Meta has invested billions of dollars in artificial intelligence development and the massive data center infrastructure required to support it. Now, the company appears to be exploring ways to turn that infrastructure into a more immediate source of revenue.

 

According to a Bloomberg report published Wednesday, Meta is developing plans for a cloud infrastructure business that would offer customers access to AI computing power as well as AI models. Such a move would position Meta as a competitor to major cloud providers, including Amazon Web Services (AWS), Google Cloud, and Microsoft Azure.

 

Meta's interest in monetizing excess AI compute capacity comes just weeks after SpaceX, through its AI company xAI, announced a similar strategy. Earlier this year, SpaceX signed an agreement with Anthropic to utilize the full compute capacity of its Colossus 1 data center and has since entered into additional compute leasing agreements with Google and Reflection AI. The fact that Meta is pursuing a comparable approach suggests that the long-term winners in the AI race may ultimately be the companies that own and operate the underlying infrastructure rather than those offering the most advanced models or applications.

 

That assumption, however, depends on continued demand for AI compute and the sustained value of data center assets. Some industry observers have warned that the rapid expansion of AI infrastructure could create a bubble fueled by expensive hardware with relatively short depreciation cycles. Others have questioned whether AI companies can generate sufficient revenue to justify the industry's trillion-dollar investment commitments.

 

Despite these concerns, Meta continues to invest aggressively in AI infrastructure. As of the end of the first quarter, the company had committed approximately $182.9 billion toward AI-related infrastructure projects over the coming years, including major developments in Louisiana and Ohio. The Ohio facility, which CEO Mark Zuckerberg has described as being comparable in size to Manhattan, is expected to become operational this year.

 

Unlike companies such as Google and OpenAI, Meta has yet to demonstrate significant commercial demand for its AI models and services. The company does not separately report revenue generated by Meta AI or its Llama family of open-weight AI models, and executives have largely emphasized AI's internal applications in public communications. This suggests that Meta's AI initiatives may not yet represent a substantial standalone revenue business.

 

To generate returns on its enormous infrastructure investments, Meta is reportedly considering a business model similar to that of CoreWeave, offering customers access to raw AI compute capacity. Bloomberg also reported that Meta is exploring the possibility of providing hosted access to a variety of AI models, including its recently introduced closed-weight model, Muse Spark, through its own infrastructure platform.

 

The initiative is reportedly being developed under a new business unit called Meta Compute, led by Head of Infrastructure Santosh Janardhan, Meta Superintelligence Labs leader Daniel Gross, and President Dina Powell McCormick.

 

The report aligns with Zuckerberg's comments in May, when he stated that launching a cloud computing business was "definitely on the table" as Meta seeks to generate returns from its massive investments aimed at developing AI "superintelligence."