Powered land: how the data-centre and AI land rush is repricing the countryside

Something has changed in the market for large, flat, well-connected land, and a landowner would be forgiven for hearing about it before understanding it. From January 2026, the House of Commons Library notes in its briefing CBP-10315, large data centres can opt into the Nationally Significant Infrastructure Projects regime, the route normally reserved for airports and power stations. The Government is consulting on a data-centre National Policy Statement, and AI Growth Zones are being marked out to funnel this kind of development to particular places. Savills, in its Data Centre Spotlight for the first half of 2026, records power-ready industrial land reaching as high as £15m an acre in prime cases. Numbers like that travel fast, and by the time they reach a farm gate they have usually lost all their conditions.

So it is worth being precise about what is actually being priced, because it is not the field. We promote residential land for the most part, but we take commercial, logistics and data-centre sites through the system too, and the single most useful thing we can tell a landowner about this boom is that the value almost never sits in the acreage. It sits in the power, and in a grid connection you can rely on.

What actually makes a site viable

A data centre is, in land terms, a large shed that drinks electricity. The building is the easy part. What a developer is really buying is the ability to draw tens or hundreds of megawatts reliably, and to keep drawing it for decades. That means a firm connection to the grid, and firm connections are the scarce thing.

Baker McKenzie and Burges Salmon, both of whom advise on these schemes, keep returning to the same short list of what a site needs before it is worth anything to an operator. Start with real grid capacity nearby, not a line on a map but headroom at a substation that the network operator will actually contract. It needs a position in the connection queue managed by the National Energy System Operator, NESO, that resolves in a workable timeframe. Water matters too, because cooling these buildings at scale is a genuine constraint and increasingly a planning one. Then there is consent, whether through the ordinary planning system, the NSIP regime or an AI Growth Zone designation, and a route through it that can actually be delivered. Last comes a developer with the covenant and patience to see it through.

Miss any one of those and the site is not a data-centre site. It is a field next to a data-centre developer’s wish list.

Why the queue is the whole story

The gap between demand and deliverability is where the market actually lives. Barbour ABI counts around 119 data-centre projects in the planning pipeline. Against that sits a grid-connection backlog that, on current reform timelines, runs to something like twelve to fifteen years for new large loads. Both things are true at once: enormous appetite, and a system that cannot connect most of it any time soon.

That is why so many “we have been approached about a data centre” conversations never complete. An operator or a broker options a site cheaply, tests whether a connection can be found, and walks away when it cannot, which is most of the time. The landowner is left with a lapsed agreement and the impression that the boom passed them by. In truth the boom never reached them, because the electricity never did.

The scale of real demand is not in doubt. Savills projects UK data-centre capacity rising to somewhere between 3.3 and 6.3 gigawatts by 2030, and vacancy in existing stock was around 8% in the first quarter of 2026, tight by any measure. But projected demand and a connectable site in a particular parish are different facts, and only one of them pays.

What this means for a landowner

If your land sits near a major substation with genuine capacity, close to fibre, with water available and a plausible consent route, you may hold something rare and valuable, and the recent policy changes have made it more valuable, not less. If it does not, no headline price protects you, and the surest way to waste a good asset is to tie it up cheaply with the first party who mentions the word.

The honest first questions are not about the land at all. They are about the grid: what capacity exists nearby, and where any connection would sit in the NESO queue on a timeframe a serious operator would accept. Those answers take real work and they change as network reform moves. Our free Land Potential Check reads your land against national planning and constraint data and is a sensible place to begin, though it will tell you plainly that the grid position is something we establish by hand rather than pretend to read from a feed.

The countryside is being repriced, but unevenly and for reasons that have little to do with how the land looks. Understanding that before an option is signed is the difference between selling power and giving away a field.