From the NVIDIA edition of October 21, 2026
NVIDIA says DSX reference designs can cut facility cooling water to near zero and use up to 30% less power for the same output; partner tests and a simulated UK National Grid trial cited
NVIDIA; Nscale; Lambda; Emerald AI; EPRI; Nebius; U.K. National Grid; Schneider Electric; Siemens and other ecosystem partners · AI infrastructure (data-centre energy and water) · Europe (Iceland, Portugal, UK trials); partner sites
- Publisher
- NVIDIA Newsroom
- Original publication
- 2026-10-21
- Source retrieved
- 2026-10-22
- Event date
- 2026-10-21
What happened
NVIDIA says its DSX reference designs pair 100% liquid-cooled Vera Rubin infrastructure (coolant up to 45 degrees Celsius) with dry coolers in a closed loop, which can reduce facility cooling water to near zero; it estimates 54 million litres a day saved at 2-gigawatt scale versus cooling towers, and notes local climate determines whether supplemental mechanical cooling is needed. DSX MaxLPS dynamically allocates power within a provisioned budget; NVIDIA says it can produce the same tokens with up to 30% less power. Cited partner tests: Lambda, 24% more cluster token throughput and 19% more GPU capacity on HGX B200 within a fixed budget; Nscale in Iceland, 49% higher inference throughput and 37% more managed GPU capacity across four GB300 NVL72 racks within 264 kW. In an early DSX Flex trial with Emerald AI, EPRI, Nebius and the U.K. National Grid, a Blackwell Ultra cluster cut power draw by up to 40% in response to simulated grid events while critical workloads continued. New open-source DSX OS components entered developer preview.
Why it matters
Relevant to SLED as context for local and state decisions on AI data centres, where water and grid impact dominate public debate (as in the archived Lowndes County thread). It gives officials vendor-stated design options to ask about, not evidence that any specific facility will use less water or power. A U.K. grid operator's participation in a trial is the most public-sector-relevant element.
Evidence and measured results
Vendor release. Efficiency figures are maxima ('up to') under unstated conditions; the 54-million-litre figure is a modelled comparison, not a site measurement. Partner tests are described without methods, baselines or independent review. The grid trial used simulated events.
Limitations and uncertainty
All figures from NVIDIA or partners; no independent verification; climate-dependent supplemental cooling; dry coolers' effect on power, noise and land not described.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-10-22; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- Public bodies must judge AI data-centre water and grid impact with little verified data.
- Stakeholders
- County commission or planning board
- State utility commission
- Water authority
- Facility developer
- Discovery questions
- Which cooling method will the facility use, and for how many hours a year is supplemental cooling expected?
- Will the operator participate in demand response, and on what terms?
- What metered water and power data will be reported publicly?
- Value hypothesis
- Interpretation: design choices such as dry coolers and flexible load could reduce water use and grid stress; site-level results are unknown.
- Potential engagement
- Advisory on measurable water, power and demand-response conditions for an AI facility approval or state guidance.
- Unsupported claims
- Zero water use for a complete facility
- 30% power savings at any given site
- Grid reliability benefits from live events
Pre-sales engineering
Role takeaway
- Fit
- New liquid-cooled AI facilities where cooling design is still open.
- Architecture and integration
- Liquid-cooled racks to closed-loop coolant to dry coolers; MaxLPS within provisioned power; Flex interface to utility signals.
- Prerequisites
- Climate analysis for supplemental cooling
- Utility demand-response programme
- Workload priority classification
- Constraints
- Hot climates need supplemental mechanical cooling
- Dry coolers' power and noise not quantified
- Partner test methods unpublished
- Security
- Utility signalling interfaces need authentication and segregation from compute management networks.
- Proof of value
- Meter water use over a full summer
- Compare power per unit of output with and without MaxLPS on the same workload
- Run a live demand-response event under utility observation
Delivery
Role takeaway
- Work
- Install metering, set reporting, agree demand-response procedures and publish results.
- Dependencies
- Utility agreement
- Metering infrastructure
- Operator reporting commitment
- Ownership
- Facility operator, with oversight by the approving body or regulator.
- Skills and adoption
- Facilities and operations staff need training on flexible-load procedures and coolant systems.
- Governance checkpoints
- Approval conditions specifying metering
- Quarterly public reporting
- Annual review of demand-response performance
- Acceptance criteria
- Proposed: annual metered water use per MW published
- Proposed: number and depth of live curtailment events met
- Proposed: supplemental cooling hours reported against forecast
- Risks
- Claims treated as commitments without metering
- Supplemental cooling erodes water savings in heat waves
- Curtailment affecting critical public workloads
Implementation considerations
Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.
Architecture and integration
What must connect, and where does the AI sit in the workflow?
Water savings depend on choosing dry-cooler heat rejection with high-temperature liquid cooling; power-per-output gains depend on software power allocation within a fixed budget.
Governance
Who approves, reviews and stays accountable for outcomes?
Gives regulators and local boards specific design questions (cooling method, supplemental cooling, demand-response participation) to make conditions of approval measurable.
Security and privacy
What data, permissions and controls need testing?
Grid-responsive operation implies data exchange with utilities; the release does not address security of that interface.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
Not addressed.
Procurement
What should contracts, pricing and exit terms secure?
Buyers and approving bodies could specify cooling method and metered water and power reporting rather than relying on platform maxima.
Operating model
Which teams own the service once it runs?
Treats an AI factory as a flexible grid load that can curtail on request, which requires agreements with utilities and workload priority rules.
Publication history
- 2026-10-21NVIDIA · Issue 469 resources
Stable resource ID: nvidia-newsroom-dsx-europe-water-power-grid-flex-claims-20261021