Meta is reportedly developing a cloud infrastructure business, selling access to both AI compute power and models. This move comes as SpaceX's xAI recently signed deals to sell out its own data center capacity. The strategy signals that owning data centers, not just having the best models, could be key in the AI race.

Meta has committed over $182 billion to AI infrastructure, with large projects underway in Louisiana and Ohio. However, unlike Google and OpenAI, Meta hasn’t seen significant revenue from its own AI services, leading it to explore selling raw compute capacity as a potential return on investment.

The new business line, dubbed Meta Compute, is led by key executives and aims to follow the model of AWS in offering access to various AI models hosted on their infrastructure. This could pit Meta directly against established cloud providers like Amazon Web Services, Google Cloud, and Microsoft Azure.

While this move may bring about increased competition, it also raises concerns about whether the demand for AI compute will continue to grow at the pace needed to justify such investments. Critics warn of a potential bubble in rapidly depreciating chips, with some questioning if AI companies can generate enough revenue from end-users to make these massive bets worthwhile.