According to sponsored research from ScottMadden published by Utility Dive, the U.S. nuclear fleet has moved past the question of whether to use AI. The firm identified more than 60 deployments spanning search, engineering support, corrective action, outage planning, and equipment reliability. Yet roughly 60% of those are in active use without benefits being effectively quantified — time savings and quality gains have not been translated into labor savings, outage performance, or risk reduction.
ScottMadden distinguishes three AI types: generative AI for knowledge work, machine learning for repetitive decisions and prediction, and specialized AI for engineering and operational optimization. The last category, which combines AI with physics-based models or computer vision, requires the deepest integration and scarce talent, but also shows the strongest evidence of sustained value — core optimization has been deployed fleetwide across multiple operating cycles.
Utilities are taking multiple adoption paths, including enterprise platform rollouts, targeted vendor deployments, and citizen development by plant engineers. The deciding factors in programs that scale are organizational rather than technical: one operator saw consistent use of an AI observation program built into a required process, while an optional AI review tool saw uneven use. The report recommends focusing on desired outcomes, challenging existing processes, addressing integration early, matching governance to consequence, and measuring operating value rather than usage counts.