Diablo Canyon, California’s sole remaining nuclear power plant, has adopted generative AI to tackle its massive paperwork burden. The tool, developed in collaboration with AI startup Atomic Canyon, aims to streamline document retrieval and reduce administrative overhead as the plant navigates extended operations through 2030.
Regulatory Demands Drive Innovation
In 2022, Governor Gavin Newsom signed SB 846, extending Diablo Canyon’s operational life beyond its planned 2025 closure. The decision came amid energy reliability concerns and California’s push to maintain its nuclear capacity. However, the extension brought heightened regulatory demands, requiring the plant’s operator, Pacific Gas and Electric (PG&E), to manage billions of data points across six systems.
"We had to pivot to continue to operate," said Maureen Zawalick, PG&E’s senior vice president and chief risk officer.
The AI tool, named Neutron, enables workers to search millions of documents instantly, including engineering records, design updates, and regulatory filings. Unlike cloud-based solutions, Neutron connects directly to the plant’s systems via the internet, ensuring secure, on-site accessibility.
Industry-Specific AI Development
Atomic Canyon CEO Trey Lauderdale, who previously worked in healthcare, saw parallels between the highly regulated nuclear sector and his past experience. His team trained Neutron using 53 million pages of publicly available nuclear data from the Nuclear Regulatory Commission. The project also leveraged Oak Ridge National Laboratory’s Frontier supercomputer to refine the AI model’s understanding of nuclear terminology.
Since its implementation, Neutron has helped PG&E employees efficiently retrieve critical documents, including over 9,000 operating and maintenance procedures. The tool’s success marks a milestone in the integration of AI within the nuclear energy sector, offering a model for other plants facing similar administrative challenges.