Power Shortages Stall the UK’s ‘Largest AI Supercomputer’ for Years

A flagship AI data center project in the UK, billed as the country’s “largest AI supercomputer,” is now expected to be delayed by several years because it cannot secure enough power from the national grid. The facility is being developed by Neocloud Nscale, a cloud provider backed by Nvidia, and is designed to host massive numbers of GPUs for training and running large language models and other generative AI workloads.

According to local reports, Britain’s existing transmission network is struggling to keep up with the surge in demand from new AI and cloud facilities. The Neocloud site requires hundreds of megawatts of capacity, but grid connections of that scale are now facing extended queues and complex regulatory approvals. The company is exploring phased deployments and design tweaks, yet full-scale operations may remain out of reach for years.

The story underlines a growing reality in the AI race: compute power is not just about chips and models, but also about concrete, copper and cables. Across Europe, governments are wrestling with how to accommodate energy‑hungry data centers while still hitting climate goals and addressing local concerns. Policymakers are increasingly being pushed to treat grid upgrades and AI infrastructure as a single, coordinated challenge rather than two separate debates.