Tech • AI • Robotics • Game
Jensen Huang’s Keynote Highlights at NVIDIA GTC Berlin 2026
9.4/10TL;DR
Nvidia used GTC to position itself as a full-stack infrastructure company for the emerging AI economy, spanning data-center systems, agent software, automotive platforms, robotics and new personal AI computers.
KEY POINTS
AI as a new industrial era
Nvidia framed the current moment as the start of a new industrial revolution driven by artificial intelligence. The company argued that AI will be used by individuals, embedded in businesses and reshape every major industry while creating entirely new markets.
From chips to AI factories
Nvidia described itself as an end-to-end AI factory platform rather than only a chipmaker. It highlighted the scale of modern computing production as unprecedented and said its DSX pod represents the next step in deploying large-scale AI infrastructure built to be fungible, durable and productive.
Shift from models to agents
The company said AI systems are moving beyond standalone models toward full agents. It defined five core elements of an agent as models, sandbox, harness, knowledge and tools, and argued that secure deployment technology is essential if such agents are to be used broadly across enterprises and public settings.
Launch of Open Shell
A newly introduced platform called Open Shell was presented as a system for safely deploying agents at scale. Nvidia said the product arrived as major companies are actively looking for ways to roll out AI agents globally and suggested adoption interest is already broad.
Open models and sovereign intelligence
Nvidia said building custom AI systems is becoming more accessible, starting with the model itself. The company emphasized support for open models, including not just open weights but the wider recipe needed to turn proprietary data into what it called sovereign intelligence.
CUDA X as the software core
Beyond hardware, Nvidia stressed that it is fundamentally an algorithm company that builds silicon to accelerate those algorithms. At the center of that strategy are CUDA X libraries and CUDA X Tools, which it described as the software layer that gives AI agents practical capabilities for real-world work.
Tool vendors becoming AI vendors
The company predicted that every major tools company will become an AI company. In that model, software vendors would provide highly optimized agents designed specifically to use their own tools effectively, extending AI from general assistants into specialized industrial and enterprise workflows.
Automotive and physical AI
In physical AI, Nvidia said automakers increasingly want vehicles to become part of ride-hailing networks. Its Hyperion platform was described as an industry standard and a giant virtual fleet, with new compatibility announced for the VW van, JLR, Volvo fleets and next-generation Mercedes vehicles.
Humanoid robotics and RTX Spark
The company also pointed to humanoid robotics as a major future industry and unveiled a reference robot called Nvidia Agile 1 to accelerate development. Separately, after four years of partnership with Microsoft, it introduced RTX Spark, a new line of computers aimed at expanding AI capabilities onto personal systems.
European expansion
Nvidia said it has spent the past year helping lay foundations for a new AI industry in Europe. The company cited more than 12,000 partnerships supporting that effort and called for broader collaboration to build Europe’s AI economy and put AI into practical use.
CONCLUSION
Nvidia is broadening its pitch from GPUs to a complete AI infrastructure stack, tying together data centers, agent software, vehicles, robots and PCs. The strategy reflects a race to define the platforms that could underpin the next phase of the global AI economy.
Ask a question