Skip to content
ATAI Today Brief
HomeNewsConceptsGuidesToolbox
AboutSubscribeUA
Subscribe

AI Today Brief

The daily AI-engineering brief. Built in public. EN · UA.

XTelegramLinkedInYouTubeRSS
NewsConceptsGuidesSubscribeAdvertiseAboutEditorial policyAI disclosurePrivacyTerms

© 2026 AI Today Brief. All rights reserved.

  1. Home/
  2. News/
  3. Token & cost optimization/
  4. Nvidia Warm-Water Cooling Reduces Data Center Facility Water Usage
Token & cost optimization

Nvidia Warm-Water Cooling Reduces Data Center Facility Water Usage

June 23, 2026· 4 min read
OKCurated by Oleksandr Kuzmenko, AI Product Engineer·Updated June 23, 2026·Sources cited on every story
AI-assisted · editor-reviewed·How we use AI
Nvidia Warm-Water Cooling Reduces Data Center Facility Water Usage

Nvidia introduced a warm-water cooling system operating at 45°C to 55°C, claiming to eliminate evaporative cooling needs. While it optimizes facility-level water usage, it does not account for indirect water consumption from power generation.

Impact: Medium

Why it matters

Understanding the true resource cost of AI compute helps architects better evaluate vendor sustainability claims and infrastructure overheads.

TL;DR

  • 01Closed-loop systems eliminate internal water use for cooling.
  • 02Facility-level efficiency ignores the significant water cost of power generation.
  • 03Infrastructure sustainability is tied to energy source, not just hardware cooling.
  • 04Fossil fuel power plants remain the largest indirect water consumers.

Key facts

Coolant Input Temperature45°C
Coolant Output Temperature55°C
Coolant Input Temperature
45°C
Coolant Output Temperature
55°C
Coal Plant Water Usage
2.2 liters/kWh
Natural Gas Water Usage
1.17 liters/kWh

Facility-Level Cooling Innovations

Nvidia’s system utilizes a closed-loop coolant cycle. Because the coolant enters the racks at 45°C and exits at 55°C, the temperature differential is sufficient to reject heat using passive radiators. In optimized environments, this eliminates the need for traditional chillers and fans, which typically account for the bulk of internal data center water usage.

The Hidden Water Footprint

Despite facility-level gains, the data center remains dependent on the electrical grid. Current energy sources include:

  • Natural gas plants: 1.17 liters per kWh
  • Coal plants: 2.2 liters per kWh
  • Hydropower reservoir evaporation: 6.8 liters per kWh
  • Wind/Solar: 0.01–0.03 liters per kWh (manufacturing/cleaning)

Infrastructure Outlook

As data centers continue to scale, reliance on fossil fuels—projected to provide over 40% of new electricity for data center demand through 2030—means the total water consumption will likely remain high despite internal cooling efficiency improvements.

✓ When to use

  • Designing hyperscale or edge data center facilities
  • Assessing hardware sustainability benchmarks
  • Planning long-term energy infrastructure needs

What to do today

  • →Review data center energy procurement strategies.
  • →Account for indirect water footprints in ESG reporting.
  • →Evaluate passive cooling potential for future builds.
#Nvidia Cooling System

Sources

  • Nvidia wants to cut data center water use, but that's not the same as fixing AI's water problem
ShareShare on XShare on LinkedIn

Related stories

  • Token & cost optimizationHow to Optimize Claude Session Limits and Avoid Context Bloat
  • Token & cost optimizationHow a Compiler Loop Unroller Generated 256KB of Code to Initialize 64KB

Email digest

Get the morning AI brief

One email a day — the stories that matter for engineers, founders and tech leads. Human-edited, with links to primary sources.

  • ✓120+ sources scanned daily
  • ✓Edited by a human
  • ✓1 email per day
  • ✓EN + UA

By subscribing you agree to the privacy policy.