By Ethan BrooksNvidia Chief Executive Jensen Huang urged developers and industrial leaders in Berlin on Wednesday, October 21, 2026, to accelerate Europe AI data center construction using a new modular computing system. Speaking before roughly 3,000 attendees at the Tempodrom auditorium, Huang also introduced open-source safety software to keep autonomous artificial intelligence tools under strict control.
Closing Europe’s Computing Capacity Bottleneck
During his keynote at the Nvidia GTC conference in Germany’s capital, Huang noted that Europe currently holds only about 5% of global AI computing capacity, trailing far behind the 75% share located in the United States. Lack of massive computing power, commonly referred to as compute by software engineers, has been a major obstacle for European startups training advanced models.
Venture capital leaders, including Jeannette zu Fürstenberg, president and managing director of investment firm General Catalyst, emphasized ahead of the address that limited computing infrastructure remains the primary bottleneck holding back European developers. Without local high-density server facilities, regional researchers often struggle to compete with counterparts in North America and Asia.
To overcome lengthy construction timelines, Huang showcased the Vera Rubin DSK Pod, a prefabricated modular computing unit built by manufacturing partners and shipped in a container-style enclosure. According to Huang, the ready-made pod can cut the time required to deploy an artificial intelligence facility at a prepared site from several months down to just a few days.
Huang told developers, analysts, and executives that the company has invested $10 billion into European startups and corporate partners over the past year to jump-start the regional ecosystem. Despite a brief interruption by protesters holding a banner calling on him to clean up his business, Huang promised to speak with them later and asked security personnel to avoid rough handling.
Industrial Partnerships and Physical AI
Wearing his trademark black leather jacket and sneakers, Huang emphasized that Europe’s artificial intelligence economy is gaining momentum as the Technology industry shifts from text-based language models toward physical AI integrated directly into manufacturing and robotics. He declared that every corporation in the future will operate as an artificial intelligence company.
Throughout the presentation, Huang highlighted collaborations with major German industrial and automotive manufacturers, including Siemens, Mercedes-Benz, and Volkswagen subsidiary Audi. Those partnerships pair Nvidia’s specialized semiconductors and software — already relied upon by leading AI developers such as OpenAI and Anthropic — with European engineering and factory automation.
Mercedes-Benz CEO Ola Källenius joined the discussion to highlight the automaker’s joint work with Nvidia on next-generation autonomous driving systems. Källenius stated that full vehicle autonomy has the potential to drive traffic fatalities caused by human error down to zero and deliver substantially safer public roadways across Europe.
Containing Autonomous Applications with Digital Sandboxes
Beyond hardware speed, Huang addressed growing enterprise concerns over artificial intelligence applications acting unpredictably or exceeding their authorized permissions. He outlined a new open-source software framework called OpenShell, which confines AI agents inside a secure digital sandbox during operation.
OpenShell allows organizations to exercise external oversight over what artificial intelligence applications are doing in real time and automatically block actions that violate corporate security policies. By combining rapid modular infrastructure deployment with built-in software guardrails, Nvidia aims to help European enterprises scale advanced computing safely.
