Get a jump start on this hot topic at MWC2026 with our new report which includes case studies from Airtel, Orange, Singtel, SK Telecom, Telefónica, Verizon, Virgin Media O2
Here’s an excerpt to get you started: Generating new revenues has aroused the most interest regarding GenAI and agentic AI in telecoms. Marina Koytcheva, Research Director at STL Partners, notes that investors are less lenient with telcos financing big ambitious changes, compared to Big Tech giants or hyperscalers. She fears that if telcos are subjected to the same time pressures regarding return on investment as in the previous 15 years or so, opportunities could slip away.
Telcos desperately need greater operational and business agility to generate new revenues and AI-powered autonomous networks will be a major contributor. One of the reasons the Big Tech companies are viewed more favourably by investors is because so far, they have been able to adapt and seize new opportunities more swiftly, and counter emerging threats more effectively.
Tectonic shifts in infrastructure
Interestingly, some commentators, like John Mihaljevic of MOI GLOBAL, an investor community, argue that “by ramping up AI-related capital spending to unprecedented levels, [hyperscalers] have set themselves on a perilous path, away from high-margin, capital-light models toward a capital-intensive future in which their return-on-capital and margin profiles are highly uncertain.”
Sebastian Barros notes on a LinkedIn blog that, “Hyperscaler CapEx has skyrocketed from $24B in 2015 to a projected $325B in 2025—a 13X increase in just a decade. Meanwhile, global telco CapEx is estimated at $297B in 2024, marking a 5% decline from the previous year.
“For the first time, the four cloud giants (Amazon, Google, Microsoft, Meta) are outspending the entire +1.150 telecom players.”
Mihaljevic adds, “This AI-driven capex frenzy is eerily similar to the telecom bubble of the late 1990s and early 2000s. Extravagant spending on fiber optics and network infrastructure promised growth but delivered catastrophic oversupply and collapsing prices. Today’s hyperscalers may be repeating history’s costly mistakes.”
Telcos are better placed
Telcos, often held back by legacy infrastructure and operations, have traditionally been much slower to shift direction. Now greater virtualisation of the network, via software-defined capabilities and cloud deployment, have helped to address this – as long as telcos understand where else this could take them, strategically and commercially. Add AI to the scenario, and real-time adaptation starts to become achievable.
Juniper Research expects operators’ AI investment to exceed $86 billion by 2029, driven by attempts to achieve zero-touch operations within mobile networks, according to findings published in April. Zero-touch operations significantly minimise or eliminate human intervention in network operations, but that forecast relies heavily on operators exploiting agentic AI, which is still an emerging technology but very much in telcos’ sights (see Verizon case study on next page). Juniper claims the biggest ROI will come from deploying agentic AI in the RAN and work is underway.
Autonomous RAN opportunities
In February 2025, Deutsche Telekom and Google Cloud announced a new partnership to improve RAN operations through the development of a network AI capability – the RAN Guardian agent, built using Gemini 2.0 in Vertex AI from Google Cloud. This AI-powered assistant can analyse networks’ behaviour, detect performance issues and implement corrective actions to improve network reliability, reduce operational costs and enhance customer experiences. For customers, it should mean fewer disruptions, more optimal speeds and an enhanced mobile experience.
Speaking at the time of the announcement, Abdu Mudesir, Group CTO at Deutsche Telekom, said, “Traditional network management approaches are no longer sufficient to meet the demands of 5G and beyond. We are pioneering AI agents for networks, working with key partners like Google Cloud to unlock a new level of intelligence and automation in RAN operations as a step towards autonomous, self-healing networks.”
AI in the RAN
Telecoms analyst Larbi Belkhit is part of ABI Research’s Strategic Technologies research group focused on 5G, 6G, and Open RAN research. He has highlighted the rise of ‘AI RAN’ as key to RAN automation for 5G as it will determine how mobile operators manage, deploy and monetise their mobile infrastructure in future. RAN automation is required for orchestrating slices end-to-end and enforcing service-level agreements (SLAs).
The AI RAN offers monetisation opportunities such as: automated SLA management and the dynamic adjustment of SLA metrics; quality of experience (QoE) monitoring and optimisation, using predictive models to refine service delivery; assurance for RAN slices and SLA assurance; fast-loop optimisation to meet SLAs; and slice-aware admission control which prioritises access for different user types.
It is still very early days. ABI Research doesn’t anticipate a ramp up in deployments much before 2029. Meanwhile, to support broader adoption of AI RAN for automation, he says, the telecommunications industry must:
• Standardise interfaces for multi-agent orchestration;
• Partner hyperscalers such as Amazon Web Services (AWS) and Google Cloud to accelerate AI integration;
• Mature data strategies to enable higher granularity and context awareness; and
• Develop more transparent and explainable LLMs – investigated on page 6 of the report.
Zero-touch can be too much
Inside and beyond the RAN, a comprehensive approach to zero-touch networks would appear to be the obvious way to leverage AI, but it may be unrealistic and hard to justify financially across the board. This is reflected in the most recent TM Forum regional benchmark report, published in summer 2025, which finds that only 4% of communication service providers (CSPs) have achieved TM Forum’s definition of Level 4 autonomy to date, and just 23% expect to get there by 2026.
A more discerning and incremental approach is now seen as more pragmatic, targeting specific ‘high-value scenarios’ rather than sweeping, end-to-end automation. This is based on a greater understanding that different parts of the network warrant different levels of autonomy – and investment.
The Chapter includes these case studies…
The Verizon case study outlines how the operator sees agentic as a continuation of its “productivity and efficiency play”.
The Virgin Media O2 case study explains how the operator is planning to gain new value from doing what it does already, but very much better, as it readies its network platform to deliver customised network experiences and tailored services by leveraging network automation and AI.
The Orange case study shows how the operator group is already monetising its investment in AI operations. Then in the next chapter we look at leading edge and future services.
