A new Nvidia breakthrough may save billions of gallons of water, but not everywhere

The artificial intelligence boom has created a new problem that many people rarely think about: water. Massive data centers powering AI systems consume huge amounts of electricity, and keeping their servers cool often requires significant amounts of water.

Now, Nvidia says it has developed a new cooling system that could dramatically reduce that demand. The company recently unveiled a warm-water cooling technology that it claims can eliminate nearly all water usage inside a data center.

How Nvidia’s new cooling system works

The new system relies on a closed-loop design that is filled once and then recirculates the same coolant for the life of the facility. According to Nvidia, the approach can reduce on-site water use by as much as 100 percent in favorable climates.

The coolant enters server racks at about 113 degrees Fahrenheit and leaves at roughly 131 degrees Fahrenheit. Because the liquid is already so warm, outside air in many places can remove the heat through passive radiators without the need for water-hungry cooling systems.

That could make future data centers quieter and more efficient while dramatically reducing their direct water consumption. NVIDIA’s chief sustainability officer, Josh Parker, recently said that the water challenge for data centers is largely solved.

The water problem does not end at the data center

The company’s claims, however, come with an important caveat. NVIDIA’s measurements focus mainly on what happens inside the data center and do not fully account for water consumed elsewhere in the AI ecosystem.

Electricity generation and chip manufacturing can add a substantial amount to a facility’s overall water footprint. Some estimates suggest these outside factors can double or even triple the total water use associated with a data center.

That means Nvidia’s breakthrough addresses only part of the issue. While the technology could slash water consumption at the facility itself, it cannot eliminate the hidden water demands created by the energy needed to run AI systems.

Fossil fuels remain a major challenge

Power generation is one of the biggest reasons why AI still carries a large water footprint. Many power plants rely on enormous amounts of water for cooling, especially facilities fueled by natural gas and coal.

According to the U.S. Geological Survey, water-using utility-scale thermoelectric power plants in the United States consumed about 2.7 billion gallons of water per day in 2015. A 2025 data center water-use review lists total water-consumption factors of about 1.17 liters per kilowatt-hour for natural gas generation and 2.2 liters per kilowatt-hour for coal.

The issue is particularly important because fossil fuels still provide around half of the electricity used by data centers today. Even if Nvidia’s cooling technology becomes widespread, data centers powered by fossil fuels will continue to indirectly consume large amounts of water.

Renewable energy paints a different picture

Not every energy source has the same water demands. Wind and solar power require only tiny amounts of water compared with traditional power plants, even after accounting for manufacturing and maintenance.

Hydropower presents a more complicated picture because reservoirs can lose significant amounts of water through evaporation. Geothermal energy also varies widely depending on the specific technology being used.

The difference between these energy sources means Nvidia’s innovation could have very different results depending on where a data center is located and how it gets its electricity. Facilities powered primarily by renewable energy could see much larger overall water savings.

Little-known fact: According to the International Energy Agency, a single large data center can consume as much electricity as 100,000 households, and some of the largest facilities under development could use even more.

AI’s growing appetite for resources

The timing of Nvidia’s announcement is significant because demand for AI computing continues to surge. Data centers are expected to require huge amounts of additional electricity in the coming years as companies race to build more AI infrastructure.

The International Energy Agency projects that natural gas and coal will still provide more than 40 percent of the new electricity needed to meet data center demand through 2030. If those projections hold true, indirect water consumption will remain a major challenge.

That means the future environmental impact of AI may depend on two separate transitions happening at the same time. More efficient cooling systems can reduce direct water use, but cleaner electricity sources are needed to tackle the broader problem.

A partial solution with big potential

NVIDIA’s new cooling system represents a meaningful technological step for an industry under growing scrutiny for its environmental impact. Eliminating nearly all on-site water use could save enormous amounts of water as more AI data centers are built around the world.

Still, the technology does not completely solve AI’s water problem. As long as many data centers continue to depend on electricity generated by fossil fuels, a large portion of their water footprint will remain hidden beyond the walls of the facilities themselves.

TL;DR;

• NVIDIA has introduced a warm-water cooling system that could nearly eliminate water use inside data centers.
• The system uses a closed-loop design that recirculates coolant instead of constantly consuming new water.
• Electricity generation and chip manufacturing still account for a large share of AI’s total water footprint.
• Fossil fuel power plants remain major water users and currently provide about half of the data center electricity.
• The biggest overall water savings may come when efficient cooling systems are paired with renewable energy.

This article was made with AI assistance and human editing.

Don’t forget to follow us for more exclusive content.

If you liked this, you might also like:

• Microsoft Chatbot can now design custom apps
• China targets long chatbot sessions with log-off nudges
• Perplexity Launches a Smarter Way to Research