With hundreds of billions earmarked for AI data centers, Meta is weighing whether to rent out excess capacity in a move that would pit it against Amazon, Microsoft, and Google.
Mark Zuckerberg’s Meta is quietly weighing whether to do something it has never done before: sell cloud computing services. The idea hinges on a simple premise. If you’re spending hundreds of billions on AI data centers, and some of that capacity sits idle, why not rent it out?
It’s the kind of move that would put Meta on a direct collision course with Amazon Web Services, Microsoft Azure, and Google Cloud, the three companies that collectively dominate the cloud infrastructure market.
The infrastructure binge behind the idea #
The company currently operates over 30 data centers. It is constructing AI-optimized facilities with capacities ranging from 1 gigawatt to 5 gigawatts.
In January 2026, Zuckerberg introduced what he called the “Meta Compute” initiative, an effort to accumulate “tens of gigawatts this decade.” The long-term ambition stretches even further, with talk of “hundreds of gigawatts or more.”
The capital expenditures to support all of this could reach into the hundreds of billions of dollars. One project alone, a Louisiana site, carries a reported $27 billion price tag. Meta has also secured over $10 billion in partnerships with Google Cloud to supplement its own capacity while it builds.
The Prometheus supercluster is planned for 2026. An even more ambitious project called Hyperion is projected to have a physical footprint comparable to Manhattan.
Meta is also working with external partners like CoreWeave and Nebius to bridge the gap between what it needs now and what its owned infrastructure can deliver once construction wraps up.
Why cloud makes strategic sense, and why it’s still a maybe #
As of now, Meta has not formally announced plans to commercialize its data center capacity. The company’s current strategy is focused on building self-sufficiency for its own AI workloads, not on becoming a platform for other companies’ workloads.
Running a cloud business is also fundamentally different from running data centers for internal use. It requires enterprise sales teams, service-level agreements, developer tools, billing systems, and a support apparatus that Meta has never had to build.
What this means for investors #
If Meta is simultaneously a customer and a competitor of Google Cloud, that relationship gets awkward fast. The $10 billion Google Cloud partnership could look very different if Meta starts competing for the same enterprise clients. The signal to watch is whether Meta begins hiring enterprise cloud sales talent or announces developer-facing tools for external compute access. Those would be the clearest indicators that this has moved from boardroom discussion to actual execution.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our