Meta has signed a five-year AI infrastructure agreement worth up to $27 billion with Dutch AI cloud company Nebius, expanding an existing partnership between the two companies. The agreement secures large volumes of compute capacity needed to train and operate advanced artificial intelligence systems.
The contract combines $12 billion in dedicated infrastructure capacity with up to $15 billion in additional compute purchases across several Nebius clusters.
Following the announcement, Nebius shares rose about 14% in early trading, reflecting investor interest in companies building AI computing infrastructure.
Delivery is expected to begin in early 2027, supported by one of the first large deployments of NVIDIA’s upcoming Vera Rubin AI platform, a new generation of hardware designed for demanding AI workloads.
Why It Matters: Developing and operating modern AI systems depends on access to specialized data centers filled with GPU clusters. Long-term infrastructure agreements help ensure companies can obtain the compute resources needed to train large models and run AI services. They also give infrastructure providers clearer demand forecasts when planning new data centers and hardware deployments.
- Meta Secures A Large Supply Of AI Compute Through A Multi-Year Agreement: The agreement reserves $12 billion in dedicated capacity for Meta across several Nebius locations, ensuring guaranteed access to AI infrastructure. Meta can also purchase up to $15 billion in additional compute from upcoming Nebius clusters over the next five years. Nebius plans to sell available capacity from those clusters to other customers first, with remaining capacity available to Meta under the agreement. This approach allows Nebius to support other AI cloud customers while maintaining a large committed buyer.
- New Infrastructure Will Deploy NVIDIA’s Next Generation AI Hardware: The clusters supporting the agreement will run on NVIDIA’s Vera Rubin architecture, which succeeds current AI accelerator systems used for training large models. These chips are designed to support higher performance for demanding machine learning workloads. Nebius plans to begin delivering capacity built on this platform in early 2027, placing the deployment among the first large implementations of the architecture.
- Nebius Is Building An AI Cloud Platform Focused On The Full Model Lifecycle: Nebius provides infrastructure designed for developers and companies building AI systems. Its platform supports the full process of AI development, including data management and model training. The company operates a growing network of AI data centers and provides cloud services tailored to GPU-intensive workloads.
- Major Technology Partnerships Have Expanded Nebius’ Customer Base: The Meta agreement adds to several large infrastructure partnerships the company has announced. In September 2024, Nebius signed a five-year agreement with Microsoft worth up to $19.4 billion to provide computing resources. The company also received a $2 billion investment from NVIDIA, highlighting interest from companies involved in AI hardware and infrastructure.
- Large Technology Companies Continue To Increase Spending On AI Infrastructure: Meta expects AI-related capital expenditures between $115 billion and $135 billion this year as it builds the data centers and computing clusters required to support AI systems. Companies such as Amazon, Microsoft, and Alphabet are making similar investments in GPU infrastructure and data centers to support the growing demand for AI computing.
Nebius jumps 14% after inking $27 billion infrastructure deal with Meta – CNBC
Trusted insights for technology leaders
Our readers are CIOs, CTOs, and senior IT executives who rely on The National CIO Review for smart, curated takes on the trends shaping the enterprise, from GenAI to cybersecurity and beyond.
Subscribe to our 4x a week newsletter to keep up with the insights that matter.
