Once computing power lands on a map, its costs are no longer abstract

The scale of the proposal is already concrete

A report in this time window says British company Xlinks has proposed a 1.5GW AI data centre and a 1.8GW battery-storage facility near Great Torrington in north Devon, on about 344 hectares within a UNESCO-designated biosphere reserve. The plan includes five large buildings; the company says only about a third of the land would be developed, with the rest open or used for habitat creation.

The reporting also describes the scale of the opposition: local groups have been filing information requests, and a recent petition had received about 30,000 signatures. Those numbers do not settle whether the project is right or wrong, but they show it is no longer just a line in an industrial plan.

A proposal is not an approval, and a promise is not an outcome

At the time of publication, the district council said it had not received a formal planning application; the earlier pre-application discussions were standard procedure and did not amount to approval. The planned public consultation had also been postponed, with no new date announced.

The developer says the project would bring £13.8 billion in investment, 600 to 1,200 jobs and clean-energy infrastructure. Residents and campaigners point to farmland and the protected landscape, water use, noise and battery-facility risks. It matters to keep both accounts visible: the economic benefits are currently the developer’s projections, not delivered outcomes; environmental concerns still need to be tested against a complete design and assessment.

The water estimate still depends on a new design that has not been published

North Devon was still under a hosepipe ban at the time. The developer says the campus would recirculate cooling water and rely mainly on rainwater for top-ups; however, a freedom-of-information response found that under extreme heat and severe drought, annual water use in the old design could reach about 2.9 billion litres. The company later said that estimate referred to a superseded design and that a new design would be published during the consultation and planning process.

That is not a settled water-use figure but a question the new design still has to answer. Compute capacity and investment totals are easy to headline; the details of water, electricity, land and long-term operation are what more directly determine whether a community is willing to host a project.

The real disagreement is about who gets to see the trade-offs

To me, the debate is not simply for or against AI. The more concrete questions are whether residents can see a sufficiently complete, verifiable account of energy and water needs before a decision is made; how the developer’s employment and investment promises would reach the area; which effects could be mitigated; and which costs could not be avoided.

At the time covered by the report, the formal application, new design and a date for the public consultation were still undecided. Whether the project will be built, how much resource it will actually require and whether the community will have a meaningful opportunity to participate remain unknown. Keeping those unknowns in the discussion is more honest than treating a proposal as a fait accompli.

Sources for this reading

- The Guardian, published 27 Sep 2026 at 03:00 EDT, updated 06:02 EDT (Beijing time 15:00, updated 18:02): “People are standing up and fighting back”: the north Devon revolt against a vast AI datacentre