Meta and Anthropic announced a computing capacity leasing deal reportedly worth up to $10 billion, positioning Meta as a cloud-compute supplier to rival AI labs.

Meta Platforms is in talks with Anthropic to lease computing power from its AI data centers under a potential two-year agreement worth up to $10 billion, according to The New York Times. Anthropic proposed the plan in June, and Meta is currently evaluating it with terms still being negotiated. Anthropic would make monthly payments, with either party retaining the right to terminate early.

The computing capacity represents approximately one-third of Anthropic's contract with SpaceX, signed in May. Negotiations remain in early stages, and the deal is not guaranteed to close.

Meta's move reflects a broader strategy of monetizing its substantial data center infrastructure, a path similar to Amazon Web Services' early approach to leasing computing capacity to external AI labs. For Meta, the arrangement optimizes capital returns while deepening industry collaboration. For Anthropic, it reduces supplier concentration by diversifying its computing power sources.

This dynamic mirrors patterns elsewhere in the AI infrastructure ecosystem. Microsoft and OpenAI have deep computing commitments, while Google Cloud provides computing services to external models. Currently, both supply and demand sides of AI infrastructure are accelerating resource matching through flexible leasing arrangements.

Fundamentally, the trend reflects capital consolidation: leading AI labs are hedging risk by distributing workloads across multiple suppliers, while cloud giants lock in long-term cash flows by leasing idle capacity. Pricing power concentrates with platforms controlling large-scale data center assets, and the industry is maturing toward a dedicated computing-power-as-a-service market.