Nvidia unveils DSX Pod modular data centers to accelerate European AI infrastructure
Jensen Huang announced a plug-and-play data center solution at GTC Berlin, aiming to bypass slow construction timelines and boost local AI production capabilities.
Key points
- Nvidia introduced the DSX Pod, a modular data center that can be deployed in 14 days.
- The system uses Vera Rubin compute racks to reduce idle time and accelerate infrastructure build-outs.
- Huang announced a robotics platform and Agile One reference robot to support physical AI development.
Nvidia CEO Jensen Huang unveiled the DSX Pod, a modular data center solution, at the GTC conference in Berlin on Wednesday. This announcement marks a strategic shift toward rapid infrastructure deployment, addressing the significant delays often associated with building traditional AI facilities in Europe.
The DSX Pod functions as a plug-and-play AI factory, arriving fully assembled in a large metal container. According to Wired, the unit is packed with Nvidia’s latest Vera Rubin compute racks and can be brought online in just 14 days. This speed contrasts sharply with the months or years typically required for conventional data center construction.
Infrastructure and Deployment Strategy
Huang emphasized that Europe must transition from merely consuming AI to producing it to participate fully in the AI economy. The DSX Pod aims to slash the time compute resources sit idle by accelerating the commissioning process. This approach directly addresses the notorious slowness of European infrastructure projects, offering a faster path to operational capacity.
The event took place at Berlin’s Tempodrom, where Huang delivered a 90-minute pitch to a crowd of 4,000 attendees. The presentation highlighted Nvidia’s role as a fundamental technology provider across various industries. This follows the previous GTC in San Jose in March, which focused on AI inference and compute efficiency, as Threat Times reported earlier.
Robotics and Industry Partnerships
Beyond data centers, Huang positioned humanoid robotics as the next great industry, leveraging Europe’s manufacturing heritage. He announced a full-stack robotics platform to standardize configuration and a preconfigured reference robot called Agile One, expected in the first quarter of next year. These tools are designed to help developers build physical AI systems more efficiently.
Huang also revealed partnerships with European carmakers Volkswagen and Mercedes-Benz on autonomous vehicles. He noted that physical AI extends beyond robotics, with telecom operators integrating Nvidia GPUs into cell towers. During the speech, two Greenpeace protesters unfurled a banner criticizing Nvidia’s business practices, but were removed after Huang instructed security to handle them gently.
What should you do?
Security and technology professionals should monitor the deployment timelines for the DSX Pod as organizations begin integrating these modular units. Understanding the rapid 14-day deployment window is crucial for planning security assessments and compliance checks before systems go live.
Organizations considering physical AI or robotics should evaluate the upcoming Agile One reference robot and the new full-stack platform. These tools may change how physical AI systems are configured and secured, requiring updated risk assessments for industrial environments.
What to do and how to stay safe: Nvidia
- Review procurement contracts for modular data centers to ensure security requirements are met during the accelerated 14-day deployment window.
- Assess the security implications of integrating new robotics platforms and reference hardware like Agile One into existing industrial networks.
- Monitor Nvidia’s announcements regarding partnerships with automotive and telecom sectors for potential supply chain or integration vulnerabilities.
- Evaluate current infrastructure strategies to determine if plug-and-play solutions offer a viable alternative to traditional, slower construction methods.
Step-by-step guide: Prompt Injection Warning Signs: How to Spot AI Manipulation
General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
How long does it take to deploy the DSX Pod?
The DSX Pod can be brought online in 14 days, significantly faster than traditional data center construction.
What compute hardware is included in the DSX Pod?
The modular unit is packed with Nvidia’s latest Vera Rubin compute racks.
When will the Agile One reference robot be available?
Nvidia plans to release the Agile One reference robot in the first quarter of next year.
