Meta Negotiates to Lease AI Data Center Computing Power to Anthropic
According to The New York Times, Meta Platforms is in talks with Anthropic to lease computing power from its AI data centers, with a two-year agreement potentially worth up to $10 billion.
Anthropic proposed the plan in June, and Meta is currently evaluating it, with transaction details being adjusted. Anthropic will make monthly payments, and either party can terminate the agreement early.
The scale of this computing power procurement is only one-third of the contract Anthropic signed with SpaceX in May this year, and negotiations are in the early stages, so the deal may not materialize.
Source: Public Information
ABAB AI Insight
Meta has previously invested heavily in building its own AI data centers, and this leasing of idle computing power continues its strategy of monetizing through infrastructure, similar to Amazon AWS's early path of leasing computing power to external AI labs.
Meta's allocation of data center resources to Anthropic is motivated by optimizing capital returns and deepening industry collaboration, while Anthropic aims to reduce reliance on a single supplier by diversifying its sources of computing power.
Similar to the deep computing power binding between Microsoft and OpenAI, as well as Google Cloud providing computing power services to external models, Meta and Anthropic are currently in a phase where both sides of AI infrastructure supply and demand are accelerating resource matching through flexible leasing.
Essentially, this represents capital concentration: top AI labs are diversifying risks across a multi-supplier computing network, while cloud giants are locking in long-term cash flow by leasing idle capacity. Pricing power is concentrating with platforms that own large-scale data center assets, and the industry chain is being restructured into a mature market for computing power as a service.
ABAB News · Cognitive Laws
- During the computing power surplus period, idle resources quickly convert into high rental income.
- Top labs diversify supplier reliance, while infrastructure giants seize the opportunity to secure long-term cash flow.
- The outcome of the AI race is determined by model capabilities, but the scale of computing power leasing often indicates the next round of capital pathways.