Nebius commits £1.7 billion to expand UK AI capacity using NVIDIA infrastructure

Nebius, an AI cloud infrastructure provider listed on Nasdaq, said it is investing about £1.7 billion to expand its artificial intelligence compute capacity in the United Kingdom, including three new deployments built around NVIDIA infrastructure. The move is positioned as both an expansion of Nebius’s hosting capabilities and an effort to deepen its role as a UK-based base layer for companies that want to train, test, and run advanced AI systems.

In a statement announcing the expansion, Nebius said its first UK deployment using NVIDIA Blackwell Ultra infrastructure began in November 2025. The company said the three new sites will also deploy what it described as NVIDIA’s “full-stack, end-to-end AI factory platform technology,” and that the combined deployments are expected to reach 65 megawatts when fully ramped up in 2027.

The company framed the additional power and deployment sites as critical for what it called “agentic and enterprise AI,” referring to AI systems designed to take actions on behalf of users or business workflows, rather than simply generating text or outputs. Nebius said UK customers include organizations in financial services and healthcare and life sciences, though it did not specify new customer names tied directly to the new deployments.

Nebius also tied its investment to UK policy priorities. The company said the build-out aligns with the UK Government’s AI Opportunities Action Plan, and it argued that local capacity reduces the friction for British enterprises, researchers, and public services that want to build and deploy AI at scale within the country.

For NVIDIA, the announcements are another indicator of demand for data center GPU capacity and the ecosystem that wraps around it. NVIDIA’s Blackwell platform is part of its broader data center strategy, and Nebius’s use of Blackwell Ultra plus additional NVIDIA “AI factory” stack components suggests Nebius expects its customers to need not only GPUs but also an end-to-end path from model development to production deployment.

From Nebius’s perspective, the expansion also reinforces its positioning as a compute and platform provider rather than a traditional software vendor. Nebius described the London hub as both a commercial base and an AI research and development center, and the investment is presented as the physical capacity behind those efforts. If executed on schedule, the company’s ramp to 65 MW by 2027 would materially increase its ability to serve organizations that need sustained, high-density GPU access.

Even with the added detail, key operating specifics were not fully disclosed in the public statement. Nebius did not provide the identities of the four sites contributing to the UK capacity figure, did not break down how much of the £1.7 billion is expected to go to power and data center build-out versus hardware and software, and did not specify the exact NVIDIA components beyond Blackwell Ultra and the reference to NVIDIA’s end-to-end “AI factory” technology.

Investors and the market will likely focus next on whether the timeline is met and how quickly Nebius can fill the capacity as new deployments come online. Watch for further disclosures on customer commitments tied to the new UK sites, updates on ramp schedules toward 65 MW, and any additional partnerships or leasing agreements that would announcement long-term demand for GPU-backed AI hosting in the UK.