Nebius Group has signed a $27 billion agreement to provide AI capacity to Meta Platforms. The deal, announced Tuesday, is one of the largest single contracts for AI infrastructure and underscores the escalating demand for scalable compute resources across the technology industry.
The agreement gives Meta access to a substantial block of Nebius's AI capacity, though the companies did not disclose the specific types or locations of the hardware involved. The contract's value — $27 billion — places it among the biggest AI infrastructure commitments made public to date.
Why Meta Needs the Capacity
Meta has been expanding its AI operations, from recommendation systems to generative models, and that expansion requires enormous amounts of compute. Buying capacity from Nebius allows Meta to add resources without waiting to build and equip its own data centers. The arrangement also gives Meta flexibility to scale up or down as its AI products evolve.
Nebius, which was spun out of Yandex in 2024, has positioned itself as a provider of AI cloud services. The company operates data centers designed for intensive machine learning workloads and has been courting large technology customers. The Meta contract is a significant validation of that strategy.
What Nebius Gets Out of It
For Nebius, the deal provides long-term revenue visibility. A $27 billion commitment from a company of Meta's size reduces the risk that comes with building out expensive infrastructure on speculation. It also signals to other potential customers that Nebius can handle contracts at this scale.
The agreement comes as demand for AI compute continues to outstrip supply. Companies across the tech sector are competing for access to the latest chips and the data center space to house them. Deals like this one show that the largest players are willing to commit billions of dollars years in advance to secure their positions.
The Broader AI Infrastructure Race
The Nebius-Meta agreement reflects a broader pattern: AI developers are locking in capacity through long-term contracts rather than relying solely on their own buildouts. This approach shifts some of the capital burden to infrastructure providers and gives AI companies more predictable access to the resources they need.
It also highlights the strategic importance of scalable compute resources. As AI models grow larger and more complex, the ability to secure enough processing power becomes a competitive advantage. Companies that can't get it may fall behind.
Neither Nebius nor Meta has said when the capacity will come online or how it will be distributed across regions. The companies also haven't specified which hardware platforms will be used, a detail that matters for performance and cost.
For now, the $27 billion figure stands as a marker of how much the industry is willing to spend on AI infrastructure. The deal is expected to close in stages, with capacity becoming available over the coming years. More details may emerge when either company reports earnings or when the first tranches of capacity are delivered.
