ItInsightsWorld AIapplications: Europe’s Cooling-Water and Power Questions Behind the NVIDIA DSX AI Factory Platform

For ItInsightsWorld AIapplications readers, the practical question is not only whether compute exists, but whether Europe’s enterprises, developers and governments can run it within water, electricity and grid limits. The public materials from NVIDIA describe the NVIDIA DSX AI Factory Platform as a full-stack reference architecture for that constraint. They also name NVIDIA DSX components that let partners adapt cooling, power consumption and grid response.

What the application layer should notice about NVIDIA DSX

Compute, networking, storage, software, power and cooling are codesigned as one system. The materials position NVIDIA DSX platform as an architecture that combines design-time tools, runtime software and facility systems. Within it, DSX MaxLPS handles dynamic power allocation, DSX Flex coordinates workload power consumption with grid conditions, DSX Reference Design provides validated infrastructure architectures, DSX Sim enables simulation, DSX OS supports infrastructure management, and DSX Exchange connects computing and facility systems.

Cooling assumption change: from evaporative water use to dry-cooler designs

Traditional towers remove heat by evaporating water and require replacement water. The reference designs connect liquid-cooled computing with dry-cooler-based heat rejection, avoiding evaporation. A closed loop lets cooling liquid absorb heat from computing equipment, while outdoor dry coolers release that heat into the air without evaporating water. NVIDIA Vera Rubin supports 100% liquid cooling with coolant temperatures of up to 45 degrees Celsius. Dry-cooler-based designs can reduce facility cooling water consumption to near zero. At 2-gigawatt scale, the platform can save 54 million liters of water a day, equivalent to the daily water use of 180,000 households. Whether supplemental mechanical cooling is needed during the hottest periods depends on local climate.

For application owners, this reframes an infrastructure assumption: the cooling path is part of the AI factory, not a downstream facility detail.

Power efficiency: using DSX MaxLPS inside the same budget

Power availability defines how much capacity can be brought online. DSX MaxLPS dynamically allocates power within an AI factory’s provisioned budget. The stated effect is generating the same number of tokens with up to 30% less power. This can also support more GPU capacity and higher throughput within existing power budgets. The software is being explored by CoreWeave, Firmus, Google Cloud, IREN, Lambda, Nebius, Nscale and Sharon AI.

In a proof of concept on NVIDIA HGX B200 systems, Lambda demonstrated 24% more cluster token throughput and 19% more GPU capacity within a fixed power budget.

Grid alignment: DSX Flex, demand response and storage

By helping AI factories become flexible assets for the power grid, DSX Flex enables operators to adapt workload power consumption to utility needs, demand-response events and energy prices. It can also coordinate electricity from the grid, onsite renewables and energy storage, so AI factories can operate as flexible power loads that support grid reliability and enable faster grid connection.

In an early trial, Emerald AI, EPRI, Nebius, the U.K. National Grid and NVIDIA showed that reductions of up to 40% in the cluster’s power draw were possible on an NVIDIA Blackwell Ultra GPU cluster in response to simulated grid events, while critical workloads continued running.

Case view: Nscale in Iceland and Portugal facilities

Through DSX MaxLPS, DSX Flex and DSX Sim, Nscale is demonstrating how NVIDIA DSX connects power optimization, grid-responsive operations and design.

An evaluation run in an Nscale AI factory in Iceland used DSX MaxLPS across four NVIDIA GB300 NVL72 racks and showed 49% higher inference throughput and 37% more managed GPU capacity within the same 264-kilowatt provisioned power budget. These results show how dynamic power allocation can unlock additional capacity from existing infrastructure.

Nscale is also exploring DSX Flex to align workload power consumption with grid conditions. For facilities in Portugal and Iceland, Nscale is using DSX Sim to create digital twins that help evaluate power, cooling and infrastructure configurations before deployment. Virtual validation lets Nscale identify potential issues earlier, reduce deployment risk and accelerate the path from planning to productive capacity.

Design assets: DSX Sim and SimReady Central Library

NVIDIA is launching SimReady Central Library, which includes validated power and cooling equipment assets from partners like Vertiv and accelerates the creation of DSX AI factory digital twins.

When used with DSX Sim, ecosystem partners including PTC are helping create simulation-ready assets with computer-aided design. These assets enable SimReady workflows that let builders model infrastructure configurations and evaluate power and cooling decisions before equipment is installed, so operators can validate designs earlier, reduce deployment risk and move faster from planning to construction.

Production readiness: NVIDIA DSX OS and its three components

NVIDIA is expanding its modular, open source NVIDIA DSX OS portfolio with three components in developer preview to help operators turn installed infrastructure into productive capacity:

• NVIDIA Config Manager automates network configuration for AI factories.
• NVIDIA Rack Management Service provides rack-level management of power, inventory, firmware and switches.
• NVIDIA Cluster Readiness Engine runs distributed workloads before production workloads arrive to validate GPU clusters and isolate failures to specific nodes.

These components, together with NVIDIA Infrastructure Controller, NVIDIA AI Cluster Runtime and NVSentinel, help operators automate infrastructure lifecycle management, maintain consistent runtime environments and strengthen health validation, fault attribution and recovery.

Aleria, Cassava, CoreWeave, Crusoe, Firmus, Firebird, G42, GMI Cloud, IREN, Lambda, Nebius, Nscale, Radiant, Sharon AI and Volta are cloud partners exploring the latest components of NVIDIA DSX OS.

Comparison table: what each platform block addresses

Block	Function described	Partners or components mentioned	Evidence or qualification
Power allocation	Dynamically allocate power within a provisioned budget	DSX MaxLPS; CoreWeave, Firmus, Google Cloud, IREN, Lambda, Nebius, Nscale, Sharon AI	Lambda on NVIDIA HGX B200 showed 24% more cluster token throughput and 19% more GPU capacity
Grid response	Coordinate workload power with grid conditions, demand-response events and energy prices	DSX Flex; Emerald AI, EPRI, Nebius, U.K. National Grid, NVIDIA	Trial showed reductions of up to 40% in the cluster’s power draw on NVIDIA Blackwell Ultra
Simulation	Model power, cooling and infrastructure before deployment	DSX Sim; SimReady Central Library, PTC, Procore, Siemens, Dassault Systèmes, ABB	SimReady assets and digital twins support earlier validation
Equipment	Provide qualified facility components	Trane Technologies, Vertiv, Hitachi Energy	NVIDIA DSX Ready categories include LiquidStack GigaModular, 2.3-megawatt CoolChip and NVIDIA DSX Ready BESS
Operations	Manage infrastructure lifecycle, runtime and fault validation	NVIDIA DSX OS, NVIDIA Infrastructure Controller, NVIDIA AI Cluster Runtime, NVSentinel	three components in developer preview; partners exploring latest components

Europe’s ecosystem: design, power and cooling partners

NVIDIA’s industrial software, power and cooling partners are advancing Europe’s AI factory buildout through DSX Reference Design, lifecycle digital twins and qualified equipment that connect facility infrastructure with accelerated computing.

For NVIDIA Vera Rubin infrastructure, Schneider Electric has developed power and cooling reference designs and is working with AVEVA on digital twins spanning infrastructure design, construction and operations. With its construction management platform integrated with DSX Sim, Procore is helping manage AI factory builds. Siemens is using DSX Sim to simulate power, cooling and acoustics. Dassault Systèmes and Vertiv are developing digital twin models of Vertiv’s OneCore infrastructure, enabling simulation and optimization through model-based systems engineering.

ABB is developing SimReady assets with DSX Sim. This enables engineers to evaluate source-to-rack architectures before construction while advancing 800 VDC development through its newly announced Infinitus Direct Current portfolio.

Trane Technologies and Eaton recently collaborated with NVIDIA to help boost data center efficiency and cut installation costs with an integrated design built for the NVIDIA Vera Rubin DSX AI Factory Reference Design. The design brings together Eaton’s Beam platform and power infrastructure with Trane Continuum cooling and coordinated controls in a grid-to-chip architecture.

To support thermal infrastructure planning at scale, Trane Technologies has published seven thermal management system reference designs, including three 250-megawatt DSX-based cooling reference designs.

Builders can select equipment qualified under applicable NVIDIA DSX Ready categories. Trane Technologies’ LiquidStack GigaModular and Vertiv’s 2.3-megawatt CoolChip — both coolant distribution units — have qualified under their applicable categories, while Hitachi Energy’s battery energy storage system has earned NVIDIA DSX Ready BESS qualification.

GE Vernova, Hitachi Energy and Siemens Energy are collaborating with NVIDIA on power infrastructure and digital twin initiatives for energy supply and grid integration, connecting energy-system engineering and grid planning based on the facility’s requirements. Hitachi Energy is developing an integration between its energy management system and DSX Flex to optimize grid, generation and storage resources with workload demand.

FAQ: What should AI application owners ask?

Can the platform reduce cooling water?

Dry-cooler-based designs can reduce facility cooling water consumption to near zero, while local climate may require supplemental mechanical cooling.

Does more throughput require more electricity?

Within a fixed power budget, DSX MaxLPS is described as supporting the same tokens with up to 30% less power. In an Nscale AI factory in Iceland, four NVIDIA GB300 NVL72 racks showed 49% higher inference throughput and 37% more managed GPU capacity within the same 264-kilowatt provisioned power budget.

Can grid events interrupt critical workloads?

The early trial showed reductions of up to 40% in the cluster’s power draw while critical workloads continued running.

Is this only for hardware teams?

The materials describe infrastructure that serves Europe’s enterprises, developers and governments, so application, product and operations owners need visibility into cooling water, power allocation and grid alignment.

Checklist for evaluating an NVIDIA DSX deployment

• Confirm that compute, networking, storage, software, power and cooling are treated as one system.
• Review DSX Reference Design options and DSX Sim digital twins for sites in Portugal or Iceland-like climates.
• Check whether DSX MaxLPS can be tested against a provisioned power budget.
• Test DSX Flex against utility needs, demand-response events and energy prices.
• Validate NVIDIA DSX OS lifecycle components: NVIDIA Config Manager, NVIDIA Rack Management Service and NVIDIA Cluster Readiness Engine.
• Look for DSX Exchange connections between computing and facility systems.
• Qualify equipment through NVIDIA DSX Ready categories, including LiquidStack GigaModular, Vertiv’s 2.3-megawatt CoolChip and Hitachi Energy’s NVIDIA DSX Ready BESS.

Step tutorial: from planning to productive AI capacity

1. Define the application workload and the site’s power budget.
2. Use DSX Reference Design and DSX Sim to evaluate cooling, power and infrastructure configurations before installation.
3. Import SimReady assets from SimReady Central Library, where partners such as Vertiv provide validated power and cooling equipment assets.
4. Model grid conditions with DSX Flex and evaluate whether onsite renewables and energy storage can support flexible loads.
5. Run DSX MaxLPS tests to determine whether the same number of tokens can be generated with up to 30% less power.
6. Bring NVIDIA DSX OS components into production: automate network configuration, rack management and cluster readiness before production workloads arrive.

Common misconceptions to avoid

• Liquid cooling always means evaporative water loss: DSX reference designs connect liquid-cooled computing with dry-cooler-based heat rejection, and NVIDIA Vera Rubin supports 100% liquid cooling with coolant temperatures of up to 45 degrees Celsius.
• Power efficiency is only a hardware issue: DSX MaxLPS, DSX Flex and NVIDIA DSX OS connect allocation, grid response and lifecycle operations.
• Grid flexibility requires stopping production: the trial showed reductions of up to 40% in the cluster’s power draw while critical workloads continued running.

Action suggestions and final read

Application and operations teams reviewing AI capacity in Europe’s markets should ask partners for evaluation evidence around DSX MaxLPS, DSX Flex, DSX Sim and NVIDIA DSX OS. The Nscale, Lambda and U.K. National Grid trial examples are useful comparison points because they show measured effects under constrained power budgets or simulated grid events.

Where local climate requires supplemental mechanical cooling, DSX Sim digital twins and qualified equipment such as Trane Technologies’ LiquidStack GigaModular or Vertiv’s CoolChip can support earlier validation. The final decision point is not whether a platform can model or allocate resources, but whether those capabilities match the cooling, grid and workload conditions of a specific deployment.