Meta has signed a long-term agreement to spend up to $27 billion on AI infrastructure from Dutch cloud provider Nebius Group, the companies announced Monday — marking one of the largest single compute-procurement contracts in tech history.

The Deal

The five-year arrangement has two parts: $12 billion in dedicated, high-density AI cluster capacity, and up to $15 billion in additional elastic compute. The dedicated capacity will include an early large-scale deployment of NVIDIA's next-generation Vera Rubin chips, positioning Nebius as a frontline partner in the post-Blackwell AI era.

Nebius shares surged 14% on the news, adding to a run that has already seen the stock climb 35% in 2026 alone.

Why It Matters

Meta is competing aggressively on the frontier model frontier, and this deal signals it won't rely solely on hyperscaler infrastructure. By locking in multi-year compute from a specialized AI neocloud — one backed by NVIDIA's own $2 billion investment last week — Meta is diversifying away from commodity cloud pricing.

Nebius already holds a $19.4 billion compute deal with Microsoft signed in September 2025. The Meta partnership makes it arguably the most strategically positioned independent AI cloud provider in Europe.

The Bigger Picture

The Meta-Nebius contract is part of a broader wave of infrastructure commitment. Hyperscalers including Amazon, Alphabet, and Microsoft are together expected to deploy around $700 billion in AI infrastructure this year. Meta alone has guided $115–$135 billion in AI capex for 2026.

For Nebius — originally carved out of the Russian internet giant Yandex in 2022 — the deal validates its aggressive push into GPU cloud infrastructure since listing in New York in 2024.