Meta is building a cloud business to sell AI compute, putting pressure on AWS, Azure, and Google Cloud

Meta is forming a dedicated cloud infrastructure business to sell excess AI computing power and hosted AI models to enterprise customers, according to Bloomberg, which first reported the development on July 1. The move positions the social media giant as a direct competitor to Amazon Web Services, Microsoft Azure, and Google Cloud, three incumbents that have collectively defined the enterprise cloud market for years.

CNBC confirmed the plan the following day, reporting that Meta will sell outside customers access to its compute infrastructure, with the company still weighing whether to lead with raw computing capacity or with AI models running on its own hardware. What is no longer in question, according to both outlets, is that a cloud business is coming.

A $145 billion bet that now needs a second revenue model

The context matters for enterprise operators evaluating where to place infrastructure spend. In April, Meta raised the high end of its 2026 capital expenditure guidance by $10 billion to $145 billion, according to CNBC, and simultaneously issued $25 billion in bonds to help fund that buildout. That level of infrastructure investment has long prompted questions from the analyst community about whether Meta could generate returns beyond its advertising business.

CEO Mark Zuckerberg has been signaling this direction for months. At Meta's annual shareholder meeting in May, he called a potential cloud computing business "definitely on the table," according to CNBC. On an earnings call seven months earlier, he noted that companies were routinely asking to buy compute capacity from Meta at a premium, suggesting organic demand already exists before a formal product is launched.

Paul Meeks, head of technology research at Freedom Capital Markets, told CNBC the move looks like a direct response to investor skepticism about whether Meta's capex would ever generate commensurate returns. His observation: virtually all the financial benefits of Meta's AI spending so far have flowed through its advertising segment, meaning the core infrastructure has yet to be independently monetized.

What entering cloud actually means for margins

Enterprise procurement teams should understand the structural economics at play here. Cloud infrastructure, while lucrative in absolute dollar terms for AWS, Azure, and Google Cloud, carries meaningfully lower operating margins than a pure software or advertising business. Meta's advertising segment is one of the highest-margin businesses in technology. Selling compute hours operates on a fundamentally different cost structure, and analysts cited by CNBC are already flagging that investors will need to recalibrate margin expectations as the cloud unit scales.

Karan Ramchandani, managing director at Post Oak Group, told CNBC that turning excess compute into a B2B revenue stream is, in his assessment, a straightforward strategic call. The trickier question is execution: how Meta prices its capacity relative to established hyperscalers, how it builds out the sales and support infrastructure enterprise buyers expect, and whether it leads with model access or raw compute.

A new hyperscaler on the procurement shortlist

For CIOs and procurement directors currently running vendor evaluations, Meta's entry changes the competitive dynamics of enterprise cloud sourcing, even before a product is formally available. Meta's AI infrastructure is built around its own custom silicon and is purpose-built for large-scale model training and inference. That hardware profile could be particularly attractive to enterprises running or evaluating large language model workloads, especially those already familiar with Meta's open-source Llama model family.

The three hyperscaler incumbents each built cloud businesses on top of internal infrastructure originally scaled for their own operations, a pattern Meta is now following. AWS grew out of Amazon's retail infrastructure; Azure extended Microsoft's enterprise software presence; Google Cloud leveraged search and ads infrastructure. Meta's path looks structurally similar, though it enters a more crowded and mature market.

Bloomberg reported that the business is still in formation, with key strategic decisions, including pricing model and product structure, yet to be finalized. What is clear is that Meta's infrastructure ambition has moved from a capital expenditure line item to an active go-to-market priority. Enterprise buyers watching the cloud market will want to track how quickly Meta formalizes its offering and whether it enters through a partner channel or a direct sales model, a distinction that will determine how quickly it becomes relevant for procurement teams to evaluate.