Meta's 'AI overbuild' finds a buyer as Anthropic eyes $10 billion computing power deal

Published 18 July 2026 · tech

Meta's massive investment in artificial intelligence infrastructure, once seen as a potential liability, may be turning into an unexpected business opportunity. Currently, Meta is in

Meta's massive investment in artificial intelligence infrastructure, once seen as a potential liability, may be turning into an unexpected business opportunity. Currently, Meta is in talks to lease computing power from its AI data centres to Anthropic in a deal worth up to $10 billion over two years, three people with knowledge of the discussions told The New York Times. While negotiations are still at an early stage and may not result in an agreement, if they are finalised, Meta could generate revenue from the enormous computing capacity it has built for its own AI ambitions. The talks come at a time when Wall Street has questioned whether technology companies are spending too aggressively on AI infrastructure.

Meta accepts it has overbuilt on AI Meta will spend as much as $145 billion this year, much of it on A.I., which would be more than double the $72 billion it spent last year. The company has faced questions about whether its own A.I. models can compete with those developed by rivals like Anthropic and OpenAI. And it has acknowledged that it may build more data centers than it needs, relative to the number of customers using its A.I. products. Ironically, the same concern might now be creating an opportunity. Selling excess computing power to companies like Anthropic could provide Meta with a new revenue stream until demand for its own A.I. services catches up.

Zuckerberg, in a recent call with investors, hinted that selling computing power could be one way for Meta to see some return on its A.I. investments. “Almost every week, there are different companies that come to us from the outside asking us about computing power that they could buy from us at some premium to what we’ve bought it at,” Zuckerberg said on an investor call in May. “We haven’t done that yet because we think we have a use for the compute.” “But obviously if we get to a point where we feel that we have overbuilt, then that is an option that we have, and that is partially what gives us confidence in investing in building this out,” he said.

The proposed Anthropic deal suggests that scenario may be closer than expected. Anthropic, which has seen soaring demand for its Claude Code software, has been scrambling to secure computing resources. Anthropic’s proposal to Meta was about a third of the size of a deal that the A.I. start-up signed with Elon Musk’s SpaceX in May. Under that agreement, Anthropic is paying the rocket company $45 billion over three years — or $1.25 billion a month — for computing power. Also Read | Microsoft trains sales team to talk down OpenAI, Anthropic AI models