NVIDIA unveils new AI factory design, robotics and safety tools at Berlin conference
NVIDIA CEO Jensen Huang has unveiled new technologies for AI infrastructure, autonomous agents, robotics and healthcare at GTC Berlin 2026, outlining how European companies can build their own AI capabilities, Qazinform News Agency correspondent reports.
Speaking at the three-day conference in Berlin, Huang said Europe was entering a new industrial revolution driven by AI. NVIDIA has committed nearly $10 billion to Europe’s technology ecosystem and works with more than 12,000 partners across the region, he said.
Huang described AI as a five-layer system, spanning energy, computing infrastructure, AI factories, models and applications. He argued that Europe must produce AI rather than simply consume it.
One of the main announcements was NVIDIA DSX POD, a reference design for AI factories built with NVIDIA Vera Rubin computing systems. Partners can assemble and test the facilities inside modular containers before delivery. Once a site has the necessary utilities, deployment could take as little as 14 days, down from six months.
NVIDIA also said its DSX AI Factory Platform can reduce water consumption for cooling by up to 100% and deliver the same AI output with up to 30% less power.
New tools for safer AI agents
Huang presented AI agents as a new computing platform. Unlike systems that only answer questions, agents can reason, use tools, access information and carry out tasks.
However, their ability to access files, networks and other resources creates security risks. NVIDIA introduced the OpenShell Reference Architecture for Safe AI Agents, designed to control agent permissions, isolate their activities and monitor their actions.
AI moves into factories, hospitals and vehicles
NVIDIA demonstrated how its Agent Toolkit can combine company data, AI models and specialized tools to support work across industries.
At Lila Sciences, agents help researchers explore potential treatments for B-cell lymphoma through virtual experiments and automated laboratory work. At Siemens, they help prepare products for manufacturing by testing plans in digital twins. At Dassault Systèmes, engineering agents use simulations to identify problems and refine vehicle designs.
In healthcare, NVIDIA and Siemens Healthineers are developing AI enabled adaptive radiation therapy. The system combines AI agents, digital twins and live imaging to adjust radiation targeting as a patient's body and tumor change during treatment.
In the automotive sector, NVIDIA said Volvo Cars was adopting its Hyperion computing platform and Alpamayo software for automated driving. Volkswagen, Jaguar Land Rover and Mercedes-Benz are also preparing vehicles compatible with Hyperion. Mobility partners, including Bolt, Lucid and Einride, are showcasing selected autonomous driving applications.
New humanoid robot platform and personal AI computers
Huang announced Agile ONE SN, a reference humanoid robot built on NVIDIA Isaac GR00T technology. Expected to become available in 2027, the platform features 31 degrees of freedom and aims to reduce researchers' setup time from weeks to days.
The system builds on NVIDIA's tools for collecting robotics data, training models in simulation and running robot software. Partners including Agility Robotics, Figure, KUKA and NEURA Robotics are also using NVIDIA's physical AI technologies.
Huang closed the keynote by demonstrating RTX Spark, a compact computer designed for local AI workloads. The machine delivers one petaflop of performance and 128GB of memory, specifications he compared with NVIDIA's DGX-1 system delivered to OpenAI in 2016.
He showed how an AI agent could train small Microduck robots by creating a simulated course, running training jobs in the cloud and evaluating the results on RTX Spark before deploying them to physical robots.
Four Microducks then appeared onstage, demonstrating movements learned through simulation.
Earlier, Qazinform News Agency reported on NVIDIA’s Rev Lebaredian discussing the potential of digital twins in Kazakhstan and the challenges of implementing the technology.