Google and Kairos Power have announced a long-term deal for Google to purchase nuclear power from Kairos Power’s upcoming small modular reactors (SMRs). The goal is to add 500 MW of electricity to US power grids, with the intention of powering AI technology, such as Google’s own Gemini.
Kairos Power will construct an unspecified number of SMRs, which use a low-pressure system with molten-salt cooling. Google claims that this makes them safer and more cost-effective than other types of nuclear reactors. According to Jeff Olson, Kairos Power’s Vice President of Business Development & Finance, having an agreement for multiple deployments is crucial for accelerating the commercialization of advanced nuclear energy and demonstrating its technical and market viability.
Google presents nuclear energy as a “clean” alternative to carbon dioxide-producing options like coal-burning plants. However, this is not entirely accurate. SMRs still produce various types of waste, and a 2022 study led by Stanford researchers found that they will “exacerbate” existing problems with radioactive nuclear waste. A report on the US Nuclear Regulatory Commission’s website outlines the different types of waste produced by Kairos reactors and how the company plans to manage it. Additionally, research published in the Proceedings of the National Academy of Sciences found that molten salt-cooled SMRs, as well as sodium- and water-cooled SMRs, actually increase the volume of nuclear waste compared to other reactor types.
Despite these concerns, some may still see nuclear energy as a worthwhile option. When it comes to AI, energy experts have expressed concerns about the high power needs of AI and whether the US grid can keep up with demand. Adding new power sources, such as a nuclear power plant, to existing grids can be a slow process due to regulations. However, AI models from companies like Microsoft, Meta, and Google are already using significant amounts of energy.
As Robert Stoner, MIT researcher and director of the university’s Tata Center for Technology and Design, explained to PCMag, utilities are often resistant to adding renewable power sources to their systems. This can make it challenging to meet the energy demands of AI technology.
