NVIDIA Announces European Partner Adoption of DSX AI-Factory Platform
Редакция ЦОД:Главное
NVIDIA has announced that European partners are adopting its NVIDIA DSX reference architecture for the design and operation of AI factories. The platform combines compute, networking, storage, software, power and cooling, with NVIDIA positioning it around lower cooling-water use and reduced energy consumption for comparable AI inference output.
- NVIDIA has announced European partner adoption of its DSX AI-factory reference architecture.
- DSX combines IT infrastructure with power, cooling, lifecycle software and grid-oriented load management.
- NVIDIA says closed-loop liquid cooling and dry coolers can enable near-zero facility cooling-water consumption.
- DSX MaxLPS is claimed to use up to 30% less power for the same token output; named operators plan to use it.
- The announcement expands the AI-factory ecosystem to industrial software, power-equipment and cooling suppliers.
NVIDIA presented NVIDIA DSX for the European market at GTC Berlin, describing it as a full-stack reference architecture for AI factories. The company said European partners are adopting the platform to design and operate AI infrastructure, while software, industrial-design, power and cooling suppliers are contributing to the ecosystem.
The reference designs use closed-loop liquid cooling and dry coolers, which NVIDIA says can enable near-zero facility cooling-water consumption. NVIDIA also cites potential cooling-water savings of up to 54 million litres per day for a hypothetical 2 GW deployment; this is an illustrative energy-capacity scenario rather than the capacity of a disclosed European project. The company characterises this volume as equivalent to the daily water use of 180,000 households.
For inference operations, NVIDIA says its DSX MaxLPS software can reduce power consumption by up to 30% while generating the same number of tokens. CoreWeave, Firmus, Google Cloud, IREN, Lambda, Nebius, Nscale and Sharon AI plan to use DSX MaxLPS, according to NVIDIA. The announcement does not confirm that any associated AI factories have entered service or achieved the stated efficiency levels.
NVIDIA is also positioning DSX Flex as a mechanism to make AI factories flexible grid loads, supporting power-system reliability and potentially accelerating grid interconnection. This places workload and load management alongside conventional data-center power design as part of the AI-factory infrastructure stack.
The announced DSX ecosystem includes industrial software providers AVEVA, Dassault Systèmes and Siemens, alongside power and cooling suppliers ABB, Eaton, GE Vernova, Hitachi Energy, Schneider Electric, Siemens Energy, Trane Technologies and Vertiv. NVIDIA SimReady Central is being launched on Hugging Face with validated power and cooling-equipment models for AI-factory digital twins. NVIDIA also describes DSX OS as open-source software for infrastructure-lifecycle automation, health monitoring and operational resilience.
For operators and infrastructure suppliers, the significance is the integration of accelerated computing with facility power, cooling, modelling and grid-facing load controls. Whether the platform delivers the claimed water and energy effects will depend on deployment-specific design and operating conditions.
The announcement signals a shift in AI data-center competition from discrete IT components toward integrated design of compute, power and cooling infrastructure. If adopted, DSX could increase demand for solutions that link GPU systems with liquid cooling, digital twins and flexible-load management, addressing grid-capacity and cooling-resource constraints. The cited performance figures are NVIDIA’s potential-effect claims, not reported operating results from specific facilities.
- NVIDIA — Newsroom(первоисточник) / АнглийскийПервоисточник →
Материал подготовлен редакцией ЦОД: Главное на основе указанных первоисточников.