Meta Signs a $27 Billion AI Infrastructure Deal With Nebius — The Largest External Cloud Commitment in AI History
Nebius Group surges 14% as Meta commits up to $27B over five years for dedicated GPU capacity, including one of the first large-scale NVIDIA Vera Rubin deployments starting early 2027.
A Deal That Reshapes AI Infrastructure
Meta Platforms and Nebius Group announced on March 16 what may be the largest AI infrastructure deal ever signed with an external cloud provider: a five-year agreement worth up to $27 billion. The deal immediately sent Nebius shares surging 14% in early trading, transforming the relatively under-the-radar infrastructure company into one of the most closely watched names in AI.
The agreement has two components. First, Nebius will provide $12 billion of dedicated GPU capacity across multiple data center locations, built around one of the first large-scale deployments of NVIDIA's next-generation Vera Rubin platform — the same GPU architecture Jensen Huang unveiled at GTC 2026 just hours earlier. Second, Meta has committed to purchasing up to $15 billion in additional compute capacity across upcoming Nebius clusters over the contract's five-year term.
Why Meta Is Going External
The deal signals a strategic shift for Meta, which has historically built and operated its own AI infrastructure. With AI compute demand growing faster than even Meta's massive capital expenditure program can address, partnering with external providers like Nebius allows the company to scale capacity without the multi-year lead times required to build new data centers from scratch.
For Meta's AI research and product teams — which are training increasingly large foundation models and deploying AI features across Facebook, Instagram, WhatsApp, and the metaverse — access to Vera Rubin-generation hardware starting in early 2027 provides a meaningful competitive advantage. The timing aligns with Meta's roadmap for next-generation AI models that will require significantly more compute than current Llama architectures.
What This Means for the AI Infrastructure Market
The Meta-Nebius deal validates a broader trend: hyperscalers are supplementing their owned infrastructure with long-term commitments to specialized AI cloud providers. Nebius, which emerged from the restructuring of Yandex's international businesses, has been quietly building GPU-dense data centers optimized specifically for AI training and inference workloads.
For investors, the deal transforms the AI infrastructure narrative. The question is no longer whether companies will spend aggressively on AI compute — it is which infrastructure providers will capture that spending. With $27 billion committed from a single customer, Nebius has answered that question decisively.
Sources: Bloomberg (March 16, 2026), CNBC (March 16, 2026), BNN Bloomberg (March 16, 2026), The Motley Fool (March 16, 2026)