Overview

• Meta confirmed its Prometheus supercluster will deliver 1 GW of computing power and is set to be operational in 2026.
• The company plans to scale its Hyperion project to 5 GW over the coming years as part of a broader multi-gigawatt AI infrastructure push.
• Meta has increased its capital expenditure forecast for 2025 to between $64 billion and $72 billion to fund the build-out of its mega-clusters.
• Through its Superintelligence Labs, Meta has recruited top AI talent including Scale AI’s former CEO Alexandr Wang and leading researchers from OpenAI.
• Meta is positioning its expansion against OpenAI’s Stargate and Alphabet’s South Carolina centers, raising scrutiny over the environmental and public health impacts of large-scale AI facilities.