Nvidia DSX Blueprint Promises Near-Zero Water Use for European AI Data Centres
European infrastructure firms adopt Nvidia’s DSX reference design to slash water consumption and boost efficiency in new AI factory builds.
Key points
- Nvidia states its DSX blueprint can reduce cooling water usage to almost zero via closed-loop liquid cooling systems.
- A two-gigawatt site using this design could save 54 million litres of water daily, according to the chipmaker.
- European partners including Siemens, Schneider Electric, and Nscale are integrating or testing DSX components and tools.
Nvidia announced on Wednesday at its GTC Berlin conference that European infrastructure companies are building AI data centres using its DSX reference design. The chipmaker stated that this blueprint enables facilities to cut cooling water consumption to nearly zero levels.
The announcement followed a day after Germany signed an AI partnership with Nvidia at a Berlin investment summit. Nvidia describes these facilities as AI factories, which are data centres constructed to train and run artificial intelligence models at scale.
Technical Specifications and Efficiency Gains
Nvidia first presented the DSX platform at GTC Taipei on May 31, 2026. The design treats chips, networking, power, and cooling as a single integrated system. Water savings result from closed-loop liquid cooling, where coolant passes chips and releases heat to outside air via dry coolers rather than evaporating water.
The company estimated that a two-gigawatt site built with this method could save 54 million litres of water per day. Local weather conditions determine whether additional cooling is required on the hottest days, Nvidia added.
European Partner Integration
London-based Nscale tested the DSX tool MaxLPS on four racks of GB300 servers at its Iceland site. Nvidia reported that inference throughput increased by 49% within the same 264-kilowatt power budget. MaxLPS can produce identical AI output using up to 30% less electricity.
Nscale is also constructing a data centre in Portugal and using DSX to create digital models of both sites before construction begins. Josh Payne, Nscale’s chief executive, stated that Europe’s AI ambitions require maximizing every megawatt of power.
A broader group of European engineering firms is designing platform components. Siemens is simulating power, cooling, and noise, while Schneider Electric is publishing power and cooling designs. Dassault Systèmes is building digital twins of Vertiv equipment.
ABB, Siemens Energy, Hitachi Energy, and Trane Technologies are also participating in the initiative. In the UK, a trial with National Grid and cloud firm Nebius tested how Nvidia chip clusters respond to grid strain.
Nvidia said the cluster reduced power draw by up to 40% during simulated grid events while critical workloads continued running. Warren Barkley, who leads DSX product management at Nvidia, said Europe’s technology leadership depends on efficiently converting power and infrastructure into intelligence.
What should you do?
Nvidia provided no specific dates for when new features will be generally available. Several DSX operating software tools remain in developer preview status. This follows earlier developments where Nvidia released a factory operations blueprint for European industrial AI agents and saw Vera CPUs gain traction among sovereign AI partners, as Threat Times reported on October 21, 2026.
What to do and how to stay safe: Nvidia
- Monitor vendor announcements for general availability dates of DSX operating software tools currently in developer preview.
- Evaluate current data centre cooling infrastructure to assess compatibility with closed-loop liquid cooling systems.
- Review power budgets for AI workloads to determine potential efficiency gains from tools like MaxLPS.
- Track regulatory developments regarding water usage and energy consumption for large-scale AI facilities in Europe.
Step-by-step guide: Mobile Device Security: How to Protect Data Beyond Passwords
General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
How much water can a DSX-based data centre save?
Nvidia stated that a two-gigawatt site built with the DSX blueprint could save 54 million litres of water per day.
Which European companies are involved in the DSX platform?
Partners include Siemens, Schneider Electric, Dassault Systèmes, ABB, Siemens Energy, Hitachi Energy, Trane Technologies, Nscale, and Nebius.
Are DSX tools available for general use now?
No, Nvidia gave no dates for general availability, and several DSX operating software tools are currently in developer preview.
