Ofgem's Proposed Data Centre Connection Reforms: What Would Change?

Executive summary

Ofgem is consulting on making data centre developers put money and evidence behind their grid connection applications. Its proposed reforms — published on 29 July 2026 and closed to responses in September 2026 — would introduce a refundable Data Centre Commitment Fee for projects of 40 MW and above, plus additional queue milestones requiring evidence of commercial demand and procurement progress. Nothing here is yet in force. Ofgem has said it intends to decide later in 2026, and the detail that matters most to developers — the exact fee level, exemptions and transitional treatment — is precisely the detail still open.

Why this matters to the UK

Britain's connections queue is the gatekeeper for almost every large electricity project, whether that is a wind farm, a battery site or a data centre. When the queue fills with applications that may never be built, projects that would be built wait behind them. Ofgem's consultation states that contracted demand connection offers rose from 41 GW in November 2024 to 125 GW in June 2025, of which around 73 GW across roughly 315 projects are data centres. For comparison, Britain's peak electricity demand in winter 2025/26 was around 45 GW.

That comparison is easy to misuse. Queue capacity is requested or contracted connection capacity — a maximum permitted import at a point on the network, at some future date. It is not electricity being consumed today, and it is not a forecast of consumption. A site can hold a 100 MW connection and draw far less. We explain the distinction in why data centre megawatt announcements can mislead.

A figure worth watching carefully

Ofgem's own materials are not internally consistent on the data centre share. The consultation document works from roughly 73 GW; the accompanying press release refers to data centres seeking "at least 80GWs" of capacity while elsewhere citing the same 73 GW consultation figure. Both statements come from the same regulator on the same day. Until Ofgem reconciles them, the honest position is that the data centre share of the queue is somewhere in the region of 73–80 GW, depending on the cut-off date and how projects are classified — not a single settled number.

The problem Ofgem says it is trying to solve

Ofgem's argument is that the connections process currently costs an applicant very little relative to the value of holding a queue position. A developer can secure a place, hold it while options are explored, and withdraw late with limited consequence. The consultation also notes that around 9 GW of transmission queue projects were modified from battery storage to data centres between May 2024 and August 2025 — a change of technology within an existing queue position rather than a new application.

The consultation frames the scale in capital terms: at an indicative average of around £9.5m per MW, the data centre capacity in the queue would imply capital expenditure of a magnitude comparable to a substantial share of UK GDP. Ofgem's inference is not that all of this will be built — it is that plainly not all of it can be, and the network is currently planning around applications that carry no test of seriousness.

What is actually being proposed

Who would be in scope

The threshold Ofgem consulted on is 40 MW, applied to transmission-connected projects and to distribution projects subject to a Transmission Entry Assessment. Ofgem's supporting analysis indicates that projects above 40 MW account for the overwhelming majority of queued data centre capacity, and that very few large distribution-connected projects would fall outside the test. The consultation also sets out exemptions for projects close to connection and transitional treatment for projects already holding offers — the principle is consulted on, the precise cut-offs are not settled.

What it could mean in practice

For developers. Capital would be tied up earlier, and optionality would become expensive. Well-funded operators with confirmed customers are the least affected; the model of securing several sites and choosing later becomes markedly harder. Smaller developers and those building speculatively for a future tenant carry the most exposure.

For network planning. If the fee and milestones work as intended, the queue would shrink towards projects with evidenced demand, and network investment could be planned against something closer to real load. If they do not, the risk is a queue that is smaller but no better evidenced — filtered by access to capital rather than by likelihood of delivery.

For consumers. Ofgem's stated rationale is that speculative applications distort network investment, and that consumers ultimately fund networks built for demand that does not arrive. That is a reasonable argument, but it is a proposition about avoided future cost, not a measured saving. No verified consumer bill impact has been published, and we would treat any specific figure with caution until Ofgem quantifies it in a decision document.

What remains unknown

• The final fee level within — or outside — the consulted 2.5–7.5% range, and which adjustment governance option is adopted.
• The precise exemptions and transitional arrangements for projects already holding connection offers.
• Whether the 40 MW threshold survives consultation unchanged.
• How the data centre definition operates where a site is mixed-use, or where compute is one part of a larger campus.
• Whether the reforms measurably reduce the queue. That can only be judged after implementation, against published queue data.

Readers tracking this should watch for Ofgem's decision document, the resulting CUSC modification, and NESO's subsequent queue reporting. We record connection-queue evidence as it is published in the grid pressure section of the UK AI Energy Index, and individual UK projects in the data centre tracker.

Sources

• Ofgem, Proposed data centre connection reforms (consultation, 29 July 2026) — the consultation document itself, including the fee design, thresholds and milestones.
• Ofgem press release, Ofgem acts to free grid capacity by tackling speculative data centre projects (29 July 2026).