California’s only remaining nuclear power plant is turning to artificial intelligence to boost efficiency, sparking both interest and concern from lawmakers and watchdogs.
The Diablo Canyon nuclear facility, located in San Luis Obispo on the central coast, has started using AI to streamline document management. While the technology is currently limited to document retrieval, critics worry it may open the door to wider automation in a highly sensitive sector.
Diablo Canyon is scheduled to begin shutting down in 2029 after years of gradual decline. Still, the plant remains vital to California’s power grid. It supplies nearly 9% of the state’s total electricity and 17% of its carbon-free energy.
Last year, plant operator Pacific Gas & Electric (PG&E) partnered with a local startup, Atomic Canyon, to install eight NVIDIA H100 processors—some of the most powerful AI chips available. The processors run Atomic Canyon’s AI platform, called Neutron Enterprise, which is designed to help staff quickly find key technical documents.
“Nuclear power plants are required by law to manage billions of pages of documents,” PG&E said in a press release. “These records are spread across many systems, and retrieving them takes significant time and resources.” The company says its AI solution can reduce search times from hours to minutes, improving safety and collaboration.
Maureen Zawalick, a PG&E vice president, described the initiative as a major step forward. “We’re proud to be the first nuclear plant to use Neutron Enterprise on the NVIDIA platform,” she said. “Faster data access means more time spent ensuring safe, clean, and affordable energy.”
Despite the limited scope of AI’s current use, lawmakers and advocacy groups are sounding alarms. They want strict guidelines to govern the future use of AI in nuclear facilities. But progress on federal regulation remains uncertain. Former President Trump recently rolled back a Biden-era executive order aimed at managing AI risks, calling it a barrier to innovation.
Some experts are skeptical of PG&E’s assurances. “The idea that you can limit generative AI to just one task at a nuclear plant seems naive,” said Tamara Kneese, director of the Climate, Technology, and Justice program at nonprofit Data & Society. “There needs to be more scrutiny over how this technology is used in such a high-stakes environment.”
As AI continues to grow, so does its demand for energy. Data centers require vast amounts of power, putting pressure on energy systems. At the same time, nuclear energy is being promoted by tech leaders like Bill Gates and Sam Altman as a clean solution to this rising demand. Some envision a future where AI and nuclear power work hand in hand, with machine learning helping to design and operate next-generation power plants more efficiently.
The federal government appears to support this vision. The Department of Energy recently identified 16 federal sites as ideal locations for new data centers. These areas already have strong energy infrastructure and could be fast-tracked for new power generation, including nuclear.
Related Topics:
- Traders Eager to Reenter Russian Crude Market, But the Door Remains Closed
- Oil Prices Rise Due to Venezuela Tariffs
- Iraq Gives BP Final Approval for Kirkuk Oil Development