Meta shares surge after report says it is planning an AI cloud business

Meta shares jumped Wednesday after a report said the company is developing plans for a new cloud-infrastructure business focused on artificial intelligence. According to the report, the initiative would be built around Meta’s existing AI and data center capacity, and could allow external customers to buy access to both hosted AI models and the computing power required to run them.

The approach described in the reporting resembles major cloud services that offer model hosting and managed access. One potential line involves selling developer access to AI models running on Meta’s infrastructure, in a setup compared to AWS’s Bedrock program, which provides a managed way for developers to use foundation models.

A second option mentioned is selling more direct computing capacity, closer to so-called “neocloud” providers that rent GPU capacity to customers. In that scenario, Meta would supply raw capacity rather than only model endpoints, turning part of its internal buildout into a separately sold service.

The reporting also places the effort inside Meta Compute, an internal initiative that is described as being led by Santosh Janardhan, Daniel Gross, and Meta President Dina Powell McCormick. The companies involved did not provide additional public details in the report, and the plan is described as still in development, with key specifics not yet disclosed.

Market reaction was immediate. Meta stock rose about 9.3% to $615.55 in New York, marking its biggest intraday gain since April, according to the coverage. The same report said rival AI infrastructure stocks were hit at the same time, with CoreWeave shares falling as much as 14% and Nebius Group shares down as much as 17%.

Cloud services have become a critical battleground as AI demand strains hardware supply and pushes enterprise customers to look for ways to access accelerators and models without building everything themselves. If Meta can monetize excess capacity, it would convert significant capital spending on data centers and AI infrastructure into a potentially recurring revenue stream.

Meta has not publicly confirmed the business details described in the report, and it is not clear from the available information when such a service could launch, what pricing would be, or how broadly it would be offered. The reporting also does not specify whether the company would prioritize particular model families, customer types, or regulatory constraints.

What to watch next is whether Meta acknowledges the initiative on an official channel, provides timing and scope, or addresses how it plans to manage capacity between internal AI projects and external customers. Investors will likely look for follow-through in earnings calls, product updates, or infrastructure announcements, alongside any changes to competitive positioning versus AWS, Microsoft Azure, Google Cloud, and specialist “neocloud” providers.