Meta plans cloud business to monetise surplus AI computing power

Meta is developing a cloud infrastructure business to sell excess artificial intelligence computing capacity to external customers, according to reports published on Wednesday, in a move that could diversify its revenue beyond advertising while challenging established cloud providers.

Reporting from Bloomberg News said Meta is considering offering developers access to AI models hosted on its own infrastructure, including its Muse Spark models, while charging for the computing power required to run them.

The company is said to be weighing a second option that would sell raw AI computing capacity, similar to specialist providers such as CoreWeave, through an internal initiative known as Meta Compute. Meta declined to comment, and Reuters said it could not independently verify the plans.

The strategy would place Meta in competition with cloud market leaders Amazon Web Services, Microsoft Azure and Google Cloud, while potentially reducing its reliance on third-party infrastructure partners. Investors welcomed the news, with Meta shares rising more than 7 per cent in early trading, while shares in neocloud providers CoreWeave and Nebius fell between 10 and 12 per cent, according to Reuters.

Speaking at Meta’s annual shareholder meeting in May, chief executive officer Mark Zuckerberg said entering the cloud computing market was “definitely on the table”. He added: “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.” Zuckerberg said the company had not pursued the opportunity because it still expected to use the capacity internally, but selling surplus infrastructure remained an option if it overbuilt.

The plans come as technology companies continue investing heavily in AI infrastructure despite investor concerns over spending levels. Reuters reported that major technology firms are expected to spend more than $700 billion on AI infrastructure this year, compared with about $400 billion in 2025, while Meta has forecast capital expenditure of between $125 billion and $145 billion in 2026.

Gil Luria, managing director at D.A. Davidson, told Reuters that the proposal could have the greatest impact on specialist cloud providers rather than established hyperscalers. “The impact of adding Meta’s capacity to the market is more likely to be on neoclouds than the big hyperscalers,” he said, adding that companies such as CoreWeave and Nebius “rely on Meta for their growth and Meta may not need them anymore”.

The reported expansion comes as demand for AI computing power continues to outstrip supply across the industry. The Financial Times reported in June that Google had limited Meta’s access to its Gemini AI models after the company requested more computing capacity than Google could provide, disrupting some internal projects.

According to the report, the shortage has prompted Google to secure additional capacity through a $920 million-a-month agreement with SpaceX, highlighting the intense competition for AI infrastructure.