An analysis of the World Economic Forum and Accenture white paper, The Strategic Role of Telecom Providers Across the AI Value Chain(March 2026).
The structural reset no telco can avoid
For three decades, communications service providers built the infrastructure on which the digital economy runs. They carried the traffic. They carried the trust. And they carried the regulatory weight. And then they watched the value accrue to someone else.
The numbers, drawn from the World Economic Forum and Accenture’s March 2026 white paper The Strategic Role of Telecom Providers Across the AI Value Chain, tell the story with uncomfortable clarity. Global communications revenue is forecast to grow just 2.3% through 2026.
Margins remain flat, with global EBITDA holding at approximately 33%. Post-tax return on invested capital sits at approximately 8%, far below digital platforms’ approximate 27% in 2025. Capital intensity continues to constrain investment capacity and innovation.
That is the inheritance. The question now is whether telcos accept it, or whether AI gives them the conditions to rewrite it.
The WEF and Accenture argue, persuasively, that AI is not just the next wave of technology adoption. It is a structural realignment of the digital economy. As AI scales, connectivity becomes the critical bridge between users and edge devices on one side, and centralised or distributed compute resources on the other. Large language models remain too heavy to run fully on end devices. Workloads distribute across devices, edge locations, and data centres. Networks move from interaction-based AI to continuous exchanges between autonomous agents, platforms, and enterprise systems.
That shift, combined with the geopolitical pull toward digital sovereignty, hands telcos a window they have not had in a generation.
Why is AI different from previous telco growth waves?
Every previous platform wave, from cloud to mobile to streaming, ended with telcos owning the pipe and platform players owning the value. AI looks different for three reasons.
First, performance now has to be guaranteed, not best-effort. Real-time inference, multimodal AI, agentic systems, and physical AI in robotics and autonomous platforms all require predictable, low-latency, secure connectivity. The TM Forum’s analysis cited in the WEF paper is unambiguous: operators will need a new networking core built on autonomous operations and experience-based charging to participate meaningfully in the AI value chain. Best-effort pricing is structurally incompatible with the workload.
Second, compute is moving back toward the network. IDC projects spending on sovereign cloud worldwide to grow 27% on average from 2022 to 2027, reaching $258.5 billion by the end of the forecast period. By then, localised data centres will contribute nearly a quarter of new computing capacity. The geographic centre of gravity in compute is shifting back toward where the telcos already are.
Third, sovereignty is now a buying criterion. 81% of enterprises are seeking telcos as partners for sovereign AI adoption, according to Accenture research cited in the report. Governments across the United Arab Emirates, Indonesia, India, Thailand, and now Europe and Canada are explicitly looking at national telco players to anchor sovereign AI agendas.
This is what makes the moment different. The structural advantages telcos have always had, namely licensed spectrum, fibre backbones, trusted national-scale operations, regulatory standing, and customer reach, are now the assets that matter most in the layer where AI value is being created.
What are the three strategic pathways for telcos in the AI era?
The WEF and Accenture frame the opportunity as three pathways, each of which a telco can pursue alone or in combination. Each pathway reflects a distinct vector of value creation, and each comes with its own enabling capabilities and execution risks.
Pathway 1: The modern telco that protects the core
The modern telco pathway is about defending and monetising the connectivity foundation itself. It modernises networks into cloud-native and AI-native architectures, uses AI to drive operational productivity, and delivers predictable, experience-led performance.
Four strategic plays sit underneath this pathway. AI-first connectivity turns the network from best-effort transport into experience-guaranteed, programmable connectivity, exposed through APIs and network slicing. Omdia, cited in the paper, forecasts that AI-enriched interactions will expand at approximately 120% compound annual growth rate through 2030. Data centre interconnect (DCI) monetises the long-haul optical transport linking hyperscale and edge compute, with IDC projecting DCI revenue to grow at 10.9% CAGR, reaching $5.5 billion by 2029. AI-optimised colocation retrofits existing telco real estate into power-dense, fibre-connected sites, sitting inside an IDC global colocation services market growing from $56.4 billion in 2024 to $81.2 billion by 2029. Dedicated networks provide secure, low-latency campus and private 5G connectivity to mission-critical sectors.
The proof is already on the ground. T-Mobile US launched Edge Control and T-Platform in October 2025, pitching mission-critical, low-latency connectivity for AI and edge applications across media, sports, entertainment, and the public sector. Verizon Business signed a Verizon AI Connect agreement with Amazon Web Services to build long-haul, high-capacity fibre pathways linking multiple AWS data centre locations. Cellnex is repurposing tower sites, fibre networks, and existing data centre facilities to deliver AI-optimised colocation at the edge, with partnerships such as the one with Vapor IO accelerating real-time AI deployment.
Pathway 2: The AI techco that grows through managed services
The AI techco pathway is where telcos stop being purely connectivity providers and become trusted orchestrators of AI-ready infrastructure and managed services.
Four plays sit here too. Network-as-a-service (NaaS) is the commercialised service layer that delivers flexible, consumption-based connectivity through cloud-native networking and intent-driven APIs. The demand signal is sharp: an Accenture 2025 survey shows 86% of enterprise leaders expect NaaS to be significant to their business within three years, and Analysys Mason projects global NaaS connectivity revenue to grow at a 42% CAGR from 2024 to 2029, reaching $14.7 billion across retail and wholesale models. GPU-as-a-service (GPUaaS) extends telco infrastructure into the compute layer, with Gartner projecting a global AI-optimised IaaS market of $108 billion by 2029. Horizontal and vertical AI solutions combine telco infrastructure with industry-specific AI agents and applications, with 94% of enterprises expecting AI productised solutions to play a significant role in their business over the next three years, per Accenture. The B2C AI aggregator play positions telcos as curators of consumer AI services, sitting inside an IDC prediction that consumers will spend $100 billion via AI agents by 2027.
Telefónica Global Solutions’ Dynamic Network is already in market as a NaaS offering with cloud-like API-based interfaces. Iliad, Indosat, Singtel, SK Telecom, Telenor, and Verizon have all launched GPUaaS platforms. Deutsche Telekom has gone furthest on the consumer side with its app-less AI phone concept, demonstrated at Mobile World Congress 2025 in partnership with Perplexity, Google Cloud, and Qualcomm.
Pathway 3: The national sovereign champion that activates resilience
The third pathway is the one that distinguishes 2026 from any previous strategic moment in telco history. Sovereignty has moved from a data residency conversation to a full-stack national strategic priority.
Global spending on sovereign AI infrastructure is projected to reach $1.5 trillion by 2028, with $145 billion in Europe alone, growing at nearly 30% annually. At least 18 telcos are already participating in state-funded AI infrastructure programmes. In Europe, 62% of organisations now seek sovereign solutions, particularly in banking, public services, and utilities. And critically, only 22% of organisations currently enforce sovereignty at the model layer, exposing a structural gap that telcos are uniquely positioned to fill.
The play stack is comprehensive. Sovereign AI infrastructure delivers AI-ready colocation and GPUaaS campuses within state jurisdictions. Sovereign AI platform and model services provide the software, orchestration, and governance layer above sovereign compute, including agent builders, LLM orchestrators, sovereign regulatory controls, and federated data layers. And sovereign AI use cases target defence, public safety, health, energy, and other mission-critical domains. AI security and safety turns telcos into the integration and assurance layer for AI deployments, with spending on AI-powered network and cybersecurity systems projected to double over the next five years.
T-Systems is partnering with Nvidia in a $1 billion initiative to renovate data centres and launch an industrial AI cloud in Germany, using up to 10,000 Nvidia Blackwell chips and projected to boost AI compute by 50%. Telenor’s AI Factory combines in-country GPU infrastructure with developer tooling, low-latency networking, and high-performance storage on NVIDIA architecture, anchored fully in Norwegian ownership and operations. Telia Cygate has launched a sovereign AI platform with Accenture in Sweden. e& has built an agentic AI platform for autonomous enterprise workflows spanning CRM, ERP, data systems, and communication channels.
Which pathway should a telco actually choose?
The honest answer, and the one the WEF and Accenture make explicitly, is that the pathways are not all equally achievable. They depend on internal AI maturity, external partnerships, regulatory environment, and overall market readiness. A telco in a market with a strong sovereign agenda has a different optimisation than one in a market with hyperscaler-dominant procurement patterns.
The WEF and Accenture’s value and feasibility matrix is instructive. The clear sweet spots (high value, high feasibility) are horizontal and vertical AI solutions, data centre interconnect, NaaS, sovereign AI use cases, AI-first connectivity, and AI security and safety. The frontier bets (high value, lower feasibility) are sovereign AI infrastructure and sovereign AI platform and model services, where the capital burden and regulatory complexity are highest but the long-term revenue pools are deepest where national investment exists. GPUaaS sits in a more cautious quadrant, with strong demand but a competitive position the report frankly describes as weak relative to hyperscalers in non-sovereign markets.
This is where the conversation has to get specific. There is no universal telco playbook for AI. There are only context-specific portfolios, prioritised against the value and feasibility realities of each market.
What does it actually take to execute on these pathways?
The WEF and Accenture group the execution capabilities into two layers, and Teraflow’s experience with enterprise AI delivery aligns closely with this framing.
Foundational enablers are the conditions every telco needs regardless of pathway. They include a disciplined AI portfolio governance approach with prioritisation and funding based on ROI and strategic fit, a shared data and AI layer with governed data products and standardised tooling, and a responsible AI, security, identity, and compliance posture embedded by design rather than bolted on later. These are the layers that decide whether AI delivers measurable outcomes or stays trapped in pilot purgatory.
Pathway-specific enablers are cumulative. The modern telco requires unified AI-coded IT architecture and interoperable, API-driven RAN, core, and transport fabric. The AI techco adds product factory capability, AI-powered managed services delivered end-to-end with SLAs, and ecosystem co-innovation with hyperscalers and AI model providers. The national sovereign champion layers on sovereign automated AI infrastructure, sector-specific LLM and SLM services, a national-grade operating model with transparent controls and audit mechanisms, and participation in sovereign AI programmes including AI gigafactories and national centres of excellence.
This is the part most strategy decks gloss over. The pathways are not just commercial choices. They are operating-model choices, talent choices, and architectural choices. The telcos that execute will be the ones that stop treating AI as a function and start treating it as the new operating logic of the business.
What does this mean for telco leaders right now?
Three things, in order.
First, pick the portfolio, not the pathway. No telco will execute one pathway purely. The modern telco capabilities are non-negotiable foundations for everything else. The AI techco services should be sequenced based on where the operator has actual differentiation, which is rarely raw compute scale and more often local presence, compliance positioning, and integration expertise. The sovereign champion role is available only in markets where the national agenda makes it available, and only to operators with the capital and capability to absorb the build.
Second, modernise the core before chasing the adjacency. AI-first connectivity, autonomous network operations, and a programmable, API-exposed network are prerequisites for nearly every play in the AI techco and sovereign pathways. Telcos that try to launch GPUaaS, NaaS, or sovereign AI platforms on top of legacy OSS and BSS will find the economics do not work. The Accenture survey embedded in the WEF paper is clear: 86% of enterprise leaders expect NaaS to matter within three years, and they are buying assured performance, not undifferentiated bandwidth.
Third, choose ecosystem positioning deliberately. Hyperscaler dominance at the edge, fragmented regulatory environments, and the capital intensity of sovereign infrastructure are real constraints. The most credible telco AI strategies in the WEF case studies are partnership strategies, not solo strategies. T-Systems with Nvidia and SAP. Telia Cygate with Accenture. Deutsche Telekom with Microsoft, Perplexity, Google Cloud, and Qualcomm. Verizon with AWS. The question is not whether to partner. It is which partners, on what terms, with what data and IP boundaries.
The bottom line
The AI era fundamentally redefines the role of telcos. Beyond delivering advanced connectivity, they are increasingly positioned to power intelligent and sovereign systems, and in some contexts to assume a central role in orchestrating the national AI value chain.
The central question is no longer how AI can make telcos more efficient. It is which roles telcos choose to play in the AI value chain, and how quickly they can execute those choices.
Over the next decade, the operators that deliberately align their core operations, growth initiatives, and sovereignty ambitions within their specific national and market contexts will see the largest increases in value creation. The ones that do not will keep operating an industry forecast to grow 2.3% on flat margins, watching the AI economy build value on top of their networks for the second time in twenty years.
That is the choice. And unlike the last one, this one is being made with the data already on the table.
How Teraflow thinks about this
Teraflow’s work with enterprise AI clients, including Vodafone Portugal where the team built 2,900+ data pipelines across 3.6 petabytes and 150+ ML models serving 37 million digital users, Cell C where digital revenue grew 10x in six months off a re-platformed digital subscriber stack, and Comair where AI-driven yield management delivered 98% aircraft fill rates across 4 million passengers, points to the same operational truth the WEF and Accenture white paper arrives at by a different route.
The strategy decisions matter. The architecture decisions matter more. AI value is captured by organisations that have the data, infrastructure, and delivery capability to put models into production at scale. Telcos have a structural advantage in that fight. Whether they convert it is now a question of execution.
The Teraflow DAPA (Digital AI Platform Accelerator) framework, Phase Zero scoping methodology, and FloJo delivery operating model are built precisely for this kind of inflection. They exist to take strategic clarity, of the kind the WEF and Accenture have laid out here, and turn it into running production systems on a timeline the business can defend.
Frequently asked questions
What is the AI value chain in telecoms?
The AI value chain in telecoms covers the layered set of capabilities required to deliver AI services at scale, from connectivity and edge compute through to sovereign infrastructure, model platforms, AI applications, and governance. The WEF and Accenture frame it across three layers: AI-first networks, AI-ready infrastructure and managed services, and sovereign AI ecosystems.
What is sovereign AI and why does it matter for telcos?
Sovereign AI refers to AI infrastructure, platforms, models, and use cases that operate within national jurisdictions under national policy controls. It matters for telcos because 81% of enterprises are seeking telcos as partners for sovereign AI adoption, and global sovereign AI infrastructure spending is projected to reach $1.5 trillion by 2028.
What is the difference between a modern telco and an AI techco?
A modern telco is focused on protecting and monetising its core connectivity business using AI to drive performance, automation, and assured services. An AI techco adds AI-powered managed services on top of connectivity, including network-as-a-service, GPU-as-a-service, and horizontal and vertical AI solutions, with consumption-based pricing and SLA-grade delivery.
Why is post-tax ROIC for telcos so much lower than for digital platforms?
Per the WEF and Accenture white paper, telco post-tax ROIC sits at approximately 8% while digital platforms’ approximate 27% in 2025. The gap reflects consumption-based pricing structures and stronger value capture by platform players, alongside continued capital intensity in connectivity infrastructure that constrains telco investment capacity.
What should a telco prioritise first when pursuing an AI strategy?
The foundational enablers come first: disciplined AI portfolio governance, a shared data and AI layer with governed data products and standardised tooling, and responsible AI, security, and compliance embedded by design. Without these, pathway-specific plays such as NaaS, GPUaaS, and sovereign AI platforms will struggle to scale.
Source: World Economic Forum and Accenture, The Strategic Role of Telecom Providers Across the AI Value Chain, March 2026. All statistics cited from the WEF white paper and its underlying sources, including IDC, Gartner, Omdia, Analysys Mason, Accenture research, GSMA, and TM Forum.
