Companies that have dug deep into artificial intelligence have fallen in love with nuclear power for its promise of reliable, around-the-clock electricity. Major tech firms including Meta, Google, and Microsoft have all made significant deals with nuclear startups and reactor operators. But does the nuclear industry love AI back? The answer is yes, but with some caveats.
While no one is proposing to let an AI run a reactor, power companies are increasingly interested in the technology’s potential to improve operations on the business side. This is according to Bradley Fox, co-founder and CEO of Nuclearn. Fox and Jerrold Vincent started the company to capitalize on that growing interest. Nuclearn says its AI tools are now being used in more than 65 nuclear reactors around the world. The startup recently raised a $10.5 million Series A funding round led by Blue Bear Capital, with participation from AZ-VC, Nucleation Capital, and SJF Ventures.
Nuclearn began when its founders were working at the Palo Verde Nuclear Generating Station just west of Phoenix. They had been experimenting with ways to streamline various repetitive tasks, starting from a data science perspective and then progressing to more advanced AI models. Soon, other reactor operators took notice and began asking if the same solutions could be applied to their plants.
This surge of interest coincided with the COVID pandemic. With extra time after work, the founders decided to focus on building a startup. Nuclearn has since developed AI models trained on nuclear industry-specific terminology. The startup can train custom models for utilities and power providers that request it. While its software primarily runs in the cloud, it can also help reactors set up on-site hardware if their security protocols require it.
The company’s software can generate routine documentation that reactor employees then review and sign off on. Most AI in the industry is currently considered a tool by the Nuclear Regulatory Commission, similar to using Excel or other engineering software. Liability always remains with a person, not the software.
Reactor operators can set thresholds for how much gets automated based on their comfort level and their confidence in the model’s ability to handle the task. If the model is unsure of something, it will send the task back to the right people for a double check. Nuclearn tells its customers to think of the AI as a junior employee.