Nebius to provide $12 billion in dedicated AI compute capacity to Meta across multiple locations

Meta has secured a substantial $12 billion, five-year dedicated AI compute capacity from Nebius, with delivery of one of the first large-scale Nvidia Vera Rubin deployments commencing in early 2027 across multiple locations. This agreement provides Meta with early access to next-generation AI hardware, critical for its technology roadmap focused on AI. The deal also includes an innovative commitment for Meta to purchase up to an additional $15 billion in compute capacity over five years from future Nebius clusters. This structure allows Amsterdam-based Nebius to leverage Meta's commitment to attract third-party customers for a portion of this additional capacity, while Meta secures the remainder, effectively de-risking Nebius's significant infrastructure expansion plans. The commercial logic centers on Meta locking in substantial, advanced compute resources well in advance, and Nebius gaining a foundational anchor client for its build-out. For Meta, a technology company committed to building the future of human connection through AI, this potential $27 billion spend ensures foundational infrastructure for its ambitious initiatives. Nebius, an AI cloud platform provider spun off from Yandex in 2022, operating its own data center infrastructure with 200-500 employees, gains immense validation and scale from this agreement. The deal provides Nebius with a foundational revenue stream to significantly expand its global GPU cloud and data platform services. This commitment from Meta accelerates Nebius's ability to deploy its own data center infrastructure and compete in the rapidly expanding AI compute market, particularly with the early 2027 delivery of Nvidia Vera Rubin deployments.