.”
A question of comfort
As confidence builds in the results that AI-aided network planning use cases are generating, telcos will become more comfortable with relying on AI for mission critical network design. Samsung expects that telco confidence will grow as real-world use cases demonstrate tangible value and as AI’s potential continues to translate into measurable results. The company said that digital twins and sandboxes can test changes under conditions that closely mirror live networks, helping operators to prepare adequately before deployment. Many of Samsung’s collaborations with global operators including TELUS, KDDI, Verizon, SK Telecom and Vodafone use its AI-powered automation solution called CognitiV Network Operations Suite (NOS), which supports network planning functions – and extends into deployment, operations, troubleshooting, optimisation and customer experience management – with agents, telco-trained AI models and intelligent applications. These real-world efforts are helping to expand the evidence and operational experience needed to validate AI’s potential, the company added.
Samsung added: “Integration of AI into network planning allows operators to align their network investment more closely with what customers actually need. It can synthesise data from the RAN, core and transport networks to identify shifting demand, variations in performance and the most impactful interventions. This dynamic perspective offers planners deeper insights so they can make better-informed decisions as network conditions and service requirements evolve. Proactive capacity management can preempt congestion, while roll-out plans can be adjusted based on real-time demands. Proposed configurations can be tested digitally before resources are committed. AI-enhanced planning can also improve energy usage, network resilience, and end customer experience.”
At the heart of efficient network planning is a flexible, software-centric infrastructure that provides telcos with the capabilities they need when their business model dictates. For many, the benefits of AI-supported network planning are compelling already but others will exercise caution and seek to mitigate perceived risks of over-trusting in automated planning. Key enablers will be digital twins and sandboxes that enable telcos to test and test again until they gain full confidence and transfer their virtualised planning rehearsals into real world actions. All indicators point to this happening in the near term, enabling telcos to roll-out vast additional capacity quickly and cost-effectively, neatly meeting the needs of AI-generated traffic.
By the numbers
0GW
Telcos are set to deliver more than 3.5GW of AI factory and edge capacity by 2030 (STL Partners)
US$0 billion
AI in the telecoms market’s size is expected to increase from US$4.18 billion in 2025 to US$5.78 billion in 2026 and reach US$29.17 billion by 2031 (Mordor Intelligence)
0%
90% of survey respondents said that AI is helping to increase annual revenue and reduce annual costs (NVIDIA)
0 million
The O2 Telefónica network planning team uses AI to predict mobile data usage for its 45 million customers, analysing traffic at 28,000 sites processing 250 gigabytes of data in minutes
EVENT PREVIEW MWC26 DOHA
Advanced networks in the GCC prepare for The IQ Era
The fifth-generation network that supports the evolution of AI, IoT and edge computing is entering a new phase. Across the Gulf, the question is no longer about whether nationwide 5G markets are achievable, but what benefits will be derived from advanced connectivity in a world approaching 6G.
MWC Doha has rapidly established itself as MENA’s premier platform for urgent conversations on next-gen technology. And the event theme for 2026, ‘The IQ Era’, acknowledges the age of intelligence through smarter connection: human ideas leading tech innovation, commercial impact and societal progress.
From 5G leadership to intelligent economies: why the Gulf is ready for what comes next
Ericsson plays a central role in the MENA region’s mobile economy – and we asked its views on where the conversations might flow at MWC26. The company shared expert insights and conclusions from its latest Mobility Report, exploring why Gulf nations are first movers in turning 5G adoption into tangible value.
Over the past decade, the GCC has built some of the world’s most advanced digital infrastructure. Driven by forward-thinking national strategies and a strong appetite for innovation, the region pioneered 5G adoption on the global stage. Today, Gulf states are fully engaged in the AI race to secure their digital sovereignty and technological independence. This transition from ‘crude to compute’ reflects a broader ambition to position the Gulf as a globally significant hub for AI, advanced computing and the digital economy.
The end game, however, is not the early adoption of technology and the build of frontier digital infrastructure. These are vehicles to create economic and societal value.
MWC26 Doha comes at a pivotal moment – a moment when the imperative for new value creation is gaining full momentum among the region’s operators, governments, enterprises, technology companies and innovators. And at a time when advanced sovereign digital infrastructure and AI breakthroughs can be fully exploited.
From 5G adoption to value creation
According to the Ericsson Mobility Report, 5G subscription penetration across the GCC reached 53% at the end of 2025 and is expected to reach 88% of mobile subscriptions by 2031, or around 84 million subscriptions.
Today’s networks have become the nervous system of the AI economy. Their role in interconnecting large language models (LLMs), data, devices and people with the performance and speed needed to enable accurate and timely inference is important to transitioning AI from the experimentation phase to real-world applications. This importance has amplified the need to reclassify telecommunication networks as critical national infrastructure underpinning economies, essential services, businesses and, increasingly, defence and security.
A port deploying autonomous systems, a factory running AI-enabled machinery, an energy company managing critical assets or a defence operation requiring secure and resilient communications have very different connectivity requirements. For these mission-critical environments, guaranteeing reliability, resilience, security and predictable performance is non-negotiable. Mission critical networks need to be reliable, autonomous, programmable and capable of service differentiation to offer that guarantee.
This is where the Gulf has a significant opportunity. After building advanced and capable networks, the next step is about converting that technology leadership into business leadership, industrial resilience and economic value.
AI and connectivity: a powerful reinforcing loop
AI is constantly creating new demands on networks while simultaneously driving their evolution.
Network performance will increasingly dictate the quality of the AI experience as it shifts closer to devices, enterprises and the edge. Concurrently, AI is transforming networks into intent-driven, inference-capable, highly secure and autonomous systems, fundamentally optimising operational and energy efficiency. This symbiotic relationship forms the bedrock for next-generation technological breakthroughs.
Across the GCC, at-scale investments in AI, cloud infrastructure and next-gen connectivity are already well underway. The immediate opportunity lies in capitalising on this symbiosis to architect intelligent industries, hyper-efficient enterprises and entirely new business models.
The region’s ultimate ambition must transcend mere technology adoption. It must, instead, be to pioneer and shape what this technology makes possible.
Building the bridge to 6G
The Ericsson Mobility Report identifies the GCC as a strong 5G market and an expected early adopter of 6G.
6G will be an evolution, not a reset. Rather than a technological overhaul, it will build directly upon the robust foundations of 5G Advanced. As current 5G networks become increasingly programmable, intelligent and adaptive, the long-term transition to 6G will unlock unprecedented capabilities; these include AI-native networks, integrated sensing and communications, and deeply immersive digital experiences.
Ultimately, technological leadership is not defined by who deploys the next generation first, but by who turns it into value first.
Few regions are as uniquely positioned as the GCC to capitalise on this. By combining advanced infrastructure and ambitious national visions with significant investment and a rapidly accelerating AI ecosystem, the region is primed to be among the leaders of the global 6G value economy.
From connected infrastructure to intelligent economies
Ericsson is celebrating 150 years in 2026. Looking back across those 150 years, one lesson stands out: the value of connectivity has never been the network itself, but what people and societies build with it. This is more relevant than ever today.
The next phase of the Gulf’s digital transformation demands collaboration that transcends traditional industry boundaries. Turning cutting-edge infrastructure into tangible economic opportunities requires collaboration between governments, operators, enterprises, tech pioneers, developers, academia and startups.
This imperative underscores the vital role of premium ecosystems like MWC26 Doha. Serving as a powerful platform for action, MWC26 Doha unites global technology and policy leaders. And this year, it’s happening precisely as the core conversation shifts from “How connected are we?” to “What can we achieve because we are connected?” This is the defining question for the Gulf’s next digital chapter.
Having firmly established its global leadership in connectivity, the region must now utilise this infrastructure to build the foundation for AI-powered industries, mission-critical operations, national resilience and entirely new business models.
Networks have outgrown their role as the mere background infrastructure of the digital economy. They are becoming the nervous system of the intelligent economy. True leadership will no longer be measured by the sophistication of a region’s networks, but by what those networks actively enable for its people, industries and national fortunes.
Be part of the conversation at MWC26 Doha
However advanced the infrastructure is, it is always moving between states of readiness to enable the next stage of technological development. And MWC26 Doha is where that enablement takes place – among global vendors and innovators turning 5G leadership into the intelligent economies of The IQ Era.
This year, there is a rare alignment that makes these industry conversations more important than ever. MWC26 Doha is purposefully co-located with the ITU’s Plenipotentiary Conference (PP-26), which sets global digital policy across 194 Member States until 2030. For three days, telecoms leaders, technology companies and the government teams advancing national digital visions around the world will all be in Doha together.
Buy your pass or enquire about exhibiting today and secure your place at one of the region’s most influential gatherings for 5G connectivity and 6G futures.
Find Ericsson on the MWC26 Doha showfloor, stand number F30.
EVENT PREVIEW MWC26 DOHA
Head to Doha to learn why the next frontier for agentic AI is agentic society
Artificial intelligence is entering a new phase. Across the Gulf and beyond, the conversation is no longer about what AI might achieve, but how organisations, governments and societies can put intelligence into action responsibly and at scale.
That shift makes the discussions taking place at MWC26 Doha on 8-10 November more important than ever. Bringing together leaders from technology, telecommunications, government and investment, MWC Doha has rapidly established itself as MENA's premier platform for the conversations that shape digital transformation. At a time when national visions are accelerating and AI is becoming central to economic diversification strategies, the region's most influential decision makers will gather to explore where we take things from here.
One of the voices contributing to the debate is Jad Haddad, the Global Head of Quotient – AI at Oliver Wyman, who examines why the next frontier of agentic AI may not be agentic government, but agentic society.
Jad Haddad, Oliver Wyman
With agentic AI firmly in the global spotlight, the Gulf region offers a particularly pertinent glimpse into how the technology’s next phase could look. As governments across the region move rapidly from ambition to implementation, businesses are exploring how agents can reshape operations and customer interactions, and consumers are becoming increasingly comfortable allowing AI to influence, and even make, decisions on their behalf. That convergence is key to Oliver Wyman’s decision to sponsor the Agentic AI Summit at MWC26 Doha this year.
At last year’s edition, I wrote about agentic AI moving from science fiction to strategy. The question then was whether governments could move beyond AI that assists people toward systems capable of planning, reasoning and executing multi-step work. The answer rested on something more important than the technology itself: whether autonomy could be introduced in ways that remained measurable, auditable and accountable.
One year on, as the debate has moved decisively from principle to implementation, the question has moved too: What happens when governments and businesses are not the only ones with agents?
From ambition to implementation
Across the Gulf, governments are beginning to put agentic AI into practice. In February, Qatar’s Ministry of Commerce and Industry launched an ‘AI Agent Factory’ with Microsoft to build, manage and deploy agents across its services. In April, the UAE announced a framework to transition 50% of government sectors, services and operations to agentic AI within two years. And in Saudi Arabia, where 2026 has been designated the Year of AI, the Digital Government Authority reports that adoption of emerging technologies across government entities has exceeded 76%, with agentic AI beginning to feature among those use cases.
For the public sector, that shifts the debate considerably, from whether organisations should adopt agentic AI to where it can create value, how much autonomy agents should have and how they can be embedded responsibly into operating models.
Meanwhile, an equally consequential change is taking place on the other side of the interaction.
What happens when citizens have agents too?
Research from the Oliver Wyman Forum reveals a significant shift in how and where Gulf consumers spend their money. In April 2026, 57% said they turned first to AI when looking for a product or service without a particular brand in mind, compared with 21% who favoured search engines. By August, a separate survey found 84% had now interacted with AI to find or purchase a product, with 90% of those eventually choosing an option recommended by AI.
A further shift from recommendation to delegation has also begun: 25% of people now say they would allow an AI agent to choose a mobile phone plan or broadband provider on their behalf.
This raises an important question: What happens when companies need to shift their engagement efforts from customers to the AI agents acting on their behalf?
An agent could compare hundreds of plans, prices and contract terms almost instantaneously against an individual’s needs, upending traditional assumptions around customer acquisition and loyalty, and forcing providers to consider not only how their services appeal to people, but how easily they can be discovered, understood and transacted with by authorised agents.
From access to action
While this shift does not prove citizens are ready to delegate consequential public decisions in the same way, it signals a growing expectation that digital systems should do more than simply direct them to the right information.
In public services, an authorised personal agent could, for example, identify an expiring permit, gather the necessary information, seek its owner’s approval, then interact with government systems to complete the renewal. A government agent on the other side might validate eligibility, identify exceptions and escalate cases requiring human judgment.
The important change is not simply that both sides are using AI, but that those agents may increasingly interact directly across institutional boundaries on behalf of the people and organisations they represent.
That requires services and processes designed specifically for authorised machine interaction: interoperable, machine-readable, and able to establish who an agent represents and what it is permitted to do.
From human oversight to accountable delegation
As agents gain greater ability to act, governance must evolve with them.
How does a government or business know that an agent genuinely represents an individual? What has that individual authorised it to do? For how long? With which data? And where does accountability sit when an action has material consequences?
This is the basis for accountable delegation – and for many of the discussions ahead of us at this year’s summit.
Last year, the challenge was moving from AI pilots to agentic platforms. The next frontier is larger: preparing for a world in which governments and businesses deploy agents while citizens and consumers increasingly have agents acting on their behalf too.
The GCC has an opportunity to help establish the rules of trust, authority and accountability that allow this future to work. At MWC26 Doha, that is one of the conversations worth advancing: not simply what AI agents can do, but what happens when they increasingly begin to act, and interact, on our behalf.
Continue the conversation at MWC26 Doha
The questions surrounding agentic AI cannot be answered by technology alone. Progress depends on collaboration between the people building, regulating and investing.
MWC26 Doha brings those communities together. With end-to-end access to policy makers, technology leaders and decision makers from across MENA and beyond, this is where we can understand future scenarios and direct the new dynamics of agentic societies.
Buy your pass or enquire about exhibiting today and secure your place at one of the region’s most influential gatherings for AI, 5G connectivity and 6G ambition. www.mwcdoha.com
LEAVE YOUR VOICE NOTE
Will telcos allow network planning to automate to innovate?
Leave your voice note
Give us your opinion on the topic. Ring in and leave a message or use email to comment.
