Nebius announced plans to expand its UK AI infrastructure footprint across four sites, targeting 65 MW of capacity by 2027 and adding NVIDIA-powered services for enterprise and agentic AI workloads.
Nebius announced an expansion of its UK AI infrastructure plans, saying it will commit about £1.7 billion to four sites and target 65 MW of capacity by 2027.
In a company announcement published by Nebius and distributed through Business Wire, the AI cloud provider said the UK expansion is intended to support growing demand for training, inference and enterprise AI workloads. The Business Wire release says the plan includes NVIDIA-powered infrastructure and additional cloud capabilities aimed at “agentic and enterprise AI.”
Nebius said the investment will cover four UK locations and that the buildout is expected to reach 65 MW by 2027. The company framed the move as part of its broader strategy to provide AI computing capacity for companies building and running production AI systems.
Because the figures come from Nebius’s own announcement, they should be read as company-stated plans rather than independently verified capacity now in operation. The announcement does not, in the excerpts provided, detail the full construction timeline for each site or disclose all technical specifications.
The Business Wire version of the release says Nebius cited Revolut and Prima Mente as customer examples. According to the same release, Revolut rebuilt its AI stack on Nebius. Nasdaq’s republication of the Business Wire material also states that Revolut rebuilt its AI stack on Nebius.
Nebius positioned the customer examples as evidence of demand for production AI services, rather than only experimental model development. The company also described workloads using Nebius Token Factory, a service it says is designed to support training and inference economics.
The announcement links the UK capacity expansion to a broader product push around production AI. Nasdaq’s Business Wire copy says Tavily, Eigen AI, Clarifai and Nebius AI Cloud 3.5 are associated with expanded capabilities for agentic search, inference and serverless production AI.
Nebius’s announcement emphasizes NVIDIA-powered infrastructure. The company’s release says the UK expansion includes more NVIDIA infrastructure deployments, though the provided excerpts do not specify every GPU model or the allocation across each UK site.
The emphasis reflects a broader market pattern: AI cloud providers are competing to secure and deploy high-performance accelerator capacity for customers training large models, running inference at scale and building AI applications. In this case, the available sources support only Nebius’s stated investment, capacity target, customer examples and product descriptions.
For UK companies developing AI systems, the planned expansion could increase access to local or regional AI computing options if Nebius completes the buildout as described. The company is pitching the investment not only as raw computing capacity, but as a platform for production workloads, including training, inference, agentic search and serverless AI services.
The announcement also illustrates how AI infrastructure providers are increasingly packaging hardware access together with software services intended to manage cost and deployment complexity. Nebius’s references to Token Factory, AI Cloud 3.5 and partner/customer examples suggest the company wants to differentiate its offering beyond renting accelerator capacity.
Still, the core facts currently come from Nebius and syndicated versions of its announcement by Business Wire and Nasdaq. The most concrete claims are the stated £1.7 billion commitment, the plan for four UK sites, the 65 MW target by 2027, the use of NVIDIA-powered infrastructure, and the customer and product examples named in the release.
Nebius announced an expansion of its UK AI infrastructure plans, saying it will commit about £1.7 billion to four sites and target 65 MW of capacity by 2027.
The company framed the move as part of its broader strategy to provide AI computing capacity for companies building and running production AI systems.
Because the figures come from Nebius’s own announcement, they should be read as company stated plans rather than independently verified capacity now in operation.
Continue reading
