Artificial intelligence continues to reshape the landscape of the tech business: Anthropic, the company behind the Claude model, has made Meta an offer that could fundamentally change the approach to monetizing AI infrastructure. The proposal involves a two-year contract for renting computing power worth up to $10 billion. The offer was made in June, and Meta is currently reviewing it, with discussions focusing on a monthly payment format that includes an early exit option for both parties.
This move is notable because Meta, traditionally focused on the advertising model, could gain a new revenue stream. Currently, the company does not have a ready-made business for selling computing power, but it already rents such capacity from providers like CoreWeave and Nebius. In April, Meta signed a $21 billion agreement with CoreWeave, and in March, a $27 billion deal with Nebius. Thus, Anthropic's offer not only paves the way for competing with cloud giants but also allows Meta to justify the massive capital expenditures on AI infrastructure, which are estimated to reach $145 billion in 2026—more than double last year's $72 billion.
Investors have already expressed concern over the scale of Meta's spending on data centers, and selling excess capacity could serve as a response to this criticism. The company itself acknowledges that it may build more facilities than it needs. Notably, Meta is strengthening its cloud team: former Amazon Web Services top executive Dave Brown has joined, signaling serious intentions.
Demand for computing resources at Anthropic is growing amid the popularity of Claude Code. Earlier, the startup entered into a similar deal with Elon Musk's SpaceX for $45 billion over three years, utilizing the Colossus 1 data center in Memphis. Analysts note that the supply shortage in the capacity market is forcing even competitors to negotiate with each other.
My expert commentary: This case is a vivid example of how the race for computing resources is transforming the business models of tech giants. By investing in AI infrastructure, Meta is not just preparing for the future but also creating an asset that can be monetized right now. If the deal goes through, it will set a precedent that forces other players to reconsider their strategies. However, Meta's lack of a ready-made cloud business is a key risk that could slow down implementation.