Meta Launches Meta Compute to Build Gigawatt-Scale AI Infrastructure
Meta is investing $600 billion in AI infrastructure and nuclear energy while cutting Reality Labs staff to pivot toward AI wearables and superintelligence.
Meta Platforms Inc. has launched Meta Compute, a top-level initiative to build gigawatt-scale data centers and secure the massive energy required for artificial intelligence. CEO Mark Zuckerberg announced plans to deploy tens of gigawatts of capacity this decade, with a long-term goal of hundreds of gigawatts to achieve personal superintelligence for billions of users. The company projects a total investment of $600 billion in AI infrastructure and jobs by 2028, including $72 billion in capital expenditures for 2025.
To power this expansion, Meta secured agreements with nuclear energy providers Vistra, TerraPower, and Oklo to provide up to 6.6 gigawatts of clean power by 2035. These deals include a 20-year agreement with Vistra for 2,600 megawatts from plants in Pennsylvania and Ohio, as well as new reactor projects in Pike County, Ohio. The energy will support the Prometheus AI data center in New Albany, Ohio, and other facilities like the Hyperion campus in Louisiana.
This strategic pivot coincides with a reallocation of resources away from the metaverse. Meta is laying off approximately 10% of its Reality Labs workforce—roughly 1,500 employees—and shutting down three VR game studios after recording losses of up to $70 billion since 2020. CTO Andrew Bosworth stated the company is shifting resources almost exclusively to mobile and AI wearables.
To manage this growth, Meta appointed Dina Powell McCormick as President and Vice Chairman to oversee government and sovereign partnerships for financing and deployment. The technical effort is led by Santosh Janardhan, who manages architecture and operations, and Daniel Gross, who oversees long-term capacity strategy and supply chain partnerships.
