Abstract: The supercritical carbon dioxide power cycle, with its high efficiency, compactness and wide adaptability to heat sources, is regarded as a strong contender for the next generation of advanced power conversion technologies. The research progress of AI technology in the field of sCO₂ power cycle was systematically reviewed, and a technical classification framework was constructed from two dimensions: "scientific computing acceleration" and "engineering research efficiency improvement". At the scientific computing level, the application of machine learning in thermal property prediction, heat transfer correlation formula construction, CFD proxy models and flow field reconstruction was summarized; at the engineering application level, the research status of AI in system design and topology optimization, advanced control strategies, fault diagnosis and digital twin was analyzed. Based on this, key methodological approaches such as "mechanism-data fusion modeling" and "large-small model collaboration" were proposed, and the current challenges such as data scarcity, model generalization and interpretability, as well as future research directions were discussed. The study shows that AI technology is expected to reshape the research paradigm of sCO₂ power cycle, shifting from "experience-driven" to "data-mechanism dual-driven".