The U.S. Department of Energy (DOE), working alongside Idaho National Laboratory (INL), Argonne National Lab (ANL), Microsoft, and Everstar, has demonstrated how artificial intelligence can transform the nuclear licensing process. In this collaboration, AI-driven mapping tools were applied to convert a safety analysis document prepared under DOE’s authorization pathway for advanced reactor demonstrations into U.S. Nuclear Regulatory Commission (NRC) licensing documents suitable for commercial deployment. The outcome highlights how digital tools can enhance both efficiency and precision in nuclear licensing, with the potential to significantly accelerate timelines for bringing advanced nuclear reactors to market.
At the center of the initiative is Everstar’s Gordian AI solution, developed on the Microsoft Azure platform. The system was used to transform the Preliminary Documented Safety Analysis for DOE’s National Reactor Innovation Center’s (NRIC) Generic High Temperature Gas Reactor (HTGR) into sections aligned with an NRC license application. The resulting 208-page document was completed in just one day, a process that would typically require four to six weeks of work by a team. In addition to generating content, the tool flagged missing or incomplete information critical to a successful nuclear licensing submission. Designed for nuclear-grade technical applications, Gordian integrates physics and engineering capabilities with semantic ontology mapping, ensuring outputs are computed and verified rather than inferred. While the technology accelerates nuclear licensing workflows, expert oversight remains essential, following a model where specialists design, AI accelerates, and experts validate. The generated output was subsequently reviewed by an expert for accuracy, completeness, consistency, grammar, and structure, confirming that it met rigorous professional standards while also identifying gaps in its own dataset.
The broader nuclear licensing environment has traditionally relied on repeated manual reviews and incremental adjustments, often extending over several years. This HTGR case adds to a growing body of examples demonstrating how AI can streamline these processes. Earlier this year, INL and Microsoft deployed a Microsoft Azure AI-based solution to illustrate how advanced models can produce engineering and safety analysis reports, a core component of applications for construction permits and operating licenses for nuclear power plants. According to a recent NRIC study, AI could reduce document preparation time and regulatory review cycles by as much as 50 percent, while also improving consistency, traceability, and overall accuracy in nuclear licensing.
Looking forward, the participating organizations plan to further validate and refine their approach. A reviewing agent will assess AI-generated documents against NRC guidance to confirm readiness for submission, while a benchmarking rubric is being developed to assign a confidence grade to Gordian’s performance. INL is also advancing in-house AI tools, including potential applications for fuel fabrication facilities. The effort aligns with President Trump’s Genesis Mission, which aims to accelerate innovation through AI. As part of this initiative, DOE has announced $293 million in competitive funding to address twenty-six national science and technology challenges, including efforts focused on speeding up nuclear energy deployment through advancements in nuclear licensing.
