In an effort to keep up with AI’s energy demands and keep emissions down, AI companies are racing to build up nuclear power capacity. NVIDIA is investing in a nuclear company founded by Bill Gates that is building the first commercial Natrium reactor at the site of a retiring coal plant in Wyoming. The project is an example of a small nuclear reactor that may serve as a model for clean energy growth.

Google is both investing in and has agreed to purchase electricity from a company developing a first-of-its-kind nuclear fusion power plant. Commonwealth Fusion Systems says it will deliver power to the grid by the early 2030s from a facility close to “data center alley” in northern Virginia. Commercial nuclear fusion is still an unproven technology that faces significant hurdles to scale. The upside, if it is achievable, is virtually limitless clean energy.

Nuclear ambitions aside, the reality on the ground of AI data center energy consumption is that natural gas is the default energy source to meet the exploding demand. If companies make a concerted effort to shift to renewables or nuclear, the US could be on a path to overbuild natural gas capacity, saddling regions with unneeded and polluting fossil fuel plants and soaring electricity bills for residents. More transparency is needed around how flexible companies can be in their energy demands.

Questions to consider

  1. How flexible are companies in their use of different energy sources? What options are they exploring to meet projected demand? 

  2. How are companies assessing the risks of investing in and adopting alternative or speculative energy sources? How are they mitigating the financial risks if the projects do not deliver? How are they assessing the environmental and social risks? Are they consulting affected stakeholders?

Keep Reading