Meta has invested billions of dollars into developing artificial intelligence and constructing the extensive data centers required to support it. However, the company now appears to be preparing to redirect these significant assets towards a more immediately profitable endeavor.
On Wednesday, Bloomberg reported that Meta is formulating plans for a cloud infrastructure business, which would involve selling access to both its AI compute power and its various AI models. This strategic move would position Meta in direct competition with established cloud service providers such as Amazon Web Services, Google Cloud, and Microsoft Azure.
Meta's decision to commercialize its surplus compute capacity follows similar announcements made weeks prior by SpaceX, through its xAI venture. In early May, SpaceX finalized an agreement with Anthropic to lease the entire compute capacity of its Colossus 1 data center. SpaceX has since secured comparable leasing deals with Google and Reflection AI. Meta's parallel approach signals a potential shift in the AI landscape, suggesting that the ultimate winners in the AI race may not solely be those delivering the most advanced models and services, but rather those who possess and control the underlying data center infrastructure.
This outlook, however, is contingent on the sustained demand for compute power and the enduring value of data centers. Some skeptics have cautioned that the aggressive build-out of AI infrastructure is creating an economic bubble, heavily reliant on rapidly depreciating semiconductor chips. Others have questioned whether AI companies can generate sufficient end-user revenue to justify the colossal, multi-trillion-dollar investments currently being made.
Despite these concerns, Meta has not wavered in its substantial investments in infrastructure for AI compute. By the close of the first quarter, Meta had committed an astounding $182.9 billion towards AI infrastructure in the coming years, encompassing massive ongoing projects in Louisiana and Ohio. The Ohio project, notably described by Meta CEO Mark Zuckerberg as being "the size of Manhattan," is anticipated to become operational within the current year.
Unlike companies such as Google and OpenAI, Meta has not yet experienced significant external demand for its own AI models and services. Meta does not separately disclose revenue generated from Meta AI or from Llama, its open-weight AI model family, in its earnings reports. Public statements from executives have predominantly emphasized the internal corporate applications of AI, suggesting that Meta's current AI endeavors may not yet represent a material, standalone revenue stream.
To realize a return on its immense capital outlay, Meta is reportedly considering adopting CoreWeave's business model, selling direct access to "raw" compute capacity, according to Bloomberg. The same report indicated that Meta is also exploring a strategy akin to AWS, offering access to a range of AI models — including its recently unveiled closed-weight model, Muse Spark — all hosted on its extensive AI infrastructure.
This forthcoming business segment is reportedly part of a new initiative named Meta Compute, which will be spearheaded by Santosh Janardhan, Head of Infrastructure; Daniel Gross, leader of Meta Superintelligence Labs; and President Dina Powell McCormick.
This development aligns with Mark Zuckerberg's statements in May, where he affirmed that a Meta cloud computing business was "definitely on the table" as a means to generate a return on the substantial investments made towards its ambitious strategy of developing AI "superintelligence."
TechCrunch has reached out to Meta for an official statement regarding these reports.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.