Product Launches Positive 6

Nvidia DSX Runs Coolant at 45°C to Cut AI Data Center Water Use to Nearly Zero

Nvidia's DSX system uses direct-to-chip liquid cooling with coolant entering at 45°C, enabling near-elimination of water use in some AI data centers. The approach targets high-density AI clusters but introduces an energy trade-off that AI infrastructure teams must engineer around.

· 5 min read · Verified by 3 sources ·

Beat this week

Last 7 days · Product Launches

8 stories
5.9 avg impact
50% positive
0% negative
vs prior 7 days +7 +7 stories vs prior 7 days

Impact 5.9/10 (-0.1 vs prior). Counts are stories in our record, not a market forecast.

Open the change report

Coverage balance Positive coverage leads. Positive coverage exceeds negative coverage by 50 percentage points.

  • 50% positive
  • 50% neutral

This story sits in Product Launches — the counts compare this beat's last 7 days with the previous 7 in our verified record, not a market forecast.

Figures are computed live from our source-verified story record (as of ) The volume change compares this window with the prior 7 days in the same record. — see our methodology for how impact and sentiment are derived.

AI briefing

Key takeaways

6 impact
Positivesentiment
3sources
5min read
  1. Nvidia's DSX system uses direct-to-chip liquid cooling with coolant entering at 45°C, enabling near-elimination of water use in some AI data centers.
  2. The approach targets high-density AI clusters but introduces an energy trade-off that AI infrastructure teams must engineer around.
Drawn from
  • digitaljournal.com
  • hurriyetdailynews.com
  • freemalaysiatoday.com

In this briefing

Mentioned

Key Intelligence

Key Facts

  1. 1In 2025, data centers consumed 222 billion liters (59 billion gallons) of water worldwide for cooling, according to Rystad Energy.
  2. 2Without adaptive measures, data center water consumption could nearly triple to 644 billion liters by 2030; with steps, growth could be kept under double.
  3. 3Nvidia's June report claims its DSX system can eliminate water consumption almost entirely at some AI data center facilities.
  4. 4Nvidia's DSX uses closed-loop liquid cooling with coolant entering at 45°C, 13°C warmer than the 32°C average for closed-loop systems in 2024 recorded by Uptime Institute.
  5. 5There is a direct trade-off between water use and energy use in cooling, according to independent researcher Andy Masley.
  6. 6AI chip temperatures can exceed 80°C (176°F), and simple fans circulating air are often enough in DSX deployments instead of year-round chilled air.
DSX coolant inlet temperature
45°C +13°C vs 2024 closed-loop average

Warmer coolant reduces need for year-round chilled air, relying on simple fans in many deployments

Analysis

For AI engineers and infrastructure teams, cooling is becoming a hard constraint on scaling. Nvidia's DSX system changes the thermal equation by letting coolant enter servers at 45°C—13°C warmer than the 2024 industry average—so AI clusters can shed heat with simple fans instead of chilled water. That could unlock larger training deployments in water-constrained regions, but the energy penalty deserves close scrutiny.

As pressure mounts over the water appetite of AI data centers, tech giants are signaling that the cooling problem is solvable. Nvidia claims in a June report that its DSX design and management system can nearly eliminate water consumption at some facilities. This comes amid public anger in the United States over data centers' demand for water and power. Rystad Energy reports data centers consumed 222 billion liters (59 billion gallons) for cooling in 2025; without adaptive measures that could nearly triple to 644 billion liters by 2030, though targeted steps could keep growth under a doubling.

Nvidia's design allows coolant to enter at 45°C, much warmer than the 32°C average Uptime Institute observed for closed-loop systems in 2024.

The technique behind Nvidia's claim is a closed-loop liquid cooling system that circulates coolant directly through servers as close as possible to chips. Chips can run above 80°C (176°F). Nvidia's design allows coolant to enter at 45°C, much warmer than the 32°C average Uptime Institute observed for closed-loop systems in 2024. Because the liquid is already warm, facilities do not need to chill it aggressively; Nvidia's Josh Parker says simple fans circulating air are often enough, though some deployments require a mix of methods.

This is a significant engineering shift, but it exposes a fundamental tension: water and energy are not independent cooling inputs. Andy Masley, an independent researcher covering AI and data centers, notes a direct trade-off: cutting water use usually increases electricity demand because the sealed coolant must still be cooled, often using air blown over it. Thus a data center may reduce water withdrawals while raising power consumption, a dynamic that matters for grids already strained by AI expansion. In this sense, DSX does not eliminate environmental impact so much as relocate it from water to electricity, although the source does not quantify the energy penalty.

Market implications are substantial. Nvidia frames DSX not merely as server hardware but as a system for designing and managing AI data centers, which could deepen its control over the AI infrastructure stack. Hyperscalers and colocation operators facing local water permits, drought restrictions, and community opposition may adopt the technology to clear regulatory and reputational hurdles. If Nvidia's claims hold in real-world deployments, it could set a new baseline for sustainable AI infrastructure, pressuring rivals in liquid cooling and data center design to match 45°C coolant temperatures.

There is reason for caution. Nvidia's near-zero water claim is qualified: it applies to "some facilities" and depends on climate, electricity access, and workload patterns. The Uptime Institute's 2024 baseline is only a reference point, and no independent third-party validation is cited in the AP report. The industry is under mounting pressure to make good on such claims, reflecting skepticism after years of bold sustainability pledges. Moreover, the water-energy trade-off means regulators and environmental groups may demand disclosure of total resource intensity, not just water data.

The public anger referenced points to specific US communities where data centers are sited near residential water supplies. Although the AP report does not detail local cases, it notes the industry is spending billions of dollars, and that spending is now tied not only to compute capacity but to social license to operate. Data centers are increasingly built in water-stressed regions because land and power are cheap, which amplifies conflict. A technology that reduces water withdrawals can ease permitting and legal challenges, even if it raises electricity load. Therefore the water claim has both engineering and political value.

What to Watch

For AI development specifically, data center cooling is no longer a back-office concern. As model training clusters scale to hundreds of thousands of accelerators, heat density increases and water-based cooling becomes harder to sustain. DSX's direct-to-chip approach is designed for high-density AI clusters, so its adoption could accelerate the deployment of larger training runs in constrained environments. However, the additional energy required for air-cooling the closed loop may collide with power availability and carbon goals. The result is a complex optimization problem: minimize total environmental footprint across water, carbon, and land, not single metrics.

Looking ahead, the next 12 to 24 months will reveal whether DSX-style systems can bend Rystad's curve toward the lower path. If adaptive measures keep water growth under a doubling, that would still imply more than 400 billion liters by 2030—a massive additional demand on regional watersheds. But if direct-to-chip warm-water cooling becomes standard, water consumption could decouple from AI compute growth. Energy planners, water utilities, and AI developers will need integrated metrics and independent audits. The sustainability of AI expansion now hinges on managing the water-energy nexus, and Nvidia's DSX is the most visible attempt to solve one side of it.

Timeline

Timeline

  1. Uptime Institute baseline for closed-loop cooling

  2. Global data center water use measured

  3. Nvidia releases DSX report

  4. AP report syndicated globally

Source cluster

Primary reporting

3articles

Cite This Page

"Nvidia DSX Runs Coolant at 45°C to Cut AI Data Center Water Use to Nearly Zero." AI Intelligence Brief, September 7, 2026. https://getaibrief.com/story/nvidia-dsx-45c-liquid-cooling-ai-data-centers

How we covered this story

Every story in our AI coverage is assembled from multiple primary sources, cross-referenced for factual consistency, and scored along three independent dimensions: sentiment, operational impact, and source-cluster confidence. Single-source rumors and unverifiable claims do not pass our editorial gate. When a story shows "Verified by N sources" with N≥2, the development is independently corroborated; when N=1, we mark it explicitly so readers can weigh the signal accordingly.

Impact scoring uses a 1-10 scale weighted toward regulatory, financial, and operational consequence rather than coverage volume. A topic that runs in every outlet but moves no real decisions ranks lower than a niche regulatory filing that reshapes how operators in the AI space have to behave. Read our full methodology for the scoring rubric, our glossary for term definitions, and our trends index for the longitudinal view across the beat.

Sources are only linked to a story once they clear our classification pipeline at a minimum 35 percent relevance threshold. According to that methodology, reviewed July 2026, this follows multi-source corroboration standards recommended by journalism research bodies such as the Reuters Institute for the Study of Journalism.

See something wrong in this story — a wrong fact, a broken source link, a misattributed entity? Report a data issue.