Overview
- OpenAI, which disclosed the pause Thursday, said high power costs and an uncertain regulatory environment led it to halt the UK phase of its Stargate plan.
- The project with Nvidia and Nscale was pitched to give the UK “sovereign compute” by hosting OpenAI models on local servers, starting with up to 8,000 GPUs and potentially scaling to about 31,000 across sites such as Cobalt Park and Blyth.
- Industry figures point to steep industrial electricity prices, long waits for grid connections, and unsettled copyright policy as key hurdles, with about 140 data‑center projects seeking roughly 50 gigawatts of capacity against Britain’s historical peak demand of about 45 gigawatts.
- OpenAI says London remains its largest research hub outside the US, it will keep hiring and delivering on a government memorandum of understanding for public‑sector AI, and it is still in talks with Nscale about the project’s future.
- The government says it is working with OpenAI to strengthen UK compute capacity, while opposition politicians blame energy policy, and some industry voices frame the move as a prudent pause rather than a permanent withdrawal.