As data center developers compete for a share of the hundreds of billions of dollars flowing into the artificial intelligence industry, Britain’s electricity grid is facing an influx of connection applications from projects that may never be built. The surge is worsening already lengthy delays for viable developments, while making it harder to forecast energy demand and plan the grid expansion needed to support AI infrastructure.
In July, the UK energy regulator Ofgem proposed reforms to clear speculative data center projects from the grid connection queue. The plan, which is expected to be finalized after an industry consultation ending in September, would require developers to pay substantial non-refundable deposits. For the largest data centers, those deposits could reach hundreds of millions of dollars. Developers would also need to secure customers in advance and demonstrate that they have sufficient funding to complete construction.
Ofgem faces a difficult balancing act. The reforms must be strict enough to discourage speculative applications without driving legitimate data center projects to other countries or undermining the UK’s ambition to become a leader in artificial intelligence. AI companies and cloud providers are already making substantial demands for computing capacity to train and operate advanced AI models.
Data center developers are experiencing grid connection delays in many major markets. Similar electricity network constraints are delaying data center construction in the United States and across Europe. Although the US remains a highly attractive market, industry experts warn that Ofgem’s proposed requirements could make the UK—already challenged by high energy costs and limited land availability—one of the world’s most expensive locations for data center development.
Alex Burgoyne, head of data centers at property consultancy Knight Frank, said the construction boom was “bringing huge amounts of capital investment to the UK.” He added: “We don’t want to shoot the golden goose.”
Ofgem said it would consider industry concerns before setting the final charges, while noting that differences between international energy markets make direct comparisons difficult. “We recognize that data centers are a key part of the UK’s AI ambitions and future economic growth,” Nathan McWhinney, Ofgem’s deputy director of strategic planning and connectivity, told WIRED in a statement. “Enabling viable data centers to connect faster is what will enable this.”
The UK data center grid connection queue began expanding rapidly around the end of 2024, after the government designated data centers as critical national infrastructure. According to Ofgem, the total electricity demand represented by projects in the connection queue rose from 41 gigawatts in November 2024 to 125 gigawatts in June 2025. New data centers accounted for 73 gigawatts of that total—more than 1.5 times the UK’s peak electricity demand.
The government said the projected growth in electricity demand far exceeds market forecasts and the network’s capacity. Much of that demand may never materialize. “It’s absolutely insane,” said Taco Engerer, managing director of power grid optimization company Neara. “With all these phantom projects, no one really understands what the real grid demand is going to be.”
The crowded grid connection queue is largely the result of economic incentives. Developers can wait years for access to the electricity network, while the cost of joining the queue was previously only a few thousand dollars. As a result, companies may apply for power before they have a fully developed construction plan. The application becomes a relatively low-cost option on a future in which demand for AI computing remains high.
However, speculative or “phantom” data centers can increase delays for viable projects. Grid operators must assess the potential impact of every major infrastructure proposal on network stability before approving individual connections. “When you try to study whether a system can handle a project, you have to treat every project as serious,” said Olivier Darmouni, an associate professor of finance at HEC Paris Business School. Research on AI’s impact on the power grid found that speculative projects make these assessments more complex, expensive and time-consuming.
Source: www.wired.com