Meta plans to rent AI computing power as it takes on AWS, Google Cloud: Report

Mark Zuckerberg-led Meta is working on a new cloud infrastructure business in order to sell access to its AI computing power and models and begin competing with the likes of Amazon Web Services, Microsoft Azure and Google Cloud, as per a new report by Bloomberg.

Reportedly, Meta has spent billions of dollars developing its AI models and building the data centre infrastructure required to run them. However, with the demand for AI computing rising, Meta is said to be looking to monetise this infrastructure.

Among the potential plans is reportedly selling access to various AI models hosted on Meta’s AI infrastructure, essentially mirroring the approach adopted by AWS’s Bedrock. In this system, Meta would run the data centres and chips that power the models, including its own ones like Muse Spark, while charging developers to access them.

Bloomberg also reports that Meta is considering renting out raw AI computing capacity, similar to other neocloud businesses like CoreWeave. The new initiative is said to be a part of an internal division inside the company called Meta Compute, which looks after the company’s AI infrastructure efforts.

The department is led by Santosh Janardhan, Meta’s head of infrastructure, Daniel Gross, a leader inside Meta Superintelligence Labs, and Meta President Powell McCormick.

Why is Meta entering cloud computing?

Meta has been working towards building AI ‘superintelligence’ and has committed hundreds of billions of dollars towards AI infrastructure, including data centres, Nvidia chips and computing agreements with companies such as CoreWeave, Google and Oracle.

However, the report notes that investors have increasingly questioned how the company plans to generate returns from this massive spending spree.

A cloud business could provide one answer. By renting excess AI computing capacity and offering hosted AI models, Meta could potentially create a new revenue stream beyond its advertising business.

The opportunity is also driven by surging demand for AI infrastructure. With companies racing to build and deploy AI models, cloud providers have increasingly gone beyond traditional computing services to rent specialised AI chips and high-performance computing capacity needed for training and inference.

However, operating such a business is far more complex than simply owning data centres and requires sophisticated software platforms, enterprise sales teams and customer support operations to serve developers and large businesses.

Zuckerberg had hinted at the move

The reported plans by Meta don’t come as a complete surprise. Meta CEO Mark Zuckerberg had 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.