Meta has spent billions of dollars developing AI and building data centers to support it. But now the company may be preparing to use those data centers for more directly profitable purposes.
On Wednesday, Bloomberg reported that Meta is developing plans for a cloud infrastructure business that would sell access to both AI computing power and models. This move puts it in competition with major cloud providers such as Amazon Web Services, Google Cloud, and Microsoft Azure.
Meta’s decision to sell excess computing comes weeks after SpaceX announced similar plans through xAI. In early May, SpaceX signed a deal with Anthropic to buy all of the computing power in SpaceX’s Colossus 1 data center. SpaceX has signed similar leases since Google and Reflection AI. The fact that Meta is doing the same shows that the winner of the AI race may not be the one with the best model or service, but the one who owns the data center.
That is, if the demand for computing continues and the value of the data center is maintained. Some skeptics warn that the race to build AI infrastructure is creating a bubble that relies heavily on rapidly depreciating chips. Some question whether AI companies can generate enough end-user revenue to justify multi-trillion dollar bets.
These concerns haven’t stopped Meta from investing heavily in AI computing infrastructure. As of the end of the first quarter, Meta had committed to spending $182.9 billion on AI infrastructure over the next few years, including large projects underway in Louisiana and Ohio. The Ohio project, which Zuckerberg said will be the size of Manhattan, is scheduled to go online this year.
Unlike Google and OpenAI, Meta doesn’t see a huge demand for its AI models or services. Meta does not include revenue from Meta AI or its openweight AI model family, Llama, in its revenue, and executives have emphasized in public statements that the company primarily uses AI within the enterprise. That could mean Meta’s AI efforts have yet to reach a significant independent revenue stream.
According to Bloomberg, Meta may copy Coreweave’s business model and sell access to “raw” computing power to recoup some of its own massive spending. The outlet also reported that Meta is considering following AWS’ lead and selling access to various AI models, including the recently launched closed-weight model Muse Spark, hosted on its own AI infrastructure.
This new business line will be part of a new initiative reportedly called “Meta Computing,” led by head of infrastructure Santosh Janardhan, Meta Superintelligence Lab leader Daniel Gross, and president Dina Powell McCormick.
The report backs up Zuckerberg’s comments in May that a metacloud computing business is “definitely on the table” as a way to benefit from some of the company’s massive investment in a strategy to develop an AI “superintelligence.”
TechCrunch has reached out to Meta for comment.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
