News · 2026-06-25
NVIDIA's warm-water fix for AI's thirsty data centers
The AI boom has a water problem, and it's more literal than people expect. Big data centers full of hot chips have traditionally been cooled the way a swamp cooler works, by evaporating enormous amounts of water, often millions of gallons a year for a single large facility. As AI compute explodes, so does that thirst, which has turned into a real public-relations and environmental headache. NVIDIA has now proposed a design to make most of that water disappear, detailed on its official blog.
The core idea is to cool the chips with liquid instead of air, and crucially, to do it with warm liquid. That sounds backwards, why would you cool something with warm water? The insight is about where the heat ends up. In NVIDIA's design, coolant runs right up against every chip in a sealed loop and carries the heat away. Because the system is engineered to work even when that coolant is fairly warm, warmer than a hot tub, the heat it's carrying is hot enough to be dumped straight into the outside air using simple radiators, the same principle as the radiator in a car, for most of the year. That matters because the water-guzzling part of traditional cooling is the evaporation step used to chill things down. If you can reject the heat to the open air instead, you can skip the evaporation, and skip the water.
The payoff NVIDIA claims is dramatic: a closed loop that recirculates the same liquid and consumes essentially no new water for cooling the chips, down from the millions of gallons a comparable conventional facility would evaporate. There's an energy bonus too. Cooling can eat up a large share, by some measures close to half, of a data center's total electricity, and running the system warm means you can switch off the power-hungry chillers for much of the year, in favorable climates. Less chilling means less water and less power at the same time.
Why it matters: the environmental footprint of AI has become a competitive battleground, not just an activist talking point, and NVIDIA is positioning itself as the company with a sustainable answer, the vendor that builds not just the chips but the blueprint for the building they sit in. As AI data centers multiply, a design that genuinely cuts on-site water use at scale is a real selling point to operators and to the communities, and regulators, deciding whether to let these facilities be built nearby.
The honest caveat is one the critics raised immediately, and it's a good one. Both TechCrunch and Fortune pointed out that eliminating the water used inside the data center doesn't eliminate the water used to make the electricity that powers it. A lot of that power still comes from plants that themselves consume large amounts of water for their own cooling, water that doesn't show up on the data center's books but is part of AI's true footprint. "Zero cooling water" is a real and useful efficiency win, narrowly scoped. It is not the same as "zero water," and the bigger, system-wide question of AI's energy and water appetite remains very much open.