Wall Street reacted before Meta even confirmed anything. On July 1, shares of Meta jumped 8.8% after Bloomberg and CNBC reported the company is weighing whether to turn its spare AI computing capacity into a cloud business, selling access to outside customers. Meanwhile, CoreWeave and Nebius, two of the smaller “neocloud” firms that already sell that kind of cloud capacity, fell 10.8% and 12.4% on the same news.
That reaction alone says something about how investors read this. Specifically, Meta has committed as much as $145 billion in AI capital spending this year, on top of $182.9 billion in future lease obligations tied largely to data centers built for cloud-scale AI work. For a while, the question hanging over that spending was simple: how does Meta actually make money from it. A cloud offering is one possible answer.
The effort reportedly sits inside a group called Meta Compute, led by infrastructure chief Santosh Janardhan, Superintelligence Labs’ Daniel Gross, and Meta president Dina Powell McCormick. Two cloud approaches are apparently on the table. One resembles AWS Bedrock, where developers pay for cloud access to hosted models, including Meta’s Muse Spark, without managing the underlying hardware. The other is more straightforward: selling raw cloud compute, similar to what CoreWeave already does.
Zuckerberg himself hasn’t been coy about this. On a May shareholder call, he said outside companies had already approached Meta about both cloud options, and that selling compute was “definitely on the table.” The holdup, he said, was that Meta still expects to use most of that capacity internally.
If Meta does move forward, it would be stepping into a cloud market that already looks crowded and, frankly, a little tense. AWS, Azure, and Google Cloud dominate at scale, and neoclouds like CoreWeave have built entire businesses around GPU rental. Even so, Meta’s edge in this cloud race is unusual: it isn’t starting from scratch. It’s asking whether to monetize infrastructure it was always going to buy anyway.
Whether that’s enough to justify the scale of Meta’s cloud ambitions is still an open question. For now, though, investors seem to like the odds better than they like CoreWeave’s.
