Meta has spent years pouring money into AI. Now it wants that infrastructure to start paying its own way. According to TechCrunch, Meta is developing plans for a cloud infrastructure business that would sell access to both raw computing power and AI models, putting it in direct competition with Amazon Web Services, Google Cloud, and Microsoft Azure.
The initiative, reportedly called Meta Compute, is led by a heavyweight team: Santosh Janardhan, Meta’s head of infrastructure; Daniel Gross, head of Meta Superintelligence Labs; and Dina Powell McCormick, the company’s president. The move confirms what Mark Zuckerberg said back in May, when he described a cloud computing business as “definitely on the table.”
This is not a small pivot. As of the end of Q1, Meta had committed to spending $182.9 billion on AI infrastructure over the coming years. That includes major data center projects in Louisiana and Ohio. The Ohio facility, which Zuckerberg described as the size of Manhattan, is expected to come online this year.
Meta is not the first company to think this way. SpaceX got there first. In early May, via its xAI subsidiary, SpaceX signed a deal with Anthropic to sell out all compute capacity at its Colossus 1 data center. It has since signed similar agreements with Google and Reflection AI. The pattern is clear: companies that built AI infrastructure at scale are now finding that selling access to it may be more immediately profitable than whatever AI products they built it for in the first place.
Bloomberg reports Meta is considering two approaches:
- Selling “raw” compute capacity, similar to what CoreWeave does
- Hosting and selling access to various AI models, following the AWS model, including its recently launched closed-weight model, Muse Spark
The timing makes sense when you look at Meta’s revenue picture. Unlike Google or OpenAI, Meta has not shown strong public demand for its own AI models and services. The company does not break out revenue from Meta AI or its Llama model family in earnings reports. Executives have mostly talked about AI in terms of internal business uses, which suggests AI has not yet become a standalone revenue line that moves the needle. Selling compute to others would change that math quickly.
That said, the broader question hanging over the entire industry is whether this infrastructure boom is built on solid ground. Some analysts have warned that the rush to build AI data centers looks like a bubble, one that relies on chips that lose value fast. Others have asked whether AI companies can generate enough consumer and enterprise revenue to justify the trillion-dollar investments being made across the sector. Meta’s plan to monetize spare capacity does not answer that question, but it does show that big players are increasingly hedging their bets by treating the infrastructure itself as the product.
If the compute-as-a-service model takes off for Meta, it would signal a broader shift in how the AI race gets won. The advantage may not go to whoever builds the best chatbot or the most capable model. It may go to whoever owns the most data center space at the right moment. TechCrunch has reached out to Meta for comment.
