2 hours ago
5
When OpenAI announced Stargate UK last September, British officials practically tripped over themselves celebrating what they called “a major step forward in the US-UK technology partnership.” A multibillion-pound data center project. Sovereign AI compute capacity. Thousands of GPUs humming away in North East England. The whole thing sounded transformative.
Here’s the thing: OpenAI apparently never bothered to visit one of the key planned sites. And roughly £20 billion of the £30 billion investment figure that UK ministers enthusiastically promoted turns out to have been, to use a generous word, hypothetical.
The gap between announcement and reality
The Stargate UK partnership was unveiled on September 16, 2025, bringing together OpenAI, NVIDIA, and UK cloud provider Nscale. The plan targeted sites in North East England, including Cobalt Park and Blyth, with an initial goal of securing up to 8,000 GPUs by the first quarter of 2026. Expansion to 31,000 GPUs was floated as a longer-term possibility.
That expansion never came close to materializing. On April 9, 2026, OpenAI paused the entire project. The company pointed to UK electricity costs, reportedly four times higher than in the US, and uncertainty around AI copyright regulations as the primary reasons.
UK AI Minister Kanishka Narayan pushed back hard on that framing. He argued that neither electricity pricing nor the regulatory environment had meaningfully changed since the project was announced. In other words: if these were dealbreakers, they should have been dealbreakers from day one.
Narayan implied that OpenAI’s own financial viability might have been the more honest explanation for the pause.
The £30 billion question
The scrutiny deepened on July 4, 2026, when it emerged that OpenAI had never actually visited a key planned site for the project. More damaging: roughly two-thirds of the headline £30 billion investment figure appears to have been speculative. Only about £10 billion represented commitments with any real substance. The remaining £20 billion was better described as “potential” investment.
What this means for AI infrastructure investment
The UK’s electricity costs remain a genuine competitive disadvantage. Data centers are extraordinarily power-hungry, and when your electricity runs roughly four times the cost of a rival jurisdiction, the math gets difficult fast.
The discrepancy between OpenAI’s stated reasons and the UK government’s counterarguments also introduces a credibility variable. When the two sides of a major technology partnership publicly disagree about why a project stalled, it signals that future AI investment announcements from either party deserve more skepticism, not less.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) · 