AWS AI Factories could dash Telco’s sovereign AI dreams

New on-premises managed service targets data residency needs, challenging telecommunications providers

In sum – what we know:

AWS AI factories – A new managed service allowing enterprises and governments to deploy AWS-managed AI infrastructure (Trainium or NVIDIA) within their own data centers.

Sovereign AI push – The product explicitly targets the “sovereign AI” market, ensuring data remains within specific borders and meets strict regulatory compliance.

The telco gap – Despite having the physical assets to compete, analysts suggest telcos are missing the “sovereign AI” opportunity that AWS is now aggressively seizing.

At AWS re:Invent on December 2, 2025, Amazon Web Services unveiled AI Factories, a new managed service that brings AWS infrastructure directly into customer-owned data centers. The offering allows enterprises and governments to deploy AI workloads on AWS-managed hardware within their own facilities, combining the operational simplicity of public cloud with the control and data residency guarantees of on-premises infrastructure.

Under the model, customers provide physical space and power capacity, while AWS handles deployment, management, and maintenance of the underlying systems. Hardware options include AWS’s own Trainium3 accelerators or NVIDIA’s latest Blackwell GPUs, paired with AWS proprietary networking, high-performance storage, and security infrastructure. The service also integrates with Amazon Bedrock and SageMaker, giving customers immediate access to foundation models and tools for training and deploying large language models on proprietary data.

Each AI Factory environment is built exclusively for an individual customer or designated trusted community, with complete operational separation from other deployments. AWS says the approach can accelerate AI buildouts by months or years compared to organizations attempting to construct similar infrastructure independently.

The push for sovereign AI

AI Factories are explicitly positioned around the growing demand for sovereign AI, systems that reside within national borders and meet strict data residency and regulatory requirements. For governments and enterprises operating under digital sovereignty mandates, the product offers the technological capabilities of public cloud regions in an isolated, controlled environment.

The pitch is straightforward. Organizations can access AWS’s nearly two decades of cloud infrastructure expertise without sending sensitive data outside their own facilities. That’s a significant draw for government agencies, defense contractors, and regulated industries where data residency isn’t optional. By eliminating the typical delays in procurement, setup, and optimization, AWS is betting that speed-to-deployment will be a decisive factor for customers racing to adopt AI.

Telcos are missing the boat

The AI Factories announcement also highlights a strategic gap that telecommunications providers have yet to close. Telcos hold unique advantages in the sovereign AI market, with their existing data center footprints, established power infrastructure, and regulatory relationships. In theory, they’re well-positioned to offer competing sovereign AI solutions to enterprise and government customers.

In practice, that hasn’t happened. As Fierce Network noted in its analysis of the re:Invent announcements, telcos appear to be missing the boat on the sovereign AI opportunity, even as AWS moves aggressively to claim it. While telecommunications providers have the physical assets, they haven’t capitalized on them to deliver integrated AI infrastructure offerings that rival what AWS is now bringing to market.

AWS’s advantage isn’t just hardware. By bundling AI Factories with Bedrock and SageMaker, the company creates significant ecosystem lock-in. Customers don’t just get GPUs or accelerators; they get a fully managed AI platform with access to leading foundation models, training tools, and seamless integration with broader AWS services. That’s a value proposition telcos currently struggle to match.

For telecommunications providers, the risk is clear. As enterprises and governments prioritize sovereign AI capabilities, they may increasingly turn to hyperscalers like AWS rather than telcos for infrastructure. Without a credible counter-strategy, telcos could find themselves relegated to providing connectivity while AWS captures the higher-value AI infrastructure opportunity sitting inside their own data centers.