Europe’s AI sector needs more computing capacity to support its ambitions, and Nvidia wants to help close that gap. Speaking at the company’s GTC conference in Berlin, CEO Jensen Huang presented a modular system designed to speed up the construction of AI data centers, alongside software intended to give organizations more control over AI applications.
Huang described Europe’s AI economy as gaining momentum and pointed to the region’s industrial base as an opportunity for the next phase of development. As AI moves beyond language models into manufacturing, vehicles and other physical environments, European companies will need infrastructure capable of supporting increasingly demanding workloads.
“Europe is building its AI economy,” Huang said, adding that he expects AI to become relevant across industries.
Computing capacity remains a challenge
Access to computing power is one of the main obstacles facing European AI developers. According to figures cited in the original report, Europe has approximately 5% of the world’s AI computing capacity, compared with 75% in the United States.
That gap matters because training and operating advanced AI systems require significant infrastructure: specialized chips, data centers, energy and networking capacity. Without sufficient access to these resources, companies may struggle to develop and scale their own AI products.
“The big bottleneck is compute,” said Jeannette zu Fürstenberg, president and managing director of investment firm General Catalyst, ahead of Huang’s speech.
A modular approach to AI data centers
To address the infrastructure challenge, Huang presented Nvidia’s Vera Rubin DSK Pod, a modular computing system built by partners and delivered in a container-like enclosure.
According to Nvidia, the system can reduce the time required to set up an AI data center at a prepared site from months to days. The approach is intended to make deployment more predictable by bringing together the computing infrastructure needed for AI workloads in a ready-made unit.
Faster installation, however, is only one part of the challenge. A prepared site still needs the power, cooling, connectivity and operational support required to run an AI data center at scale.
Huang also said Nvidia had invested $10 billion in European startups and partners over the previous year, highlighting the company’s interest in expanding the region’s AI ecosystem.
From language models to physical AI
Throughout the presentation, Huang highlighted Nvidia’s work with European companies, including Siemens, Mercedes-Benz and Audi. These partnerships reflect the growing interest in physical AI: systems that can perceive, interact with and operate in the physical world, including in factories and vehicles.
Mercedes-Benz CEO Ola Källenius joined Huang to discuss the company’s work with Nvidia on autonomous driving. Källenius said he believes full autonomy could make streets safer by reducing fatalities caused by human error.
Reaching that goal will depend on how well autonomous systems perform across real-world conditions, as well as on testing, regulation and public trust.
Keeping AI applications within defined boundaries
Huang also addressed the need for controls around AI applications. He presented Nvidia’s open-source OpenShell software, which is designed to place AI applications in a digital sandbox and enforce external policies on their actions.
The idea is to give organizations a way to define what an AI application can access and do, and to block actions that violate established rules. As more companies introduce AI agents into workflows that involve sensitive data, software tools or business systems, these controls become an important part of deployment.
For engineering teams, the practical question goes beyond what an AI application can do. It also includes what it is permitted to access, how its actions are monitored and who remains responsible when something goes wrong.
What this means for European businesses
Nvidia’s announcements bring together two issues shaping AI adoption in Europe: access to infrastructure and control over how AI systems operate.
More computing capacity could give European developers and businesses greater room to build and scale AI applications. At the same time, organizations adopting these tools will need to make considered decisions about security, permissions, integration and ongoing oversight.
At Control F5 Software, we see AI as most useful when it is introduced with a clear purpose and supported by sound engineering practices. Whether a business is exploring AI-powered features or integrating agents into existing workflows, the work starts with understanding the problem, defining boundaries and building systems that teams can maintain and trust over time.
We have helped 20+ companies in industries like Finance, Transportation, Health, Tourism, Events, Education, Sports.