Mark Zuckerberg’s $135 Billion AI Gamble Takes a Shocking Turn

According to a new Bloomberg report, Meta is preparing to enter the cloud computing market, aiming to monetize its massive investments in AI by renting out excess computing power to outside businesses.

The under-the-radar initiative is reportedly organized under an internal group called Meta Compute and sets up a major new front of competition with the industry’s dominant players: Amazon Web Services, Microsoft Azure, and Google Cloud.

For the past two years, Meta has been on a historic buying spree, securing every available AI chip, piece of land, and power grid contract it could find to fuel its pursuit of “superintelligence.” The company has guided its capital expenditures for 2026 to a staggering $115 billion to $135 billion.

While this massive outlay is primarily intended to power Meta’s own internal products and upcoming AI models, investors have grown increasingly vocal about when and how these investments will generate actual returns. Launching a commercial cloud service allows Meta to transform idle data center capacity into an active revenue engine rather than a sunk cost.

CEO Mark Zuckerberg directly hinted at this strategic pivot back in May during Meta’s annual shareholder meeting. At the time, Zuckerberg confirmed that selling excess compute was “definitely on the table,” revealing that outside companies were approaching Meta almost every week asking to buy capacity or access its models at a premium.

While the strategies remain in development and could evolve, the group is currently weighing two main commercial paths. The first option would allow outside developers to access and build on top of advanced AI models hosted directly on Meta’s infrastructure. The second option focuses on renting out raw computing power directly to clients.

Earlier in 2026, Google had to restrict Meta’s access to its Gemini models due to an inability to meet Meta’s massive data demands. The sudden shortfall delayed several internal AI projects and even forced Meta management to ask its own employees to ration their automated AI token usage.