Meta enters cloud market by selling AI computing power

In a significant move, Meta is developing plans to enter the cloud computing market by selling access to its AI computing power to external customers. This initiative, known as Meta Compute, is being led by Santosh Janardhan, the head of infrastructure at Meta, along with Daniel Gross from the Meta Superintelligence Labs AI unit and Meta President Dina Powell McCormick. The company aims to generate revenue from its excess computing capacity, positioning itself to compete directly with established cloud service providers like Amazon Web Services, Microsoft Azure, and Google Cloud. The plans are still in development, and various options are being considered, including a service similar to AWS's Bedrock offering, where developers would pay to access AI models hosted on Meta's infrastructure, including its own Muse Spark models. Additionally, Meta is contemplating the sale of raw computing capacity, akin to neocloud providers like CoreWeave. This strategic shift comes as Meta has committed between $125 billion and $145 billion in AI-related capital expenditure for 2026, raising concerns among investors about potential returns. Mark Zuckerberg, the CEO of Meta, indicated that launching a cloud service was a possibility during the company's annual shareholder meeting in May, noting that outside companies have been approaching Meta frequently to utilize its API services or purchase excess compute capacity. He acknowledged that while Meta has not yet pursued this option, it remains viable if the company feels it has overbuilt its computing resources. The announcement of these plans has already had an impact on the market, with Meta's shares rising by 9.3% following the news, while rival neocloud player CoreWeave saw a drop of up to 14%. As Meta continues to explore this new venture, the implications for the cloud computing landscape could be significant, potentially reshaping competition among major players in the industry.