NVIDIA GTC Berlin Launches Vera Rubin Factories, Six-Domain Open Models

At GTC Berlin 2026, NVIDIA CEO Jensen Huang unveiled a wave of product launches and partnerships. The DSX POD reference design ships Vera Rubin AI factories inside pre-integrated containers, cutting on-site deployment from six months to 14 days once site utilities are ready. NVIDIA presented an open frontier model portfolio spanning six domains: Nemotron for language, Cosmos for world simulation, GR00T for humanoid robotics, Alpamayo for autonomous vehicles, Ising for quantum-classical computing, and Earth-2 for physics AI. The Agent Knowledge Foundation adds Kumo models for structured-data prediction alongside Nemotron tools for unstructured data, enabling agents to query enterprise data without moving it or rebuilding existing workflows. Automotive partnerships expanded with Volvo Cars, Volkswagen, JLR, and Mercedes-Benz committing their next-generation vehicles to the Hyperion platform. NVIDIA reported completing its first commercial phone call on T-Mobile's network with Nokia after nearly eight years of work applying GPU-accelerated computing to telecom. cuDSS launched as generally available for engineering simulation, with Applied Materials reporting up to 10x speedup and Dassault Systèmes reporting 8-10x speedup in Abaqus electromagnetic solvers. Confidential Bare Metal extends NVIDIA's confidential computing to Grace Blackwell NVL72, HGX Rubin NVL8, and DGX Rubin NVL8 systems. European cloud providers Nebius, Nscale, and Verda are building infrastructure on Vera CPUs. Mistral and NVIDIA announced a collaboration on industrial engineering AI covering faster simulation, open physics AI benchmarks, and agentic AI evaluations for CAE and EDA.