The queue of companies seeking to plug into Great Britain’s electricity grid has reached staggering new heights, with energy regulator Ofgem warning that a surge in applications is threatening the nation’s decarbonisation goals.
At the heart of the demand is an unprecedented rush from datacentre developers, driven by the global boom in artificial intelligence. According to an Ofgem consultation document, roughly 140 proposed datacentre schemes could collectively require 50 gigawatts of electricity—a colossal figure that alone exceeds the country’s current peak demand by 5GW.
A Grid Under Siege from AI’s Appetite
This staggering power requirement is fuelled by the central role datacentres play in the AI revolution. These facilities form the essential infrastructure for training and operating complex AI models, from chatbots like ChatGPT to image generators. Their immense computing power comes with an equally immense energy cost, creating a “global challenge” for power networks.
Ofgem pointed to a “surge in demand” for new grid connections between November 2024 and June last year, a wave that has “exceeded even the most ambitious forecasts”. The regulator cautions that the work required to connect these power-hungry facilities could lead to significant delays for other projects it describes as “critical for decarbonisation and economic growth”.
This backlog is compounded by a parallel slowdown in connecting new renewable energy projects to the grid. These projects are being built to help meet the government’s clean energy targets by 2030, but they are not being integrated into the network at the necessary pace. The combined effect places the UK’s target of a virtually carbon-free power system by 2030 in further doubt, a goal already under scrutiny due to concerns over rising electricity costs.
The Environmental and Logistical Tightrope
The environmental impact of this datacentre expansion can be severe. The Guardian revealed last year that a single vast datacentre proposed for Elsham in Lincolnshire could, according to its findings, cause more greenhouse gas emissions than five international airports.
While some technology leaders and climate experts believe AI could ultimately aid the fight against global heating—by optimising power grids or accelerating green technology development—there are widespread concerns. In the near-term, the fear is that datacentres’ soaring energy demands will drive increased reliance on fossil fuels to keep them running.
Ofgem also warned that the current system is counterproductive, allowing unviable applications for grid access to block progress for important, strategic bids. This includes projects related to the government’s designated AI growth zones. These zones, announced last year as part of plans to boost UK AI adoption, were touted as offering a streamlined planning process and help in accessing energy, but they lack a mechanism for prioritisation within the grid queue.
Proposals to Unclog the System
In response, the regulator has proposed a series of tougher financial tests designed to filter out speculative projects. The aim is to prevent a backlog of poorly-funded schemes from delaying viable projects further down the queue.
One key proposal is to charge datacentre providers for access to a connection, potentially through a deposit or a non-refundable fee, which could deter “nonviable” applications. Ofgem stated that datacentres must be central to any reform of the connection application process.
The regulator is also exploring a more fundamental shift: whether datacentre developers should pay for and build their own grid connections. Ofgem suggests this approach could “accelerate connections and deliver better outcomes for consumers” by transferring the burden and cost of infrastructure expansion directly to those creating the new demand.
