Meta is considering building a cloud-computing business that would sell unused AI computing capacity from its internal infrastructure, Bloomberg reports.
The company has been investing heavily in large-scale data centers to support its AI products and research. Because this capacity is so expensive and often built with future growth in mind, Meta may end up with more compute power than it immediately needs. The idea is to monetize that surplus instead of letting it sit idle.
Under this potential model, Meta could rent out raw AI compute, similar to how modern cloud providers sell access for training and running AI models.
It could also evolve into offering more managed services, such as hosted AI models or APIs that developers and companies could build on. That would put it in partial competition with major cloud providers like Amazon Web Services, Microsoft Azure and Google Cloud.
The motivation is largely financial and strategic. AI infrastructure costs have surged across the industry, and Meta’s spending has grown rapidly as it scales up its AI ambitions across products like social platforms, advertising systems and virtual assistants. Turning excess capacity into revenue would help offset those investments and improve utilization.
This potential move also reflects a broader trend. AI companies increasingly want to control both the models and the infrastructure beneath them. If Meta proceeds, it would shift from being only a consumer of cloud-style AI infrastructure to also becoming a seller of it.
At the same time, the plan is still early and not guaranteed. It would require building customer-facing cloud services, pricing models, and support systems that are quite different from Meta’s current internal infrastructure focus.
