The Strategic Role of Telecom Providers Across the AI Value Chain 2026

GPUaaS GPUaaS extends telco infrastructure into the compute layer. Operators host shared GPU clusters in regional or edge data centres and provide enterprises with virtualized GPU capacity accessible through APIs or the cloud. GPUaaS enables model training and inference close to users, reducing latency and driving performance and security. A growing cohort of operators has launched GPUaaS platforms to address growing demand. Customer demand is bound to be significant, with AI model development and inferencing expected to outstrip the global supply of high-performance GPUs. Gartner projects a global AI-optimized infrastructure-as-a-service (IaaS) – of which GPUaaS is a segment – market size of $108 billion29 by 2029, driven by the attractiveness of on-demand AI computing without intensive capital expenditure. Telcos’ right to play is challenged. They have local infrastructure, fibre and regulatory trust, which hyperscalers cannot easily replicate. However, they need to differentiate on proximity, compliance and service integration rather than raw compute scale (which, today, is a race to the bottom). Additionally, markets with protectionist policies present an advantage to domestic suppliers, including CSPs. Monetization is driven by the fractionalization of GPU infrastructure through multi-tenancy, enabling operators that amortize capital investment across multiple customers, optimize operating costs through shared use and resell capacity as a service, often complemented by managed, orchestration or integration add-on services. Horizontal and vertical AI solutions Vertical AI solutions are industry-optimized AI offerings, co-developed with partners, that combine telco infrastructure with telco/partner AI agents, software and data expertise. These include managed AI model hosting, specialized LLMs, pre-built analytics applications and hybrid edge-to- cloud environments, designed to address industry- specific challenges tied to industry key performance indicators (KPIs). CSPs are harnessing network intelligence, interoperable APIs, edge compute and industry expertise to co-develop AI applications tailored for industries such as manufacturing, finance and healthcare. Horizontal solutions provide AI agents that are valid across industries and business functions. Examples include content creation tools and contact centre experience for customer service. Most telcos have experimented with these capabilities36 for internal use and can capitalize on their knowledge by productizing and selling these use cases to their enterprise and public administration customers. Customer demand is strong, with 94% of enterprises37 expecting AI productized solutions to play a significant role in their business over the next three years. Sectors like manufacturing, healthcare and financial services require domain-specific models, low latency and industry compliance – needs that general-purpose cloud solutions often can’t meet. CSPs’ right to play is driven by combined edge infrastructure, network data and industry partnerships to co-create AI products for manufacturing, finance and healthcare. Although many enterprise AI solutions exist, CSP advantage stems from the strong integration of use cases with network capabilities or emerges where telcos are already trusted technology providers, as for SMEs in certain regions. Monetization methods are built on use- and results-based pricing; vertical solutions are scalable across sectors after proof of concept. CASE STUDY 6 Global operators scale enterprise AI through distributed GPU infrastructure A broad set of telcos – including Iliad (via Scaleway),30 Indosat (Lintasarta),31 Singtel,32 SK Telecom,33 Telenor34 and Verizon35 – have launched GPUaaS offerings to meet surging enterprise demand for AI compute. These platforms provide low-latency, local access to GPU clusters without large upfront investment, helping enterprises shorten provisioning cycles and improve data compliance. The Strategic Role of Telecom Providers across the AI Value Chain 14