惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
V
Visual Studio Blog
IT之家
IT之家
博客园 - 聂微东
The Cloudflare Blog
月光博客
月光博客
阮一峰的网络日志
阮一峰的网络日志
S
SegmentFault 最新的问题
Apple Machine Learning Research
Apple Machine Learning Research
酷 壳 – CoolShell
酷 壳 – CoolShell
爱范儿
爱范儿
H
Help Net Security
博客园 - 叶小钗
V
V2EX
WordPress大学
WordPress大学
J
Java Code Geeks
Hugging Face - Blog
Hugging Face - Blog
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
博客园_首页
C
Check Point Blog
B
Blog
D
DataBreaches.Net
美团技术团队
罗磊的独立博客

TechCrunch

Robots beat human records at Beijing half-marathon Palantir posts mini-manifesto denouncing inclusivity and ‘regressive’ cultures TechCrunch Mobility: Uber enters its assetmaxxing era Cracks are starting to form on fusion energy’s funding boom Blue Origin successfully re-uses a New Glenn rocket for the first time ever Tesla brings its robotaxi service to Dallas and Houston VC Ron Conway says he has a ‘rare form of cancer’ AI chip startup Cerebras files for IPO Anthropic’s relationship with the Trump administration seems to be thawing The App Store is booming again, and AI may be why “Tokenmaxxing” is making developers less productive than they think Hackers are abusing unpatched Windows security flaws to hack into organizations Zoom teams up with World to verify humans in meetings Gigs turns your concert history into a personal live music archive Chef Robotics escaped the robot cooking graveyard and says it’s thriving — here’s why Uber will now pick up your returns from your doorstep Anthropic launches Claude Design, a new product for creating quick visuals Google’s AI Mode can now help you find products in stock nearby Bluesky confirms DDoS attack is cause of continued app outages Bluesky confirms DDoS attack is cause of continued app outages Netflix plans to add a vertical video feed, use AI for recommendations SaySo is a new short-form video app that aims to restore users’ trust in news Loop raises $95M to build supply chain AI that predicts disruptions Are we tokenmaxxing our way to nowhere? New leaders, new fund: Sequoia has raised $7B to expand its AI bets Netflix co-founder and chair Reed Hastings to leave board Upscale AI in talks to raise at $2B valuation, says report Physical Intelligence, a hot robotics startup, says its new robot brain can figure out tasks it was never taught From the Startup Battlefield stage to the International Space Station: geCKo Materials built a sticky product Slash, a Ramp competitor founded by teenagers, raises $100M at $1.4B valuation
Nvidia wants to cut data center water use, but that's not...
Tim De Chant · 2026-06-23 · via TechCrunch

Nvidia just announced a warm-water cooling system that it says can dramatically reduce the amount of water a data center uses — eliminating “pretty much all water usage” inside the data center, according to an Nvidia executive in a press release.  

“The water consumption challenge for data centers is largely solved,” Josh Parker, chief sustainability officer at Nvidia, recently told Axios.

But that’s only part of the water story. As long as AI data centers run on fossil fuels — a choice tech companies are increasingly making — the savings stop at the data center’s walls.

The core issue is how Nvidia measures data center water use. According to its blog post, the company essentially draws a line around the data center. Anything inside gets counted, and anything outside gets ignored.

To be fair, Nvidia’s system does appear to deliver on its facility-level promise — the coolant runs in a closed loop, filled once and recirculated for the life of the facility, meaning no new water is consumed to cool the chips. In favorable climates, the company says, that can amount to a 100% reduction in on-site water use.

TechCrunch has asked Nvidia to clarify the matter, and we’ll update this article if we receive a reply.

The problem is, water use outside of the data center — primarily in electricity generation and chip manufacturing — can double or triple the total water footprint of a facility. That means Nvidia’s solution addresses about a quarter to a third of AI data centers’ total water consumption. 

The new system is clever, pumping coolant into racks at 45°C (113°F). That’s hot for humans but not for computer chips. After passing through a server, the coolant emerges at 55°C (131°F), Nvidia said, bringing a significant amount of heat away from the hardware. 

At that temperature, the outside air in most climates can draw heat off passive radiators without evaporative cooling or, in some cases, fans. A data center without fans or chillers would not only use less water, but it would also be more efficient and quieter. 

But no data center can operate without an electricity supply, and many types of power plants are themselves major water consumers.

Fossil fuel power plants are one of the largest water users in the U.S., consuming 2.7 billion gallons per day, according to the U.S. Geological Survey — most of it for evaporative cooling. Natural gas power plants use 1.17 liters of water for every kilowatt-hour of electricity they generate, according to a recent study. Coal plants are even more water-intensive, using 2.2 liters per kilowatt-hour. Fossil fuel power plants collectively generate about half of all data center power today, according to the IEA.

Hydropower dams, which supply around 10% of data center power, don’t consume water in the same direct way, but evaporation from their reservoirs amounts to 6.8 liters lost per kilowatt-hour generated. Geothermal, a source that tech companies are starting to explore, varies widely — it can be higher or lower depending on the specific technology. Some enhanced geothermal startups, like Fervo, have pledged to use mostly “degraded” water that would otherwise go unused.

Wind and solar power, on the other hand, use vanishingly small amounts of water, about 0.01 liters and 0.03 liters per kilowatt-hour, respectively — figures that include the water needed for manufacturing and cleaning solar panels. 

While renewables are providing a growing share of new electricity capacity, natural gas and coal are expected to provide more than 40% of new electricity needed to meet data center demand through 2030, the IEA projects. Without major changes to that trajectory, data centers will still consume large amounts of water, regardless of what Nvidia does inside its walls.

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Tim De Chant is a senior climate reporter at TechCrunch. He has written for a wide range of publications, including Wired magazine, the Chicago Tribune, Ars Technica, The Wire China, and NOVA Next, where he was founding editor.

De Chant is also a lecturer in MIT’s Graduate Program in Science Writing, and he was awarded a Knight Science Journalism Fellowship at MIT in 2018, during which time he studied climate technologies and explored new business models for journalism. He received his PhD in environmental science, policy, and management from the University of California, Berkeley, and his BA degree in environmental studies, English, and biology from St. Olaf College.

You can contact or verify outreach from Tim by emailing tim.dechant@techcrunch.com.

View Bio