Meta has invested billions in artificial intelligence infrastructure but hasn't yet generated significant direct revenue from its AI products, so the company is now planning to rent out its excess computing capacity to outside businesses. According to a recent Bloomberg report, Meta is developing a cloud infrastructure service under a new division called Meta Compute, which would allow customers to access raw GPU (graphics processing unit) computing power and Meta's AI models, including its recently introduced closed-weight model, Muse Spark.

This strategic shift comes as Meta faces pressure to justify its enormous capital expenditures on AI infrastructure. By the end of the first quarter of 2026, the company had already booked around $182.9 billion worth of spending on AI infrastructure projects over the next few years, with major data center projects planned for Louisiana and Ohio. The move also mirrors similar strategies being pursued by competitors; SpaceX's AI company xAI reached an agreement with Anthropic in May to utilize the full computing capacity of its Colossus 1 data center.

Why Is Meta Suddenly Focused on Selling Computing Power?

Meta's AI products have not yet generated the same commercial impact as competitors like Google or OpenAI. The company has not shared any revenue statistics for Meta AI or its open-weight Llama family of models, and executives have focused primarily on how AI impacts Meta's own products and services rather than treating AI as a standalone business. This lack of direct revenue from AI services has prompted the company to explore alternative ways to monetize its massive infrastructure investments.

CEO Mark Zuckerberg acknowledged in May that creating a cloud computing business was "definitely on the table," suggesting that renting out Meta's computing infrastructure could help generate returns on the company's long-term vision of building AI superintelligence. The initiative is being developed under the leadership of Meta's head of infrastructure, Santosh Janardhan, alongside Meta Superintelligence Labs leader Daniel Gross and company president Dina Powell McCormick.

Mark Zuckerberg

How Will Meta's Cloud Computing Service Compare to Existing Competitors?

Meta is reportedly considering two different business models for its cloud infrastructure offering. The first would be similar to CoreWeave, providing customers with access to raw GPU compute capacity. The second would function like Amazon Web Services (AWS), offering a marketplace where businesses could access multiple AI models running on Meta's infrastructure.

If Meta moves forward with these plans, it would directly compete with major cloud providers including:

• Amazon Web Services (AWS): The dominant cloud infrastructure provider that offers computing resources and AI services to enterprises worldwide
• Google Cloud: Google's cloud platform that provides computing power and access to Google's AI models and services
• Microsoft Azure: Microsoft's cloud infrastructure service that integrates with OpenAI's models and other AI tools

Industry analysts believe that companies controlling powerful data centers may be in the strongest competitive position as AI demand continues to grow. However, not all experts are convinced that the current pace of AI infrastructure spending is sustainable. Some argue that the rush to build AI data centers could be creating an infrastructure bubble, pointing to the enormous costs of AI chips, which lose value over time, and questioning whether AI companies will generate enough revenue from customers to justify investments totaling hundreds of billions or even trillions of dollars.

Steps Meta Is Taking to Launch Its Cloud Computing Business

• Organizational Structure: Meta has created a new division called Meta Compute to oversee the cloud infrastructure initiative, with leadership from the company's top infrastructure and AI executives
• Service Models: The company is evaluating both a raw GPU compute offering similar to CoreWeave and a managed marketplace model similar to AWS, giving customers flexibility in how they access computing resources
• Model Access: Meta plans to offer customers access to multiple AI models running on its infrastructure, including both its open-weight Llama models and proprietary models like Muse Spark
• Capital Deployment: The company continues to increase its AI spending despite concerns about sustainability, with major data center projects underway in multiple US states

Meta's pivot toward monetizing its computing infrastructure represents a significant shift in how the company views its massive AI investments. Rather than relying solely on AI to improve its core products like Facebook, Instagram, and WhatsApp, Meta is now positioning itself as an infrastructure provider for the broader AI industry. This approach could help the company recover some of its enormous capital expenditures while also establishing Meta as a key player in the competitive cloud computing market.

The success of Meta's cloud computing venture will likely depend on its ability to offer competitive pricing and reliable service compared to established players like AWS and Google Cloud. It will also require Meta to demonstrate that its infrastructure can deliver the performance and reliability that enterprise customers demand. As AI infrastructure becomes increasingly central to the technology industry, Meta's ability to monetize its computing capacity could become just as important to the company's financial performance as its advertising business.