A power station just became an AI data center — and the energy company holds the cards
On November 3, 2026, Drax Group announced it has signed an agreement with Brookfield and Radiant to build a gigawatt-scale AI data center on the Drax Power Station estate in North Yorkshire, England. Brookfield estimates total investment, including the computer chips, could exceed £20 billion — about $26.8 billion. The first phase targets up to 100 megawatts of computing capacity, running from the first quarter of 2028. The deal is not really about computers. It is about what the energy company already owns: power, grid access, and land.
Why a power station is the prize
For the biggest AI data centers, power — not chips — has become the hardest thing to secure. A new grid connection in Britain can take years to arrange. Drax's site skips that queue: the 1,000-acre estate already has about 4 gigawatts of grid access and 2.6 gigawatts of generation it can switch on when needed. Radiant, Brookfield's AI infrastructure platform, will design, build, and operate the facility, and the capital comes through the Brookfield-Qai AI infrastructure joint venture. The plan runs in three phases: up to 100 megawatts from early 2028, a further 500 megawatts between 2028 and 2031, and more capacity after that, taking the site past 1 gigawatt in total. Drax chief executive Will Gardiner said the project would help secure the long-term future of the site and protect hundreds of jobs.
What it is not
It is not a finished project, and it is not a data-licensing deal. Nothing is built yet. The first phase still needs planning permission, regulatory and network approvals, a tenant to use the facility, commercial agreements, and a final investment decision from each partner. No customer has been named and no chip supplier has been disclosed. And the £20 billion figure is Brookfield's estimate of total potential investment, not money already spent. What matters is the direction: AI money is flowing to energy companies that hold scarce assets — and those assets are worth more than the power they generate.
What it means for an American energy company
The same week, OpenAI announced its Grid Protectors Program: under a 25-year agreement with Georgia Power, the utility can cut up to 1,000 megawatts from OpenAI's planned 3.2-gigawatt campus in Effingham County, Georgia, when the grid is under strain. That deal runs on the utility's grid data and its ability to flex demand — operational knowledge no AI company can buy off the shelf. The pattern is the same on both sides of the Atlantic: AI companies need energy companies twice over — for power and for data. The Drax deal is about power and land. But the same scarcity logic applies to the records energy companies already hold: maintenance logs, equipment performance histories, inspection records, operating procedures, grid data. That is the record of how physical operations actually run, and it is exactly what AI models cannot learn from the public internet. And energy is one of the few industries that can qualify for licensing at any size — the free check routes your company to the right program.
Wondering if your company qualifies? The free check takes about five minutes, and there is no obligation. The AI companies pay you, and you're never out of pocket.
Free Data Check →Source: Drax Group Plc, "Strategic partnership for GW-scale data centre," regulatory news service, November 3, 2026, via TradingView / Reuters.