Meta stock jumps on plan to sell AI cloud services

Meta Platforms shares rose 10.6% during Wednesday trading after a report said the company plans to create a cloud infrastructure business to sell AI computing power and hosted AI models to outside customers.

The proposal would let Meta monetize excess AI compute capacity by offering access to hosted models on its systems or by selling raw GPU and other compute capacity directly to buyers. The initiative would be built around Meta’s expanding network of data centers.

The new business would be organized under Meta Compute, the company’s internal program for building and managing large-scale AI infrastructure. Meta has increased investment in data centers, AI chips and related systems and has existing computing agreements with several providers, including CoreWeave, Google and Oracle.

Investors moved on the report during Wednesday trading. Smaller AI-focused cloud providers saw shares decline; CoreWeave fell about 14% and Nebius dropped roughly 15% as markets priced in potential new competition. The plan would place Meta alongside major cloud providers that already rent computing infrastructure and host AI models, including Amazon Web Services, Microsoft Azure and Google Cloud.

Meta raised its capital expenditure outlook in April by $10 billion to a range of $125 billion to $145 billion for the year, citing higher component prices and additional data center costs. Company executives have said building large-scale compute capacity is needed to support AI development and that some capacity could be available for external customers at times.

On a shareholder call in May, CEO Mark Zuckerberg said, “It’s definitely on the table,” when asked about selling unused compute. He added that outside companies frequently ask Meta about offering API services or buying compute at a premium and that selling excess capacity would be an option if the company determines it has overbuilt.

Separately, Meta is pursuing product-level AI commercialization, including its Meta AI chatbot. The company describes its infrastructure spending as necessary to advance its AI capabilities and as a potential source of future revenue if parts of that infrastructure are opened to outside customers.