New partnership touts nuclear-powered AI with minimal water use
Valar Atomics and Nvidia have announced a partnership to explore developing a nearly waterless AI data center in Orangeville, Utah, powered by Valar's advanced small modular nuclear reactor. This collaboration marks the first time advanced nuclear energy will directly power AI infrastructure, aiming to address concerns about electricity and water consumption while creating local jobs and revenue. The project is currently in the exploration phase and requires licensing from the Nuclear Regulatory Commission.
Valar Atomics, an advanced nuclear power startup, has partnered with Nvidia to explore the development of an artificial intelligence data center in Orangeville, Utah, marking the industry's first direct application of advanced nuclear energy to AI infrastructure. The proposed 30-megawatt facility would utilize Valar's Ward250 small modular reactor, which recently achieved a milestone by becoming the first startup-built reactor to generate nuclear power.
This innovative project aims to address growing concerns about data center resource consumption. Valar's Ward250, with its helium cooling system, combined with Nvidia's new water-efficient direct liquid cooling technology (DSX), is designed to operate with minimal water use. According to Max Ukropina, head of projects at Valar Atomics, this approach offers "concrete reassurance" to local families concerned about utility bills and environmental impact, promising high-skill jobs and tax revenue without straining local water supplies or the broader electrical grid.
The partnership was announced at Valar's Orangeville site, attended by employees, investors, and government officials including Utah's Office of Energy Development Director Emy Lesofski and State Senator David Hinkins, who lauded the project's potential for job creation and economic growth in Emery County. Before commercial operation, the Ward250 reactor will require licensing from the Nuclear Regulatory Commission, which is reportedly working to expedite review times.