Britain’s electricity regulator wants to stop speculative data-center projects from occupying scarce places in the national grid connection queue. The proposed reform follows a surge in applications that could leave credible AI infrastructure waiting behind facilities that may never be built.
Ofgem’s July proposal would require developers of the largest data centers to put down nonrefundable deposits, potentially worth hundreds of millions of pounds. Applicants would also need to show evidence of customers and financing. The regulator is accepting feedback until September, so the final rules and thresholds have not yet been decided.
The problem is not a shortage of ambition. It is a queue whose requested capacity has become detached from what the power system could plausibly deliver.
A cheap application created an expensive bottleneck
Requests for data-center connections rose from 41 gigawatts in November 2024 to 125 gigawatts by June 2025. New projects accounted for 73 gigawatts of that total, about one and a half times the United Kingdom’s peak electricity demand in the previous year.
Those figures do not mean 73 gigawatts of data centers will be constructed. Applying for a grid connection could cost only a few thousand dollars, according to the Wired report, giving developers little financial reason to limit speculative submissions. A queue position could increase the value of land or become an asset to sell even when the applicant lacked a firm customer or a realistic plan to finance construction.
The result is a distorted signal. Grid operators see enormous projected demand, while serious developers must wait behind projects that may exist primarily on paper. Ofgem has already tightened the process since 2024 by requiring land rights and planning milestones. Its proposed deposits would add a much stronger test of commitment.
That test is intended to make speculation costly. For an applicant seeking a very large connection, walking away would no longer mean losing only a modest filing fee. It could mean forfeiting a substantial deposit.
Deposits could also exclude viable newcomers
The regulator must decide how to discourage queue manipulation without reserving the market for the richest operators. Large cloud companies can absorb a major deposit more easily than a smaller infrastructure business, even when both have genuine projects.
Wired points to Nscale as the kind of newer operator that could face pressure from an aggressive threshold. A company may have customers and technical plans but still be raising the capital needed to build. Requiring too much cash too early could favor established groups and reduce competition before construction begins.
Evidence requirements present a similar tension. Proof of financing and customer demand can expose applications built on little more than land and a connection request. Yet infrastructure projects often assemble those elements in stages. Rules that demand every commercial commitment at the start could reject projects whose sequencing is legitimate rather than speculative.
Ofgem’s consultation therefore has to define more than a fee. It must determine when a project becomes credible, how evidence should be assessed and whether deposits should scale with capacity or development progress. The proposal describes a direction; it does not establish that every phantom project will disappear or that every viable one will survive.
Grid reform cannot remove Britain’s cost challenge
Clearing the queue would help credible data centers receive more reliable timelines, but it would not make Britain an automatically attractive place to build. Developers still face expensive energy and land, and overly burdensome rules could encourage projects to move elsewhere.
That creates a difficult policy balance. The United Kingdom wants the computing capacity associated with AI development, yet the physical system cannot reserve power indefinitely for applicants with no firm path to construction. At the same time, a screening process that only global incumbents can satisfy would narrow the field of potential builders.
The most immediate value of reform may be better information. A queue backed by meaningful deposits, customers and financing would give planners a more realistic view of future electricity demand. Developers would gain greater confidence that the projects ahead of them are real, while grid investment decisions could rely less on inflated applications.
Ofgem’s final design will determine whether that credibility is purchased at a reasonable price. The regulator has identified the phantom-project problem; its harder task is removing speculative demand without turning access to the grid into a privilege available only to companies with the deepest balance sheets.
