NVIDIA unveils DSX POD and OpenShell agent security at GTC Berlin

Europe as a producer of AI, not only a consumer

GTC Berlin runs three days at STATION-Berlin, with more than 100 sessions on agentic AI, robotics, physical AI, open models and accelerated computing. Huang's keynote took place at the Tempodrom. He opened by asking what L'Oréal, Revolut, Spotify, Merck, Vodafone, Zara, Porsche, Nestlé and HSBC have in common. His answer: they are all European companies working with NVIDIA on AI.

The keynote followed a weeklong tour of Paris, London, Stockholm and Dublin. According to NVIDIA, the company is committing nearly $10 billion to the technology ecosystem across Europe and working with 12,000+ partners.

Huang described AI as a five-layer cake: energy powers advanced computers that are part of an AI factory, which generates tokens and digital intelligence from the models running on it. Those models are then integrated into every application. "To be part of this AI economy, you don't want to just be consuming the AI; you want to be producing the AI," Huang said, adding that this could be done in Europe and had already started.

DSX POD: AI factories in modular containers

Huang framed NVIDIA's full-stack AI factory, spanning hardware, system software, CUDA-X libraries and models, as a broad platform for building and operating AI infrastructure. He said Vera Rubin, now in full production, extends NVIDIA's computing platform beyond models and large language models to process the entire AI agent. He also showed several rack-scale hardware offerings.

The main announcement was NVIDIA DSX POD, a reference design for complete NVIDIA Vera Rubin AI factories. Partners build them inside modular containers, and they are integrated and validated before delivery. NVIDIA says onsite deployment can shrink from six months to just 14 days once site utilities are ready.

"This is a complete transformation of the world supply chain," Huang said. He added that it will change how people install and build data centers, and that he hopes velocity goes up as a result.

NVIDIA also stated that the NVIDIA DSX AI Factory Platform lets companies reduce water consumption for cooling by up to 100% and generate the same AI output with up to 30% less power.

OpenShell: treating agent safety as engineering

Huang said agents think, reason, iterate and use tools, and that token consumption rises as they become more useful. He described an agent as having five elements: a model, a harness that orchestrates its work and memory, a sandbox that isolates the harness, knowledge and tools.

"The most important thing is to make sure the agent is secure," Huang said. Because an autonomous agent can access files, resources, networks and tools, its privileges should be minimal and it should be monitored.

NVIDIA's answer is the NVIDIA OpenShell Reference Architecture for Safe AI Agents. It is a full-stack design built on NVIDIA OpenShell, an open source, policy-controlled runtime with kernel-level sandboxing that sits between an agent and the systems it can affect. Huang said privileges, policy enforcement and monitoring must adapt dynamically. "Every single time an agent is launched, the privileges are different," he said.

What it means for European AI buyers

For organizations planning AI infrastructure, the DSX POD pitch centers on speed: prefabricated, validated containers instead of a long onsite build. The efficiency figures on water and power are NVIDIA's own claims. The source gives no pricing or availability details for DSX POD or the OpenShell architecture.

For companies deploying agents, NVIDIA is positioning an open source runtime and reference design as a way to limit what agents can reach and to monitor them. Both announcements fit the message of the keynote: Europe should build and run its own AI infrastructure rather than only use AI services.