Can The UK Power Its AI Ambitions?

Britain’s electricity connection queue has become rather crowded. In less than a year, demand applications jumped from 41GW to 125GW, with data centre projects accounting for at least 80GW of the total, according to Ofgem.

That doesn’t mean Britain is about to build 80GW of data centres. Many of the projects in the queue may never materialise, and Ofgem is already looking at ways to prevent speculative developments from taking up capacity that could go to projects that are actually ready to proceed. What the figure does reveal is the extraordinary scale of interest in building the infrastructure behind our growing use of AI.

The government’s own UK Compute Roadmap forecasts that Britain will need at least 6GW of AI-capable data centre capacity by 2030, around three times the capacity available when the forecast was made. If AI develops and is adopted more quickly than expected, it acknowledges that demand could be significantly higher.

Suddenly, Britain’s ambition to become an AI leader isn’t only a question of technology, investment or skills. It’s also a question of whether we have enough power and infrastructure to support it.

AI Has Become An Energy Story

When we use AI, very little about the experience feels physical. We type something into a box, wait a few seconds and get an answer back, with almost no indication of everything required to make that interaction possible.

Behind it sit chips, servers, cooling systems and enormous data centres, all of which ultimately need electricity. As more organisations adopt AI and more powerful systems are developed, expanding the UK’s computing capacity also means working out where that additional power is going to come from.

This matters particularly because Britain isn’t trying to expand AI infrastructure in isolation. At the same time, the country is investing in a cleaner electricity system and dealing with an existing grid that wasn’t designed with huge clusters of AI data centres in mind.

That doesn’t mean the UK’s AI and clean-energy ambitions are necessarily in conflict. New renewable generation, grid upgrades, storage and other forms of energy infrastructure can all contribute to meeting growing demand, and the government’s Compute Roadmap explicitly recognises that energy provision needs to be part of planning future AI infrastructure. What it does mean is that the two ambitions are increasingly connected.

Infrastructure Moves At A Different Speed

Part of the challenge is that AI and energy infrastructure operate on very different timescales. A new model can appear and attract millions of users remarkably quickly, while building data centres, securing grid connections and expanding electricity infrastructure can take years.

That makes forecasting particularly difficult. Nobody knows exactly how quickly AI adoption will grow, how efficient future hardware will become or how many of the data centre projects currently seeking connections will ever actually be built. Waiting until the demand is certain, however, isn’t much of an option when the infrastructure needed to meet it can’t be created overnight.

The enormous connection queue is therefore interesting even if much of it ultimately disappears. It shows just how much potential demand is already knocking on the door of an electricity system that has plenty of other things competing for its attention.

For all the discussion about models, investment and Britain’s position in the global AI race, one of the biggest constraints on that ambition may turn out to be considerably less glamorous. Building the AI infrastructure is one thing; making sure there’s somewhere to plug it all in is another.