Abstract: With the breakthrough development of big data and computer technology, all industries are starting to promote the process of intelligence. As a key link in the nuclear energy industry chain, design optimization is becoming an important node in the deep integration of artificial intelligence technology and nuclear energy development. However, due to the bottleneck of safety constraint rigidity, multi-physical field strength coupling and expert experience dependence, the nuclear energy field is facing the status of highly intelligent demand and slow interweaving of engineering application development. In order to analyze the factors that cause the slow process of engineering landing in the nuclear energy field and further develop the key technical route to break through the existing bottleneck, design optimization was first classified into physical component optimization and functional process optimization according to the specific needs of design optimization in the nuclear field and the principle of intelligent optimization technology. Physical component optimization is static optimization. It is a process of selecting the optimal solution that meets the design requirements and constraints from the candidate design parameter space through mathematical modeling and algorithm search for the design stage of the system and equipment. Three major types of parameter optimization-discrete, continuous, and hybrid, were reviewed in this work. Heuristic algorithm, neural network and reinforcement learning are the key technologies driving their development functional process optimization, also known as dynamic optimization, is a dynamic programming solution problem composed of decision variables, state variables, objective functions, constraints, transfer functions, etc. The research and application cases including model predictive control and reinforcement learning were summarized in this study. At the same time, research cases of intelligent design optimization in other advanced industries were used as comparison and reference objects. Based on this analysis, the core technical obstacles of intelligent design optimization in the current nuclear energy field were identified, including the partially observable states, multi-objective optimization under safety constraints, and the conflict between high fidelity and computational efficiency of simulation models. Through the combination of technical obstacles and unique characteristics of nuclear energy, a transferable technology combination based on the characteristics of nuclear energy was finally summarized and formed. How to organically integrate these unrelated advanced technologies to adapt to the complex design environment and engineering requirements in the nuclear field will become the inevitable core issue in the follow-up research. This provides theoretical and technical guidance for nuclear energy design to break through the traditional experience-driven model.