Nvidia is pushing new ways to power and cool the data centers behind artificial intelligence, as the environmental cost of AI becomes harder to ignore. As AI reshapes everything from scientific research to everyday computing, according to reporting via MSN, its rapid growth carries a rising and increasingly visible footprint.
At the center of the effort is a new liquid cooling system. According to The Times of India, the technology aims to cut the amount of water AI data centers consume. Reporting from Newsbytes adds that the system is designed to reduce energy use and recycle water inside these facilities.
But experts are urging caution. According to The Times of India and Newsbytes, the cooling system will not address AI's full water footprint, meaning the technology tackles only part of a much larger resource problem.
Nvidia is also exploring how these centers get their power. According to Newsbytes, Nvidia and a company called Valar Atomics are developing a Utah-based, AI-focused data center powered by a helium-cooled microreactor. The two companies have demonstrated the microreactor running Nvidia's Blackwell chips, with the stated goal of reducing water consumption in computing facilities.
Together, the two efforts point to a strategy of attacking AI's resource demands from both ends: cooling the hardware more efficiently and rethinking where the electricity comes from.
Why it matters: AI's explosive growth is straining water and energy supplies, and how the industry's biggest chipmaker chooses to manage that footprint will shape the environmental impact of the technology for years to come.