Meta is reportedly building a cloud infrastructure business to sell off its excess AI computing capacity. This move would pit the social media giant directly against Amazon Web Services, Google Cloud, and Microsoft Azure. According to a Bloomberg report, the plans are still in the early stages and could evolve before any formal launch, but they mark a notable shift for a company that has spent two years aggressively acquiring AI compute rather than selling it.
Meta is reportedly weighing two distinct approaches. The first involves selling developers access to AI models hosted on its own infrastructure, including its closed-weight Muse Spark model, in a setup similar to how Amazon's Bedrock service operates.
The second option would see Meta sell raw computing capacity outright, a model that ‘neocloud’ providers such as CoreWeave have built their entire businesses around. Either path would place Meta in direct competition with the three companies that currently dominate the cloud computing market.
The initiative is reportedly organized under a new unit called Meta Compute, led by the company's head of infrastructure, Santosh Janardhan, alongside Meta Superintelligence Labs' Daniel Gross and company president Dina Powell McCormick.
The idea isn't entirely new. At Meta's shareholder meeting in May, CEO Mark Zuckerberg said entering cloud computing was ‘definitely on the table,’ noting that companies were approaching Meta almost weekly to buy access to its AI models or spare compute capacity.
The logic ties back to Meta's enormous infrastructure spending. The company has guided full-year 2026 capital expenditure toward $125 billion to $145 billion, driven largely by data centers, chips, and land acquisition to support its AI ambitions. With Meta not breaking out separate revenue figures for its AI or Llama model business, investors have grown increasingly vocal about wanting to see returns on that spending.
Turning idle compute capacity into a paid service offers one visible way to convert that outlay into revenue rather than treating it purely as a sunk cost. Meta CEO Mark Zuckerberg hinted at the move during a shareholders' meeting in May, where he said that the company was already receiving frequent requests from businesses looking to buy access to its AI models or spare computing capacity.
“It’s definitely on the table.… Almost every week there are different companies that come to us from the outside asking us to both stand up an API service or asking if we have compute that they could buy from us at some premium to what we’ve bought it at.”
“We haven’t done that yet because we think we have a use for the compute… But obviously if we get to a point where we feel that we have overbuilt, then that is an option that we have, and that is partially what gives us confidence in investing in building this out,” he added.
Also Read: Meta Brings Token Billing to WhatsApp AI Agent, Restores Messaging Fees
Investors responded positively to the report, with Meta's stock rising more than 1% following the news, while shares of AI infrastructure specialists such as CoreWeave, Nebius, and IREN slipped on concerns about a larger, better-funded competitor entering the compute rental market.
The development follows a similar move by SpaceX, which, through xAI, has been selling excess compute capacity to companies including Anthropic, Google, and Reflection AI in recent months.