Meta Explores $10 Billion Computing Lease Deal with Anthropic

What happened

Meta Platforms entered discussions with Anthropic in June about leasing computing capacity from its AI data centers. Under the proposed terms, Anthropic would make monthly payments over a two-year period for access to Meta's infrastructure, with provisions allowing either party to exit early. The talks remain preliminary and may not result in an agreement. This follows Anthropic's $45 billion three-year computing deal with SpaceX in May, which the Meta proposal would be roughly one-third the size of at approximately $1.25 billion monthly. The discussions highlight the scarcity of computing power as AI companies race to expand their capabilities.

Why it matters

This potential deal represents a strategic opportunity for Meta to monetize its massive infrastructure investments and diversify revenue beyond advertising. With Meta planning to spend $125-145 billion in capital expenditures this year-more than double 2024's $72 billion-investors have questioned whether such spending can be justified. CEO Mark Zuckerberg has acknowledged the company may overbuild capacity relative to internal needs, and selling excess computing power could provide returns while demand for Meta's own AI products develops. The arrangement would position Meta alongside Amazon, Microsoft and Google in the cloud computing market, though Meta currently lacks an established business selling computing services. The company has already secured its own computing capacity through deals worth $21 billion with CoreWeave and $27 billion with Nebius, while computing power prices have surged due to limited supply.

Bigger picture

The talks illustrate the extraordinary demand for AI infrastructure across the technology sector. Major companies are collectively investing hundreds of billions of dollars in data center construction and advanced semiconductors as they compete to develop cutting-edge AI models. Anthropic's willingness to work with Meta-a company developing competing AI models through its Llama family-demonstrates how computing scarcity is reshaping traditional competitive dynamics. The AI startup has secured computing agreements with multiple providers including Google, Amazon and Microsoft alongside the SpaceX deal. Meta faces pressure to demonstrate that its AI investments can compete with models from Anthropic and OpenAI, particularly as Meta shares have declined more than 8% year-over-year. The company recently released an upgraded Muse Spark AI model and introduced its first paid AI service, signaling efforts to generate direct AI revenue.

What to watch

Monitor whether Meta and Anthropic finalize an agreement and the specific financial terms if announced. Track Meta's quarterly infrastructure spending updates and any signals about excess capacity utilization. Watch for indications that Meta is actively pursuing cloud computing as a formal business line beyond one-off deals. Observe competitive responses from established cloud providers Amazon Web Services, Microsoft Azure and Google Cloud. Pay attention to Meta's AI product adoption metrics and whether internal demand for computing resources matches the company's buildout pace.