Contributing Analysts: Jake Saunders, VP, Asia-Pacific & Advisory Services; Dimitris Mavrakis, Senior Research Director; Larbi Belkhit, Senior Analyst; Sam Bowling, Industry Analyst
Explore what ABI Research analysts have to say about how AI is transforming telecommunications in 2026. Through this Q&A, the industry gains much-needed clarity into the commercial opportunities and technical challenges in integrating AI with cellular networks.
Artificial Intelligence (AI) in telecommunications is no longer theoretical, but a strategic imperative. AI facilitates network automation, improves customer support, and expedites service delivery. In this Q&A, ABI Research analysts offer a clear view of how Communications Service Providers (CSPs), Mobile Network Operators (MNOs), and vendors are applying AI across cellular networks in 2026. From AI-ready infrastructure and monetization challenges to emerging in-building wireless use cases, this article explains what’s real, what’s scalable, and what’s next in telecom AI.
Table of Contents
1. What are the benefits of AI in telecommunication networks today?
2. How are telcos actually using AI in 2026?
3. Is AI monetization currently viable in telecom?
4. Why is AI-RAN still stuck in trials despite heavy vendor investment?
5. What role will AI play in the evolution of the 6G core network?
6. How are telcos approaching AI compute and GPUaaS?
7. Why are AI agents gaining traction in in-building wireless?
1. What are the benefits of AI in telecommunication networks today?
Samuel Bowling, Industry Analyst: “The significance of AI-ready networks lies in their role of connectivity. Instead of being a neutral transport layer, networks become active enablers of automation, intelligence, and real-time decision-making.”
The benefits of AI for telecom operators come from intelligence being embedded into operations rather than added as an overlay. Key advantages include:
- Predictive operations that move networks from reactive fault handling to proactive optimization
- Deterministic performance that enables AI workloads to run reliably at the edge
- Operational efficiency gains through reduced downtime, smaller headcount (e.g., customer service), faster issue resolution, and improved throughput
- Industry-specific enablement, including manufacturing automation, logistics routing, and healthcare diagnostics.
Bowling adds that “Any business case for AI-ready networks cannot be based on bandwidth and latency. Adoption will necessitate measurable protection from errors, downtime, throughput speed, or cost recovery.”
2. How are telcos actually using AI in 2026?
Larbi Belkhit, Senior Analyst: “Almost every telco AI infrastructure announcement is backed by confirmed demand from day one, reducing the speculative risk and ensuring immediate monetization.” He also emphasizes that this approach sharply contrasts with the speculative strategies of neoclouds.
Telcos are not trying to outspend hyperscalers. They are taking a realistic, customer-first approach that prioritizes confirmed demand over speculative build-outs.
Most telcos are using AI in targeted, commercially grounded ways:
- Deploying sovereign AI factories for regulated and national workloads
- Supporting AI inference at the edge, rather than large-scale AI training
- Applying AI in mobile networks, energy optimization, and service assurance
- Enabling enterprise AI use cases with strict data residency requirements
3. Is AI monetization currently viable in telecom?
Dimitris Mavrakis, Senior Research Director: “There appears to be a gap between the potential opportunity, ambition levels, and current capabilities when it comes to AI implementation and monetization.”
ABI Research and Amdocs surveyed 300 telco leaders worldwide to gauge their priorities for modernizing charging systems. AI integration was a recurring theme among respondents, and the survey results highlighted AI’s growing importance in delivering telecom services.
Survey findings highlight that:
- 65% of CSPs are deploying or trialing AI factories for 5G monetization
- 57% of CSPs are deploying or trialing sovereign AI cloud
- 47.5% of CSPs are deploying or trialing Graphics Processing Unit-as-a-Service (GPUaaS)
- Over 40% say AI integration remains a major monetization challenge
Mavrakis warns that “Significant gaps between intent and actual capability in relation to B2B should not be taken for granted and should be addressed immediately. There is a risk of these gaps becoming more prevalent in relation to AI as opportunities in that area rapidly emerge, especially with AI factories, GPUaaS, and edge AI deployment models. “
4. Why is AI-RAN still stuck in trials despite heavy vendor investment?
Belkhit: “The disparity between the perceived advantages associated with AI-Radio Access Network (RAN) and what has actually been documented through practical use cases is simply too great for service providers to justify committing their capital toward AI-RAN deployments at this time.”
Despite strong vendor momentum, such as the US$100 billion NVIDIA-Nokia deal, AI‑RAN adoption remains limited because:
- There is no independent, validated benchmarking process
- Return on Investment (ROI) remains unclear amid ongoing 5G monetization challenges
- Architectural uncertainty persists between Graphics Processing Unit (GPU), Central Processing Unit (CPU), and custom silicon approaches
- Power consumption and vendor lock-in remain unresolved concerns
Belkhit continues, “At best, 2026 will be a year when AI-RAN is likely to experience a continuation of conservative testing, with only modest deployments occurring.” These deployments are expected to be primarily in non-time-sensitive use cases like traffic prediction and anomaly detection.
He concludes, “In order for AI-RAN to become more than just a buzzword, the telecom industry will need to standardize metrics, validate economic models, and demonstrate interoperability between different architectures.”
5. What role will AI play in the evolution of the 6G core network?
Jake Saunders, VP, Asia-Pacific & Advisory Services: “The telco community wants to ensure that 6G deploys a native AI architecture, rather than AI as an external "plug-in” for today’s 5G network. A 6G network could then use AI as an integrated “control loop” for managing traffic, etc.” He also tells us that “From discussions that ABI Research has had with Communications Service Providers (CSPs) headquartered in Asia and Europe, they believe an AI-core network should provide trusted access, interconnection, and task cooperation for multiple AI agents.”
Key AI-driven objectives for the 6G core include:
- Simplifying architecture to reduce operational costs
- Embedding AI as a native control loop for traffic and service management
- Supporting collaboration among multiple AI agents
- Integrating Sensing and Communication (ISAC) and compute within a single core
While telco enthusiasm for native AI functionality in 6G networks is high, challenges loom. Saunders stresses that the telecom industry must reconcile the rapid pace of AI innovation with slower standards-based cycles, manage and govern massive numbers (potentially millions or even billions) of AI agents, ensure interoperability across a fragmented global ecosystem, and introduce advanced AI capabilities without increasing network complexity or costs.
6. How are telcos approaching AI compute and GPUaaS?
Belkhit: “Long term, ABI Research expects that telcos will continue to focus on AI inference as the main opportunity in the sovereign AI landscape, with perhaps AI training and fine-tuning being a secondary opportunity for customers with high sovereignty requirements (i.e., beyond data residency).”
He also mentions that “Most telcos are either facilitators or aggregators, largely depending on current data center portfolio and enterprise AI demand. Facilitators typically do not build AI factories, but leverage their existing portfolio to have an Infrastructure-as-a-Service (IaaS) model when leasing capacity for neoclouds to run their own Graphics Processing Unit (GPU) clusters.”
Telcos also frequently collaborate with other cloud service providers due to relative inexperience in offering distributed compute to third parties.
Rather than racing hyperscalers, telcos are:
- Building GPUaaS offerings tied to sovereign and regulated demand
- Partnering with hyperscalers and neoclouds instead of developing full proprietary stacks
- Leveraging existing edge facilities to support distributed inference
- Avoiding speculative capacity that lacks immediate customers
ABI Research forecasts that telco GPUaaS revenue will reach US$21.1 billion by 2030, increasing from about US$0.02 billion in 2024. The year 2027 will be a key inflection point, as most telco AI infrastructure is anticipated to be built out by then.
7. Why are AI agents gaining traction in in‑building wireless?
Belkhit: “2026 will mark the breakthrough moment when connectivity providers will lean into the capabilities of AI to simplify technology complexities and combine different connectivity technologies to provide a strong in-building wireless connectivity portfolio.”
AI agents are gaining traction in telecom network operations because they:
- Unify Distributed Antenna Systems (DAS), Distributed Radio Systems (DRS), and private cellular through a single orchestration layer
- Use the Model Context Protocol (MCP) to access tools, Application Programming Interfaces (APIs), and telemetry across systems
- Enable real-time optimization without manual intervention
- Deliver clear Operational Expenditure (OPEX) reductions and operational simplicity
The use of the MCP is especially important for AI agents in telecom, breaking down system silos. As a result, Belkhit says that AI agents can “compare Radio Frequency (RF) conditions across systems, choose the right technology for a given space, reconfigure capacity in real time, or trigger workflows (e.g., retuning DAS, optimizing small cells, reallocating private network bandwidth) through a single orchestration layer.”
Final Takeaways: Where AI in Telecom Is Headed
ABI Research sees telecom AI entering a more grounded and commercially focused phase. 5G monetization challenges are a fresh wound, so telcos are conservative about investing in AI. Our Analyst Q&A illustrates that telcos are choosing pragmatism over speculation.
Here are four strategic conclusions telecom stakeholders should keep in mind:
- AI in telecom is real, but use case–driven. Operators are using AI to optimize network operations, predict faults, and enhance energy efficiency. However, most AI deployments in telecom are within environments with deterministic connectivity and sovereign requirements.
- AI-RAN still needs proof. Despite headline partnerships like NVIDIA and Nokia, there’s no validated benchmark, ROI model, or architectural consensus to unlock scale. For these reasons, most AI-RAN projects in 2026 remain in the trial phase.
- Edge AI, GPUaaS, and inference are rising priorities. Telcos are not as focused on AI training workloads as hyperscalers. Instead, they’re preparing their AI infrastructure to serve edge inference, often in partnership with cloud and neocloud players.
- AI agents are where unification happens. While macro networks advance cautiously, AI agents are making significant progress in in-building wireless. Chinese vendors are converging DAS, DRS, and private cellular through orchestration layers and open APIs. Meanwhile, Western telcos are using the MCP to ensure AI agents work in unison.
Across all domains, ABI Research emphasizes that success with AI in telecom hinges on standardization, ecosystem openness, and measurable performance gains. The shift from AI pilots to real-world deployment is underway, but it’s gradual. The winners will be those who balance technical readiness with commercial realism.
As AI continues to transform the telecom industry, learn how ABI Research supports organizations across the telco & communications space in integrating cutting-edge AI technologies. Learn more
Related Research
Agentic AI Becomes an Essential Innovation for AI Core Architecture
How Generative AI Will Reshape Telco Business Strategies
Seven Telecom Trends to Watch in 2026 and Beyond