Meta Builds $182B Compute Cloud to Rival AWS and Google

Meta is launching a service to sell compute capacity to external customers — the project is known as Meta Compute. The company has invested $182.9B in AI infrastructure and is building a data center in Ohio the size of Manhattan, set to open in 2026. Customers will have access to both raw GPU compute and hosted AI models, including the proprietary Muse Spark.

Meta is entering a market dominated by AWS, Azure, and Google Cloud, backed by one of the largest GPU fleets on the planet. The same playbook worked for SpaceX in May 2026: excess capacity turned into a profitable business — and this precedent could reshape pricing across the entire enterprise AI cloud market.