News
NVIDIA said partners across Europe, the Nordics, Africa and Central Asia are expanding AI cloud capacity, with new work on sovereign, open and agentic AI systems.
NVIDIA said on 21 October 2026 that its cloud partners across Europe, the Nordics, Africa and Central Asia are expanding AI infrastructure to support training, inference and agentic applications closer to data and users. The company framed the push as a mix of sovereign compute, open-model services and new controls for AI agents, with the stated aim of giving developers more choice over where workloads run and what software systems are allowed to access.
Evidence and context
In the company’s announcement published on 21 October 2026, NVIDIA said European AI clouds are adding regional compute capacity and full-stack infrastructure based on its DSX platform, Vera Rubin systems, BlueField DPUs and Spectrum-X networking. NVIDIA said the effort is intended to help developers build agentic and physical AI applications while keeping data and workloads under regional control. The company also said partners are expanding across multiple geographies, but inclusion in the ecosystem does not by itself establish a new country-specific deployment or facility.
NVIDIA said Nscale used the DSX platform to coordinate power allocation with workload requirements and, in tests using GB300 NVL72 systems, said DSX MaxLPS enabled 37% more GPUs and nearly 50% higher aggregate inference throughput within the same provisioned power budget. NVIDIA also said Nscale plans to deploy Vera Rubin infrastructure for frontier-model training and production workloads and to expand across five European markets.
Other named developments included Crusoe’s liquid-cooled modular AI factories for the Vera Rubin platform; Nebius’ use of NVIDIA Groq 3 LPX with Spectrum-X networking on Nebius Token Factory; Baseten’s model for code-review routing using Nemotron 3.5 Lightning; and Fireworks’ partnership with Era4 to combine model services with direct liquid-cooled capacity powered by Blackwell Ultra GPUs at Era4’s Chesterfield facility. NVIDIA also said Mistral trained Mistral Large 4 on 3,800 NVIDIA Grace Blackwell GPUs in Mistral-operated European data centers, and that Verda plans to deploy Vera Rubin while using Nemotron, Cosmos and the Dynamo open-source framework.
The announcement also highlighted an OpenShell Reference Architecture for secure agent deployment. NVIDIA said the architecture, announced at GTC Berlin, uses Vera Rubin NVL72 for model inference, Vera CPUs with OpenShell for agent harnesses and tool use, and BlueField-4 DPUs running Sentry for in-silicon enforcement. NVIDIA said partners including Aleria, Baseten, Cassava, CoreWeave, Crusoe, DigitalOcean, Firebird, Fireworks, Firmus, G42, GMI Cloud, IREN, Lambda, Lightning AI, Nebius, Nscale, Radiant, Verda and Vultr are supporting the reference architecture.
The broader backdrop is NVIDIA’s ongoing European infrastructure push. Background material supplied with the announcement notes that on 11 June 2025 NVIDIA outlined European infrastructure plans including 14,000 Blackwell GPUs through UK cloud partners Nebius and Nscale, an initial 18,000 Grace Blackwell systems for Mistral AI in France, and 10,000 Blackwell GPUs for an industrial AI cloud in Germany. Those figures were presented as plans announced on that date, not independently verified installed capacity as of 21 October 2026.
NVIDIA shares last traded at $216.47 as of 21 October 2026 at 10:37:47 UTC, according to the market data provided.
