Meta is reportedly in discussions to lease computing power to Anthropic in a deal potentially worth up to $10 billion over two years, according to a report from The New York Times. The arrangement aims to provide Meta with a new revenue stream while meeting Anthropic’s urgent need for computational resources.
Anthropic first proposed this deal in June, and it involves monthly payments over the two-year period, with an early-exit clause available to either party. Sources noted that the proposed deal represents a significant reduction compared to a previous agreement Anthropic signed with SpaceX, in which it pays approximately $1.25 billion monthly, totaling $45 billion over three years.
Discussions between the two companies are still in the early stages, and both Meta and Anthropic have declined to comment on the negotiations. Increased demand for computational resources among leading AI firms, including Meta, Google, and Microsoft, has intensified the urgency for such arrangements, as these firms invest considerable funds into new data centers globally.
The rapid expansion in AI infrastructure has raised concerns among investors about the high costs associated with building these facilities, and whether these expenditures will yield satisfactory returns. Mark Zuckerberg announced plans for Meta to spend up to $145 billion on AI in the current year, more than doubling the $72 billion spent in the previous year, although he has expressed doubts regarding the effectiveness of these investments.
Meta has acknowledged that it may construct more data centers than currently necessary for its AI needs. This potential oversupply could make leasing excess computing capacity to others, such as Anthropic, a practical solution. Competitive pressures have led some companies, like Anthropic, to seek cooperation with direct rivals, while their valuation and demand for AI resources continue to rise.
Meta’s extensive leasing agreements include a $21 billion contract with CoreWeave and a $27 billion arrangement with Nebius. The evolving landscape of AI infrastructure leasing indicates an increasing trend where companies explore alternative revenue sources from their surplus capabilities.
