Meta Plans to Monetize AI Compute Capacity with New Cloud Service

Meta is reportedly developing a cloud infrastructure business aimed at selling access to its AI compute resources and models, according to a Bloomberg report released Wednesday. This new venture would position Meta alongside established cloud providers such as Amazon Web Services, Google Cloud, and Microsoft Azure.

The company’s move follows similar announcements from SpaceX’s xAI, which in early May signed a deal with Anthropic to lease all compute capacity at SpaceX’s Colossus 1 data center. SpaceX has also established comparable agreements with Google and Reflection AI. Meta’s entrance into this market signals that control over data center resources may become as crucial as offering advanced AI models and services.

However, the sustainability of this strategy depends on continued demand for compute power and the enduring value of data centers. Critics have raised concerns about a potential bubble in AI infrastructure investments, warning that the rapid depreciation of chips could undermine long-term returns. Additionally, questions remain about whether AI companies can generate sufficient revenue from end users to justify their substantial expenditures.

Despite these concerns, Meta has committed to extensive investments in AI infrastructure. By the end of the first quarter, the company planned to spend $182.9 billion on AI infrastructure projects, including large-scale developments in Louisiana and Ohio. The Ohio facility, described by Mark Zuckerberg as comparable in size to Manhattan, is expected to become operational this year.

Unlike Google and OpenAI, Meta has not yet experienced significant demand for its AI models and services. The company does not separately report revenue from Meta AI or its open-weight AI model family, Llama, in its earnings statements. Public communications from executives have primarily focused on AI’s internal applications within Meta, suggesting these initiatives have not yet formed a substantial revenue stream.

To monetize its substantial infrastructure investment, Meta may adopt a business model similar to CoreWeave’s by selling access to raw compute capacity. Bloomberg also reported that Meta is considering offering access to various AI models, including its newly introduced closed-weight model, Muse Spark, hosted on its AI infrastructure.

This new cloud service will be part of an initiative reportedly named Meta Compute, led by head of infrastructure Santosh Janardhan, Meta Superintelligence Labs leader Daniel Gross, and president Dina Powell McCormick.

The Bloomberg report aligns with Zuckerberg’s May comments that establishing a Meta cloud computing business is “definitely on the table” as a means to recoup investments in developing AI “superintelligence.”

Requests for comment have been made to Meta by TechCrunch.