NVIDIA (NASDAQ:NVDA) outlined a broad plan to expand Europe’s artificial intelligence economy at its GTC event in Berlin, highlighting investments, AI infrastructure, agent security, physical AI systems and new computing platforms.
The company said it has operated in Europe for 33 years and has invested nearly $10 billion across startups and partnerships in the region during the past year. NVIDIA said it has more than 12,000 partnerships across Europe and recently announced investments in 14 startups during visits to France, the U.K., Sweden and Germany.
The presentation framed AI as an industrial system rather than solely a software technology. NVIDIA described a five-layer AI economy comprising energy, advanced computing, AI factories, AI models and applications. The company said Europe is building capacity across those layers through partnerships with organizations including Schneider Electric, Siemens Energy, neocloud providers and telecommunications companies.
Vera Rubin Systems and DSX Pods
NVIDIA said its Vera Rubin platform is in full production and represents the next generation of its AI computing architecture. The company said Vera Rubin combines GPUs, CPUs, networking, storage acceleration and software intended to support the full range of agentic AI workloads, including model inference, data processing and accelerated tools.
According to NVIDIA, the Vera Rubin NVLink system connects 72 GPUs into a single computing system. The company said each GPU delivers 20 petaflops of performance. NVIDIA also described its Spectrum-X networking hardware as incorporating co-packaged optics, which directly connects silicon photonics to the chip.
NVIDIA said Vera Rubin is designed to be fully liquid cooled and can be assembled in less than five minutes per node, compared with roughly two hours for a Grace Blackwell node. It also said Vera Rubin provides another 30-fold performance increase over Blackwell for large-language-model processing, while Blackwell was about 30 times faster than Hopper for that workload.
The company also introduced NVIDIA DSX Pod, a factory-integrated AI infrastructure system based on Vera Rubin NVL72 systems. NVIDIA said the pods are fully cabled, factory tested and delivered in transportable containers, with the aim of reducing on-site deployment time from several months to days once site utilities are ready.
NVIDIA said the DSX reference architecture encompasses computing, networking, storage, software, power and cooling. It added that its DSX MaxLPS power-management system can use 30% less power for the same output, while the platform’s liquid-cooling design can reduce water use in suitable climates.
OpenShell for AI Agent Security
A central focus of the event was AI agents, which NVIDIA described as systems that can reason, plan, use tools and access enterprise knowledge. The company introduced OpenShell, an open-source secure runtime designed to control what agents can access and execute.
OpenShell uses isolated sandboxes, policy enforcement for system calls, file access and network connections, and formal verification intended to ensure agents do not receive more access than permitted. NVIDIA Sentry, an optional product running on BlueField-4 hardware, adds monitoring and policy enforcement in a hardware-isolated environment, according to the company.
NVIDIA said OpenShell is available under the Apache 2.0 license and can be used from edge systems to cloud infrastructure. It also announced a partnership with Anthropic to develop Claude OpenShell agents.
The company said businesses will increasingly use a mix of closed, open and customized AI models. NVIDIA highlighted its open-model portfolio, including Nemotron for language applications, Cosmos for world models, GR00T for robotics, Alpamayo for autonomous navigation, Ising for quantum-classical computing and Earth-2 for physics AI.
NVIDIA also described an “Agent Knowledge Foundation,” including Nemotron Parse, Nemotron Embed, Nemotron Rerank and Kumo, intended to help organizations use both unstructured documents and structured databases with AI agents. The company said CUDA-X libraries remain a core part of its strategy, providing accelerated tools for workloads such as fluid dynamics, quantum chemistry, data processing and rendering.
Physical AI, Vehicles and Robotics
NVIDIA said physical AI will be a major next phase of AI development, including telecommunications, medical applications, autonomous vehicles and humanoid robots. The company said it completed its first commercial phone call using CUDA-based technology on T-Mobile’s network with Nokia after nearly eight years of work.
In autonomous vehicles, NVIDIA said its Hyperion reference architecture includes 14 cameras, nine radar systems, three lidars and other sensors, supported by dual NVIDIA GPU systems. It announced that the Volkswagen van, Jaguar Land Rover’s new vehicle, Volvo’s fleet and Mercedes’ next-generation vehicles will be Hyperion compatible.
For humanoid robotics, NVIDIA said it is developing a reference robot called NVIDIA Agile ONE with Agile Robots. The company said the system will be available in the first quarter of next year and will arrive preassembled with integrated computing and software.
RTX Spark Personal AI Computers
NVIDIA also introduced RTX Spark, a new line of desktops, laptops and workstations developed with Microsoft and the broader PC industry. The company said RTX Spark is based on its N1X chip and delivers 1 petaflop of performance with 128 GB of memory. NVIDIA said a DGX Station running Windows will offer 20 petaflops and 754 GB of memory and is expected to be available by the end of the year.
The event concluded with a demonstration in which an AI agent used an RTX Spark system, cloud-based training resources, Omniverse digital-twin software and NVIDIA tools to train a small robot called Microduck to navigate a course.
NVIDIA said its broader objective is to provide the infrastructure, models, security software and development tools needed for European companies to deploy AI agents and physical AI systems across their operations.
About NVIDIA (NASDAQ:NVDA)
NVIDIA Corporation (NASDAQ: NVDA) is a technology company that designs graphics processing units (GPUs), accelerated computing platforms, networking products and related software. Its technologies support artificial intelligence, machine learning, scientific computing, data analytics, professional visualization and computer graphics.
The company’s products include GeForce GPUs for gaming and personal computing, data center GPUs and systems for AI and high-performance computing, and networking solutions used to connect and manage large-scale computing infrastructure.
