On Sunday 8 November 2026 the Financial Times reported that Anthropic has agreed to pay about $3.4 billion for gas-fired generators to power its data centres. Bloomberg and CNBC relayed the report the same day. Anthropic and the supplier involved declined to comment, so every detail below is as reported by the FT and its relayers, not confirmed by either company.
What was reported
- Size: about $3.4 billion for 1.7 gigawatts of large, heavy-duty generators, described by the FT as the biggest known single purchase of power equipment by an AI company.
- Counterparty and timing: the commitments were bought in late September from OnePWR Solutions, a San Antonio, Texas equipment supplier.
- Structure: according to the FT, OnePWR had ordered the generators from Caterpillar roughly two years earlier. Anthropic first planned to buy power from OnePWR, then took over the generator orders instead. The deal reportedly includes engineering and construction capability for installation. Anthropic did not acquire OnePWR, and none of its staff moved to Anthropic.
- Location: where the generators will be used has not been disclosed.
What is not established
Anthropic and OnePWR did not respond to requests for comment, and Reuters, cited by CNBC, said it could not immediately verify the FT account. The figures rest on one FT report that cites a person familiar with the matter. No company statement or regulatory filing confirming the order has been identified.
Context
CNBC notes that demand for Anthropic’s Claude models has pushed the company to secure more computing and power capacity, and that Reuters reported in September that Anthropic expects to spend more than $500 billion over the next 10 years on AI infrastructure. Taking over an existing equipment order is one way a buyer can reach on-site power faster than by joining a new queue, though the sources reviewed here do not give delivery dates for this deal.
Why it matters to universities and research offices
This is editorial interpretation rather than reported fact. First, it illustrates that AI compute growth is now being met partly with on-site fossil generation, which is relevant to energy, sustainability and environmental-health researchers studying data-centre load, emissions and local air quality. Second, institutions weighing AI infrastructure, whether purchasing cloud capacity, joining compute partnerships or hosting facilities, may see power availability and sourcing become part of due diligence, procurement and sustainability reporting. Third, community scrutiny of data centres is rising: see our coverage of local voters rejecting data-centre measures and of the AI and data-centre bills planned for the new Congress.
