Sovereign AI is becoming infrastructure, not marketing copy. On October 8, 2026, Eminence Grey and Lumen Technologies announced a partnership to connect sovereign AI environments across the United States, combining dedicated compute with programmable networking.
The target market includes government agencies, enterprises, and AI developers that need to connect distributed workloads while keeping control over their data and operating environments. Customers will be able to reserve pre-architected systems or request custom designs, with availability, configuration, and implementation determined by their requirements.
Eminence Grey brings AI nodes, GPU and high-performance computing clusters, its Whitehorse command-and-control plane, and Outrider access appliances. Lumen supplies the programmable connectivity linking those environments, supporting distributed workloads, resilience, and operational continuity—because a powerful AI system is less useful when its data cannot move safely.
The combined platform is designed around dedicated infrastructure, verifiable isolation, and zero internet exposure across the data lifecycle. Eminence Grey describes Whitehorse as a proprietary, patent-pending control plane that coordinates its private AI fabric across locations.
Control now includes the network
Jim Fowler, Lumen’s chief technology and product officer, framed the partnership around movement rather than storage: “Sovereign AI isn’t just about where data lives. It’s about how data moves.” That distinction matters for organizations spreading AI workloads across multiple environments, where network design becomes part of data control.
The American effort arrives as sovereign compute initiatives expand beyond individual facilities. The National Compute Grid has 760 megawatts connected or in sight, with a goal of reaching 2 gigawatts by 2030. Anjney Midha leads that effort, adding a national-scale dimension to the push for controlled AI resources.
Europe is pursuing the same objective under its own constraints. European organizations in regulated sectors are seeking sovereign AI solutions, with 62% of organizations across Europe and 76% in banking looking for that control. Europe was already seeking greater AI sovereignty in November 2025, while European spending on sovereign cloud infrastructure is expected to more than triple between 2025 and 2027.
TensorX shows the demand behind the infrastructure
TensorX, an Irish sovereign AI inference company, offers a useful measure of that demand. On October 8, 2026, the company reached $5 million in revenue, served more than 5,000 paying customers, and appointed Tim Grant as chief executive officer.
TensorX delivers inference on open-source AI models using dedicated NVIDIA GPUs with zero data retention. It owns a fleet of 96 NVIDIA B300 GPUs in Ireland and has additional GPUs in Finland, while expanding its infrastructure across Europe and planning to grow its team to meet client demand.
The usage numbers have moved far beyond an early-stage experiment. Monthly token volume on TensorX’s shared inference platform rose from 5.63 billion in February 2026 to 1.1 trillion in September 2026, a 171% increase over seven months; September alone added 451 billion tokens.
During the same period, active customer domains increased by more than 1,000, while monthly API requests climbed from 264,917 to 34.5 million. The arithmetic is not subtle: companies want access to capable models, but they want the processing environment under their control.
“Progressive businesses are building real AI powered applications right now and they want access to the most performant open source AI models as they evolve, but they also need control over where their data is processed and the infrastructure that runs it,” Grant said.
TensorX has raised $10 million in seed funding led by Darius Cubed Ventures. Shane Morton founded the company, and its customer Usman Khan, CEO of APEX:E3, said: “TensorX turbo charged the output of our development team and enabled us to deploy our own A3 coding assistant. TensorX is simply the only platform we trust with our most sensitive data.”
The two developments point to the same market shift: sovereign AI now depends on compute, connectivity, isolation, and predictable access working as one system. OpenAI and Anthropic may dominate public attention, but the infrastructure race is spreading toward private deployments, open-source models, and regional control. The glamorous part is the model. The difficult part is everything around it.
Based on
