You cannot run a data centre on North Sea oil.

On Sunday, 21 June, the US president named energy as one of two reasons Keir Starmer should go, with a specific instruction: "OPEN NORTH SEA OIL!" Leave the politics to others. As an engineering claim about Britain''s AI ambitions, it points at the wrong fuel entirely.

Start with what an AI data centre actually consumes. It needs firm, low-carbon electricity, delivered to one precise point on the grid, hundreds of megawatts at a time, around the clock. It does not run on oil. It barely touches gas except as expensive backup. The thing that decides whether a British AI campus gets built is not a barrel of crude in the North Sea; it is a wire, a substation and a generator with spare capacity at the right node.

And on that measure Britain is badly short. Ofgem reports that around 140 proposed data-centre projects are seeking roughly 50 gigawatts of grid capacity between them — more than Great Britain''s entire peak electricity demand, which was about 45GW on 11 February this year. The queue for demand connections tripled in seven months, from 41GW in November 2024 to about 125GW by June 2025. The government''s own forecast is that the country needs at least 6GW of AI-capable data centres by 2030 — roughly the output of four large nuclear reactors. That is the real bottleneck, and not one drop of it is about oil.

It is also a bottleneck that is already biting. Nearly £10bn of data centres were approved in 2025; less than £1bn was actually built. OpenAI''s Stargate UK, meant to anchor the North East AI Growth Zone, is paused — partly because of energy costs. The constraint is not a shortage of fuel in the ground. It is the price and availability of firm power and a connection to carry it.

So the country is having a loud argument about the wrong molecule. Whoever wins the contest that opens on 9 July inherits an AI strategy whose binding constraint is electrons — and an energy debate, imported wholesale from across the Atlantic, aimed squarely at the previous century''s fuel. North Sea oil is a real and contested question. It is simply not the question the AI build is failing on.

This is where the abstract lands in Britain physically. The work that AI removes leaves a very concrete residue on British soil: power draw, grid queues, and homes and factories waiting behind the data centres for the same scarce capacity. You can change the Prime Minister tonight. You cannot change the physics of the grid by Friday, and no amount of North Sea oil will connect a single megawatt of AI compute. If the new leader wants Britain to be an AI maker, the first honest sentence is about wires, not wells.