United States-based Meta Platforms is developing an internal initiative called "Meta Compute" to sell access to its artificial intelligence computing power and models, according to a Bloomberg report published this week. The plan would place Meta in direct competition with major cloud providers Amazon Web Services, Microsoft Azure and Google Cloud by turning its large-scale AI infrastructure into a commercial cloud service. Under one proposed model, Meta would host a range of AI models, including its flagship "Muse Spark" systems, in its own data centers and run them on advanced chips, then charge external developers and companies for access to those models. United States-based Meta is also exploring a second model that would sell raw or bare-metal computing capacity, offering high-performance GPU clusters in a manner similar to specialized neocloud providers such as CoreWeave. The development of Meta Compute is led by Santosh Janardhan, the company’s head of infrastructure, and marks a notable shift from Meta’s longstanding focus on digital advertising tied to Facebook, Instagram, Messenger and WhatsApp. By directly monetizing its hardware infrastructure, Meta aims to create a new revenue stream that leverages the same AI capabilities it already uses to improve ad targeting and user engagement across its social media platforms.

Prepared by Christopher Adams and reviewed by editorial team.