Summary

• Ofgem has proposed a Data Centre Commitment Fee and new milestones for large datacentre grid connection applicants.
• Demand connection applications have risen from 41GW to 125GW, with datacentres accounting for at least 80GW.
• The consultation links AI infrastructure growth directly to grid planning, consumer risk, and economic development.

Ofgem has proposed new rules to stop speculative datacentre projects from occupying scarce electricity grid capacity, after demand connection applications more than tripled in under a year.

The regulator has launched a consultation on a Data Centre Commitment Fee, which would be paid by large datacentre developments when accepting a grid connection offer. The fee would be refunded when a project reaches energisation, but forfeited if it exits the queue early. Ofgem has proposed a range of £237,500 to £712,500 per megawatt, equivalent to around 2.5% to 7.5% of average project costs.

The move follows a sharp increase in contracted demand offers, from 41GW in November 2024 to 125GW in June 2025. Ofgem said datacentre projects account for at least 80GW of that total, prompting concern that some applications may be speculative and could distort network planning, delay viable developments, and impose costs on consumers.

The consultation also proposes datacentre specific queue management milestones. Developers would have to provide evidence of tangible progress, including financial capability, commercial maturity, procurement activity, and credible end user commitment, if they want to retain their place in the connections queue.

Britain’s datacentre debate has often been framed around AI ambition and inward investment, but Ofgem’s intervention brings the energy system into the foreground. AI workloads, cloud expansion, and digital services all require physical infrastructure, and that infrastructure cannot be separated from grid capacity. A datacentre that looks viable in an investor presentation can still be useless if it cannot secure power at the right location and time.

The regulator is not treating all datacentre demand as illegitimate. Britain needs more compute capacity if it wants to support AI, cloud services, scientific computing, financial infrastructure, and digital public services. Speculative projects become a problem when they occupy connection capacity before they are ready to build, creating a queue that looks larger than deliverable demand and making network planning harder.

That uncertainty reaches beyond the datacentre sector. If grid capacity is reserved for projects that do not progress, hospitals, housing developments, factories, transport electrification schemes, renewable energy projects, and deliverable digital infrastructure can all be affected. Consumers may also be exposed to inefficient network investment if planners respond to inflated demand signals.

The proposals would change the economics of early stage datacentre planning. A credible operator with committed customers, capital, and procurement plans may view the fee as an acceptable cost of certainty. A speculative landholder or underdeveloped project could face a material financial risk. Ofgem is trying to separate real demand from options trading on grid access.

The consultation also complicates the UK’s AI growth agenda. Support for datacentres, AI growth zones, and digital infrastructure will have to be matched by energy planning that can distinguish strategic projects from paper demand. The issue is not simply whether Britain wants more datacentres, but where they should be located, how they should be powered, and which projects deserve priority when the grid is constrained.

Cloud providers and enterprise buyers will watch the consultation closely because grid access ultimately affects capacity availability, resilience, and cost. If the queue becomes cleaner and more credible, viable projects may connect faster. If fees are set too high or milestones are poorly designed, the regime could deter investment or favour larger players with deeper balance sheets.