When ministers and tech executives hailed Microsoft’s £22 billion UK investment package, one project stood out from the dense jargon of “compute capacity” and “data infrastructure”: Stargate UK.
The name might sound like a science fiction franchise, but Stargate UK is in fact Britain’s most ambitious supercomputing initiative to date — a programme designed to provide the raw computing power needed to train the next generation of artificial intelligence (AI) models on British soil.
Stargate UK takes its name from Stargate, the $500 billion US project announced earlier this year by OpenAI and SoftBank. That American initiative aims to build an unprecedented network of AI data centres capable of hosting trillions of operations per second, fuelling the world’s most advanced generative AI.
Britain’s version is more modest in scale, but strategically vital. It is being positioned by the government as Europe’s largest AI supercomputer effort — a joint partnership between Nvidia, the US chipmaker behind the world’s most powerful AI processors; NScale, a British data-centre business; and OpenAI, the San Francisco-based creator of ChatGPT.
The government hopes that Stargate UK will secure Britain’s place as an “AI maker, not an AI taker”, ensuring British researchers, startups and industries have direct access to cutting-edge computing power rather than relying entirely on US or Chinese capacity.
AI models such as ChatGPT, Google Gemini or Anthropic’s Claude require astronomical amounts of computing power to train. This is provided by GPUs (graphics processing units), which can handle many calculations in parallel. Training a state-of-the-art model can take tens of thousands of GPUs running for months, consuming as much electricity as a small town.
Until now, Britain’s computing capacity has lagged behind. The government’s flagship Isambard-AI project, launched in 2023, was designed to run on around 5,500 GPUs. By comparison, Stargate UK is expected to scale up to 31,000 GPUs by 2026 — many of them Nvidia’s new Grace-Blackwell Ultra processors, among the most powerful AI chips in existence.
That leap would put the UK closer to global peers, offering a viable domestic platform for training advanced models in areas like finance, defence, life sciences and climate science, where sovereignty over sensitive data is critical.
The initiative is being structured around guaranteed demand. Microsoft has committed billions to Britain’s AI infrastructure and is partnering with NScale, while OpenAI is expected to be one of Stargate UK’s first anchor customers, potentially using up to 8,000 GPUs in early 2026 and scaling up sharply from there.
The idea is simple: by locking in customers in advance, the consortium makes it financially feasible to build data centres of this scale. Construction will be spread across multiple sites, with Cobalt Park in northeast England earmarked as a central hub. The area has been designated an AI Growth Zone, where planning approvals and energy connections are expected to be fast-tracked.
Why “sovereign compute” matters
AI has become not just an economic race, but a geopolitical one. Nations are vying to ensure they control at least some of the “compute” — the hardware and software capacity — needed to run large AI models. Without it, countries risk being locked into dependence on foreign suppliers, with sensitive data leaving national borders.
By hosting OpenAI’s most advanced systems in UK-based data centres, operating under British regulatory rules, the government hopes to guarantee that national security, defence and financial institutions can use these tools without compromising confidentiality.
David Hogan, Nvidia’s vice president of enterprise, summed it up: “The only thing that’s been missing in the UK is infrastructure. This will truly make Britain an AI maker, not an AI taker.”
The numbers underline the ambition. Microsoft has pledged £22 billion over four years, half of it for capital expansion. Nvidia is allocating 120,000 GPUs to the UK, its largest European deployment, with about half being the ultra-powerful Grace-Blackwell Ultra chips. CoreWeave, a US AI infrastructure firm, will also invest £1.5 billion to expand capacity in Britain.
Of these chips, about 60,000 are expected to go directly into the Stargate UK supercomputer. For context, that is more than ten times the capacity of Isambard-AI.
The potential payoff is significant: faster training of AI models, new breakthroughs in science, and thousands of high-skilled jobs in data-centre management, engineering and research.
The challenges ahead
Despite the fanfare, Stargate UK faces real hurdles.
• Energy demand: Data centres of this scale consume huge amounts of electricity. Google has already announced a partnership with Shell to stabilise clean energy supplies for its new UK centre, and Stargate UK will face similar scrutiny over its environmental footprint.
• Cost overruns: Building AI infrastructure is capital-intensive. Locking in customers like OpenAI helps, but delays or budget overruns could strain finances.
• Talent shortages: The UK will need more data scientists, engineers and technicians to run and maintain this infrastructure.
• Geopolitics: With President Trump pushing hard for US dominance in AI, Britain will need to balance partnership with Washington while ensuring domestic priorities are not sidelined.
The “Stargate” brand is symbolic. It references both the US project and the sense of entering a new era of computing, where AI is not just another software tool but an operating layer for the economy.
For ministers, the name conveys ambition: that Britain is not retreating from global competition, but stepping through its own “stargate” into a future powered by AI.
Stargate UK is not yet built, but if successful it will mark a turning point for Britain’s digital economy. It represents a bid to anchor world-class AI capacity on UK soil, reduce dependence on foreign compute, and keep Britain at the forefront of the next technological revolution.
As one government insider put it, “This is about sovereignty, science, and staying in the race.”
Read more:
What is Stargate UK? Britain’s new AI supercomputer project explained
