Meta is planning to sell off its spare AI compute, and Wall Street likes the idea a lot more than the companies already doing it.
On Wednesday, Bloomberg reported that Meta is building a cloud business to lease its excess GPU capacity to outside companies. Meta shares jumped 9.3%, the stock's biggest intraday gain since April. CoreWeave dropped as much as 14%, and Nebius fell as much as 17%. Both are neoclouds, which is to say companies whose entire business is renting AI compute to firms that don't have their own.
That's a notable pivot for a company that just got cut off from Google's Gemini API. The pitch to shareholders writes itself: if Meta's own AI ambitions come up short, someone else's AI still needs GPUs to run on.
The plan has been quietly taking shape for months. At Meta's shareholder meeting in May, Mark Zuckerberg said outside companies were coming to Meta "almost every week" asking to buy compute or spin up an API service. If Meta ends up with more infrastructure than it needs, he added, selling the excess is "an option that we have, and that is partially what gives us confidence in investing in building this out."
Which is a polite way of hedging a very expensive bet. Meta has committed hundreds of billions of dollars to data centers and GPUs over the next few years, most of it earmarked to train and serve its own models. A cloud business gives that spending a second life if the models don't earn it back.
Gil Luria, managing director at D.A. Davidson, told Reuters that Meta entering the space is bad news for the neoclouds and not much of a threat to the giants. "CoreWeave and Nebius depend on Meta to grow, but Meta might not be as reliant on them," he said. The hyperscalers (AWS, Microsoft Azure, and Google Cloud) come with enterprise sales teams, compliance certifications, and a decade of trust that Meta doesn't have. The neoclouds don't have any of that either, which is why their stock got hit the hardest.
There's a weird wrinkle in the timing, though. Earlier this year Meta signed an infrastructure agreement with Nebius to buy GPU-as-a-service, disclosed in Nebius's SEC filings. So Meta is currently a customer of one of the companies it's about to compete with. Nebius CEO Arkady Volozh called the deal a way to "accelerate the build-out and growth of our core AI cloud business," a line that reads a little differently now that Meta is planning to enter that same business.
None of this is officially a product yet. Meta declined to comment, and the plans could still change. But the shareholder comments, the SEC filings, and the market reaction all point the same way. Meta has more compute than it can use, and it would rather sell the excess than let it sit idle.
There's a version of this where Meta becomes the fourth hyperscaler, undercuts everyone on price with ad-subsidized inference, and reshapes the cloud market the way DeepSeek reshaped reasoning models. There's also a version where enterprise buyers take one look at Meta's compliance story and stick with AWS. The real question is whether Meta actually wants to run a cloud business or just wants a hedge in case its own AI doesn't pay off. One takes a decade of enterprise sales work. The other takes a press release. So far Meta has only committed to the press release.
