Meta follows SpaceX in monetizing excess AI compute power
▼ Summary
– Meta is developing plans for a cloud infrastructure business to sell access to AI compute power and models, competing with AWS, Google Cloud, and Microsoft Azure.
– The move follows SpaceX’s similar decision to sell excess compute capacity, signaling that data center ownership may be key to winning the AI race.
– Meta has committed to spending $182.9 billion on AI infrastructure, including a massive project in Ohio expected to come online this year.
– Meta has not seen significant revenue from its own AI models like Llama, and may sell access to raw compute and models like Muse Spark to generate returns.
– The new initiative, reportedly called Meta Compute, is led by Santosh Janardhan, Daniel Gross, and Dina Powell McCormick.
Meta has poured tens of billions into artificial intelligence and the sprawling data centers that power it. Now, the company appears ready to turn those facilities into a more direct moneymaker.
According to a report from Bloomberg on Wednesday, Meta is laying groundwork for a cloud infrastructure business that would sell access to both AI compute power and its own models. That would place the social media giant in direct competition with the likes of Amazon Web Services, Google Cloud, and Microsoft Azure.
This move to monetize excess compute capacity mirrors a recent strategy from SpaceX. Through its xAI arm, SpaceX struck a deal with Anthropic in early May to lease the entirety of its Colossus 1 data center’s compute power. Similar leases have since been signed with Google and Reflection AI. Meta’s decision signals a potential shift in the AI race: the ultimate winners may not be those with the best models or services, but rather those who control the data centers themselves.
That thesis, however, hinges on sustained demand for compute and the long-term value of these facilities. Skeptics warn that the frantic buildout of AI infrastructure could be a bubble, one heavily dependent on rapidly depreciating chips. There are also lingering doubts about whether AI companies can generate enough end-user revenue to justify the trillion-dollar bets being placed.
Those concerns haven’t slowed Meta’s spending. As of the end of the first quarter, the company had committed $182.9 billion to AI infrastructure over the coming years, including massive projects in Louisiana and Ohio. The Ohio facility, which CEO Mark Zuckerberg once said would be the size of Manhattan, is expected to come online this year.
Unlike Google or OpenAI, Meta has not yet generated significant standalone revenue from its own AI models or services. The company does not break out earnings for Meta AI or its open-weight Llama model family. Executives have instead emphasized internal corporate uses of AI in public statements, suggesting the technology hasn’t yet become a material revenue line on its own.
To recoup some of its colossal investment, Meta may adopt a model similar to CoreWeave’s, selling access to “raw” compute capacity, according to Bloomberg. The report also indicates Meta is considering following AWS’s lead by offering access to various AI models hosted on its infrastructure, including its recently launched closed-weight model, Muse Spark.
This new business line is reportedly part of an initiative called Meta Compute, led by infrastructure head Santosh Janardhan, Meta Superintelligence Labs leader Daniel Gross, and president Dina Powell McCormick.
The report aligns with Zuckerberg’s comments from May, when he said a cloud computing business is “definitely on the table” as a way to generate returns on the massive investment in Meta’s push toward AI superintelligence.
TechCrunch has reached out to Meta for comment.
(Source: TechCrunch)
