The UK Power Grid Faces Issues with Undetected Data Centers

As data center developers vie for a piece of the massive investment pouring into the artificial intelligence sector, the queue to connect to the UK’s power grid has become congested with projects that may never materialize. This backlog is prolonging wait times for viable projects and complicating efforts to anticipate energy demand and plan for grid enhancements.

In July, the UK’s energy regulator, Ofgem, proposed a plan aimed at pushing phantom data centers out of the overcrowded queue. Under this proposal, which will be finalized post-industry feedback in September, developers will be required to submit a hefty, nonrefundable deposit that could soar into the hundreds of millions for the largest data centers. They will also need to secure customers ahead of time and demonstrate they have sufficient funding to complete their projects.

However, Ofgem faces a balancing act: the reforms must be stringent enough to dissuade speculators but not so harsh that they drive legitimate data center projects abroad, jeopardizing the UK’s goal to meet the skyrocketing demand for the computing power needed to support advanced AI models.

Data center developers are experiencing delays across the board; similar grid congestion issues plague the US and various European nations. Yet while the US remains a highly attractive market, industry experts warn that Ofgem’s reforms could render the UK—already hindered by high energy costs and limited land—one of the priciest locations globally for constructing data centers.

The rapid data center expansion “is bringing substantial capital investment into the UK,” states Alex Burgoyne, head of data centers at real estate consultancy Knight Frank. “We don’t want to damage the golden goose.”

Ofgem has stated that it will consider industry feedback before finalizing the fees, but the discrepancies between international energy markets make direct comparisons challenging. “We recognize that data centers are crucial to the UK’s AI ambitions and future economic growth. Facilitating quicker connections for viable data centers supports this,” Nathan Macwhinnie, deputy director of strategic planning and connections at Ofgem, tells WIRED in a statement.

The queue to connect to the UK grid began to swell towards the end of 2024, coinciding with the government’s designation of data centers as “critical national infrastructure.” From November 2024 to June 2025, the total energy demand from projects in the connection queue surged from 41 gigawatts to 125 gigawatts, according to Ofgem. Newly proposed data centers account for 73 gigawatts of this demand, which is equivalent to one and a half times the peak energy consumption for the entire UK last year—and that figure is still growing.

The government posits that a significant portion of this demand may be illusory. “It’s absolutely insane,” remarks Taco Engelaar, managing director at grid optimization company Neara. “No one really comprehends … what the actual grid demand will be due to these phantom projects.”

The crowded queue primarily stems from an incentives issue. Given the lengthy wait for grid access and the historically low cost of joining the queue—often just a few thousand dollars—developers have little reason not to apply for power, even without a solid plan for site development. This creates a risk-free proposition that allows developers to hedge against a future where demand for computing remains exceptionally high.

However, because grid operators need to evaluate the cumulative impact of large infrastructure proposals on network stability before granting individual connections, speculative data centers can worsen the already lengthy delays for genuine projects. “They must consider every project seriously when assessing whether the system can accommodate them,” explains Olivier Darmouni, associate professor of finance at HEC Paris Business School, who has researched the effects of AI on power grids. “The more speculative projects are particularly detrimental because they complicate these assessments, making them more expensive and time-consuming.”