Meta and Nebius Group strike long-term AI infrastructure pact
Nebius Group, an Amsterdam-based AI cloud provider, has signed a five-year AI infrastructure agreement with Meta that could reach $27 billion in total value as demand for large-scale computing accelerates across the industry.
Under the deal, Nebius will deliver $12 billion worth of dedicated AI computing capacity across multiple locations. The company said capacity is expected to begin coming online in early 2027.
Early deployment of NVIDIA’s Vera Rubin platform
The buildout is set to use one of the first large-scale deployments of NVIDIA’s Vera Rubin platform, a next-generation AI computing architecture aimed at supporting the training and deployment of advanced models. The choice underscores how hyperscalers and AI cloud specialists are locking in next-wave hardware as part of multi-year capacity planning.
Additional compute option could add $15B
Beyond the dedicated capacity, Meta also committed to purchasing up to $15 billion of additional computing power from future Nebius AI clusters over the same five-year period. Nebius said this incremental capacity is initially intended for third-party customers on its AI cloud platform, with any remaining capacity to be purchased by Meta under the agreement.
Arkady Volozh, founder and CEO of Nebius, said the company is expanding its partnership with Meta as it secures more large, long-term capacity contracts to accelerate growth in its core AI cloud business.
Deal follows NVIDIA investment
The announcement comes days after NVIDIA disclosed a strategic partnership and a $2 billion investment in Nebius to support the development of hyperscale AI cloud infrastructure.
Nebius is building a full-stack cloud platform used by startups and enterprises for data processing, model training, and AI product deployment—use cases increasingly reliant on long-duration compute commitments as competition for capacity intensifies.
Related
