Nvidia announced at London Climate Week that its next-generation AI infrastructure can be fully cooled with warm liquid, potentially reducing water consumption in data centers. The company’s chief sustainability officer, Josh Parker, claimed that “the water consumption challenge for data centers is largely solved” through a coolant system that operates at 113 degrees Fahrenheit, eliminating the need for additional chilling equipment.
Data centers have faced growing scrutiny over their energy and water usage as AI infrastructure expands. Tech companies including Google and Amazon have recently defended their water practices amid local opposition, with industry leaders arguing that efficiency gains will mitigate environmental impacts. Microsoft’s vice president of data center engineering suggested the technology could eliminate the need for mechanical chillers in most climates, even in hot regions like Arizona.
However, the technology faces significant limitations in addressing the broader water challenge. The new systems would take years to spread across the industry, existing data centers will continue using older cooling technologies, and costs remain unclear. Additionally, the technology only addresses water use within data centers themselves, not the substantial water required to produce electricity for AI infrastructure, particularly from fossil fuel power plants.
Efficiency improvements could accelerate AI infrastructure buildout rather than reduce overall environmental impact. Nvidia explicitly frames its gains as supporting continued growth in AI workloads, suggesting that efficiency gains may simply enable more expansion rather than solve underlying environmental concerns.
Sources:
