Meta Platforms is reportedly in discussions to lease computing capacity to Anthropic in a potential arrangement valued at up to $10bn over a two-year period.
The talks are at an early stage and may not lead to a deal, Reuters reported, citing an undisclosed source.
Go deeper with GlobalData
Access deeper industry intelligence
Experience unmatched clarity with a single platform that combines unique data, AI, and human expertise.
The proposal, initiated by Anthropic in June, would see the Claude developer pay Meta in monthly instalments for access to its infrastructure, the source indicated. Both parties would retain the option to exit the agreement before the two-year period ends.
Meta, which has primarily focused on advertising and does not currently operate a business selling computing power, is considering Anthropic’s offer but the negotiations have become complex as a result.
Shares of Meta fell more than 2% at close on Friday, retracting some earlier losses following reports of the talks.
The potential partnership signals a move by Meta to expand beyond advertising revenue by offering access to its infrastructure. This would align the company with firms like CoreWeave and Nebius, which provide leased computing resources amid rising demand from companies developing advanced AI tools.
Anthropic has previously sought access to significant computing resources. In May, the company entered an agreement with SpaceX to use the Colossus 1 data centre in Memphis, Tennessee.
Meta’s CEO Mark Zuckerberg, addressing shareholders in May, said that moving into cloud computing was “definitely on the table”. Zuckerberg added that the company was approached “almost every week” with requests for access to AI models or surplus computing power.
In product developments, Meta discontinued the AI image generation tool Muse Image earlier this month following criticism from privacy advocates and industry groups. The tool had enabled the creation of images using public Instagram accounts. Concerns were also voiced by a Hollywood union.
Last month, Anthropic suspended access to its Fable 5 and Mythos 5 AI models for all foreign nationals, following a directive from the US government under export control law. The restriction applied to both external users and Anthropic staff who are classified as foreign nationals, regardless of their physical location.
However, by June-end, the US government had removed the export restrictions, enabling Anthropic to put the models back into service after agreeing to tighter safety standards and enhanced monitoring measures.
