NVIDIA Cloud Partners Scale Across Europe — Sovereign, Open and Built for Safe AI Agents

Europe’s AI cloud crowd is moving fast because demand is forcing the issue. Developers want compute near their data, open models they can adapt, and infrastructure that’s secure by design. NVIDIA says partners across Europe, the Nordics, Africa and Central Asia are expanding capacity for training and inference, with a clear pitch: keep workloads closer to home and make the stack easier to control.

The sovereignty angle is doing a lot of the work here. Nscale is using NVIDIA’s DSX Platform to match power allocation to workload needs, and in tests on GB300 NVL72 systems it says DSX MaxLPS delivered 37% more GPUs and nearly 50% higher aggregate inference throughput inside the same power budget. Nscale also plans to deploy Vera Rubin infrastructure for frontier-model training and production workloads, with expansion across five European markets. Crusoe is taking a different route with Spark, its liquid-cooled modular AI factories for the Vera Rubin platform, to push low-latency inference closer to users.

There’s a broader pattern underneath that. Regional AI clouds are being built around NVIDIA DSX, Vera Rubin, BlueField and Spectrum-X Ethernet, so customers can decide where data and workloads live instead of just renting whatever happens to be nearby. Aleria, Cassava Technologies, Firebird, G42, Radiant, SWI Compute and Vultr are all part of that push, spreading infrastructure across Europe, the Nordics, Africa and Central Asia.

The open-model side is getting just as much attention. Nebius has taken on NVIDIA Groq 3 LPX for production inference on Nebius Token Factory, tied together with Spectrum-X and Vera Rubin NVL72. It is also pushing robotics and physical AI with its Physical AI Workbench, while Baseten says CodeRabbit used Nemotron 3.5 Lightning to improve routing agreement from 75.8% to 80.7% and cut estimated inference costs by 50% for the evaluated workload.

Then comes the part that matters once agents start touching real systems. NVIDIA’s OpenShell Reference Architecture, announced at GTC Berlin, is meant to put clear boundaries around what agents can access and do, using OpenShell secure runtime software plus Vera Rubin NVL72, Vera CPUs, BlueField-4 DPUs and Sentry. 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 lining up behind it. The message is simple: Europe wants more AI, but it wants the reins too.