Meta in early talks to lease AI compute to Anthropic
Anthropic is exploring a deal to rent computing power from Meta, signaling a potential new revenue stream for the social media giant to offset its massive AI infrastructure costs.
Anthropic is in preliminary discussions to lease computing power from Meta, according to a person familiar with the negotiations. While the talks remain in very early stages, a finalized agreement would mark a notable alignment between two of the largest players in the artificial intelligence sector.
The proposed structure follows a similar playbook Anthropic recently employed with Elon Musk’s SpaceX. Anthropic agreed to use the computing capacity at SpaceX’s Colossus 1 data center to alleviate capacity constraints for its paid subscriber base. Access to advanced Nvidia chips remains a persistent bottleneck for the AI industry, forcing Anthropic to implement strict usage limits on its most sophisticated models, including one known as Fable.
For Meta's investors, the potential leasing deal validates a strategy to monetize the company's staggering infrastructure outlays. Chief Executive Mark Zuckerberg told investors in May that Meta is considering entering the cloud computing market specifically to demonstrate that it can generate revenue from its AI hardware beyond internal product upgrades.
The financial scale of that infrastructure is immense. Meta forecasts capital expenditures could reach $145 billion in 2026, heavily weighted toward AI buildout. Transforming those servers into revenue-generating assets through third-party leasing would significantly improve the return profile of Meta's AI investments. Zuckerberg indicated last October that companies are regularly "asking if we have compute that they could buy from us at some premium to what we've bought it at."
Executing this cloud pivot requires institutional expertise, which Meta is actively acquiring. The company is bringing on Dave Brown, a former senior executive at Amazon Web Services, to guide these efforts. If Anthropic becomes a tenant, it would signal that Meta can successfully compete for external AI workloads against established cloud providers. Meta declined to comment on the ongoing discussions.
