Nvidia says its AI data center design runs hotter to use a lot less water
TL;DR
Nvidia is promoting a Rubin-generation reference design for fully liquid-cooled AI data centers that it says can cut water use to near zero. The core idea is to let servers run hotter, up to 45 degrees Celsius. Heat is captured at the chip and moved through liquid loops to outdoor dry coolers. Nvidia compares this with conventional cooling-tower systems that can use about 2.6 million gallons of water per megawatt per year. The missing piece is cost and how widely operators can actually deploy it.
Nauti's Take
This is technically interesting, but it is not a blank check for ever-larger AI factories. Water is the most visible pressure point, so near-zero water use is a powerful argument for operators.
Still, a more efficiently cooled data center can remain a huge consumer of electricity, land, and grid capacity. Nvidia is selling an important optimization, not a full answer to the AI infrastructure problem.
Briefingshow
AI data centers are facing more local resistance because communities see the water, power, and infrastructure impact up close. If Nvidia's design scales, it could meaningfully reduce one part of the footprint. But it shifts the debate rather than ends it: the bigger question is still how much energy these facilities need and who pays for grids, land, and cooling upgrades.
Sources