One of the biggest criticisms of the artificial intelligence boom has little to do with chips, software or valuations. It’s water.
As tech companies build increasingly power-hungry AI data centers, critics have warned that the water needed to cool those facilities could strain local supplies, especially in drought-prone regions.
But Nvidia Corp. (NASDAQ:NVDA) and Elon Musk say those concerns may be overstated.
The Tesla, Inc. (NASDAQ:TSLA) and Space Exploration Technologies Corp. (NASDAQ:SPCX) CEO responded with a simple “True” to an Nvidia post highlighting new data showing that data centers account for 0.2% of daily U.S. water consumption and that advances in liquid cooling technology are dramatically reducing water requirements even further.
But as TechCrunch points out, the data doesn’t address water usage outside the data center (i.e., electricity generation and chip manufacturing). That increases a facility’s total water footprint.
Nvidia’s Case Against The AI Water Narrative
At the center of Nvidia’s argument is its new Rubin AI platform, which the company says is the first generation of its AI infrastructure to operate using a fully liquid-cooled architecture.
Unlike traditional facilities that rely heavily on cooling towers and large volumes of evaporated water to remove heat, Nvidia’s latest systems use a closed-loop liquid cooling design that circulates coolant directly through servers.
According to Nvidia, the approach allows facilities in favorable climates to replace conventional cooling towers with dry coolers, reducing cooling-related water consumption from roughly 2.6 million gallons per megawatt annually to nearly zero.
The company says the system can operate with coolant temperatures as high as 45 degrees Celsius, enabling data centers to reject heat more efficiently while reducing reliance on energy-intensive cooling equipment.
$4 Million In Annual Savings
The potential benefits extend beyond water conservation. Nvidia estimates a 50-megawatt hyperscale data center could save more than $4 million annually in combined cooling, energy and water expenses by adopting liquid-cooled infrastructure.
Cooling has historically represented one of the largest operating expenses for data centers, accounting for as much as 40% of total electricity consumption in some facilities. By capturing heat directly at the chip level, Nvidia says operators can significantly reduce both energy use and physical infrastructure requirements.
Why Investors Should Care
The debate over AI infrastructure has increasingly shifted from computing power to resource consumption. Investors have spent months evaluating whether electricity demand, water availability and environmental concerns could become obstacles to the industry’s rapid expansion.
Nvidia’s latest claims suggest at least one of those concerns may be less severe than many critics believe. If the company’s liquid-cooling approach delivers the efficiency gains it describes, AI data centers could become both cheaper to operate and less dependent on local water resources.
Musk’s endorsement may have consisted of a single word, but, again, Nvidia’s only addressing about a quarter to a third of AI data centers’ total water consumption.
Musk has also faced backlash tied to AI infrastructure in Mississippi, where he’s facing a proposed class-action lawsuit alleging noise pollution from AI data center operations.
Photo courtesy: Shutterstock
Login to comment