OpenAI pauses UK data centre
OpenAI has paused its Stargate UK data‑centre project after concluding industrial electricity prices in the UK are about four times higher than in the U.S. and citing unresolved copyright rules. The company framed the decision around costs and regulation, showing how energy pricing and legal uncertainty can redirect where AI infrastructure is built. That pause is a concrete reminder that geography, energy and policy now factor into platform strategy. (thenextweb.com)
OpenAI has paused its Stargate UK data-centre plan less than seven months after unveiling it with Nvidia and Nscale in September 2025, and the company says the two blockers are power prices and regulation. A data centre for artificial intelligence is basically a warehouse full of specialised chips, and those chips turn electricity into model training and chatbot answers all day. When power is expensive, every prompt, fine-tune, and new model gets more expensive too. The original Stargate UK pitch was not small. Nscale said the broader UK commitment involved up to 58,640 Nvidia graphics processing units, and OpenAI’s own UK project was meant to start by leasing up to 8,000 advanced chips before scaling further. The first site was tied to Cobalt Park in North Tyneside, inside one of Britain’s new Artificial Intelligence Growth Zones. Those zones were designed to speed up planning decisions and improve access to the electricity grid for large computing projects. That is why the pause is awkward for London. The UK government signed a memorandum of understanding with OpenAI in July 2025 to explore infrastructure, public-service use, and technical cooperation, then spent months selling Britain as a place where big AI facilities could be built quickly. The energy problem is concrete, not rhetorical. UK government data shows Britain had the highest industrial electricity prices in the European comparison set in late 2024, and OpenAI said UK industrial power costs are about four times US levels. The regulation problem is also specific. Britain is still arguing over whether AI companies can train models on copyrighted material under a broader text-and-data-mining exception, and the government’s March 2026 consultation kept that fight alive instead of settling it. For an AI company, that uncertainty works like building a factory on land where the property lines are still being debated. If the legal rules on training data change after the servers are installed, the economics of the whole site can change with them. OpenAI has not walked away from Britain entirely. The company said it still sees “huge potential” in the UK, will keep working with the government, and will move forward when energy costs and regulation support long-term infrastructure investment. What changed in one week is not the physics of AI chips but the map of where they make financial sense. The race to host artificial intelligence is now being decided not just by talent and tax breaks, but by who can offer cheap power, fast grid access, and rules that do not move halfway through construction.
