Britain’s plan to build its largest AI supercomputer has hit the most mundane of obstacles: it cannot get enough electricity quickly enough. The datacentre project in Loughton, Essex, promoted by the government in 2025 as a sovereign AI flagship, will miss its planned launch next year and may not come fully online until the early to mid-2030s, according to reporting by The Guardian.
The operator, Nscale, had been counting on a 90 megawatt power link to the site. Connections of that size are not plug-and-play; they require grid reinforcement work that National Grid has warned could take years across the country. The delay crystallises a problem ministers have discussed in general terms for months. The UK wants a leading role in artificial intelligence, and AI runs on power, cooling and land, three things the British planning and grid systems ration carefully.
The government framed the Loughton facility as critical national infrastructure when it was announced, part of a package of commitments to make Britain an AI hub. A project that slips from 2027 into the 2030s weakens that story, and rivals have not been idle. The BBC has reported on the vast scale of Chinese datacentre construction in Inner Mongolia, built at what it described as “China speed”, while United States operators secure power agreements measured in gigawatts.
Analysts say the episode should reshape how AI infrastructure policy is written. Grid connection queues in Britain are among the longest in Europe, with some renewable projects waiting a decade for a slot. Data centres, which consume power constantly rather than seasonally, compete in that same queue. Until government treats compute sites as schedulable loads, perhaps with battery buffering or demand flexibility obligations, the queues will keep lengthening and the launch dates will keep moving.
Nscale has reportedly explored alternative power solutions, including on-site generation, which carries its own political difficulty in a country committed to clean power by 2030. None of that stops the immediate harm to the UK’s positioning. Universities and startups that were promised access to domestic supercomputing capacity will keep booking time on American clouds instead, and the money will follow it.