Abstract: In new power system dominated by renewable energy, pumped storage faces challenges in construction cost allocation, value realization, and cost recovery due to imperfect market mechanisms. This paper proposes a pumped storage capacity optimization method considering multiple revenue pathways. By accounting for uncertainties in service demand and compensation prices, the value of pumped storage in capacity leasing, peak-cutting and valley-filling, and frequency modulation is quantitatively evaluated. A multi-objective optimization model is established with the objectives of maximizing investment benefits and minimizing net load fluctuation variance. The improved NSGA-II algorithm is used to solve the model that incorporates adaptive scaling factor and normal distribution crossover operator to maintain population diversity. Case studies based on multiple scenarios reveal that 160 MW capacity proves optimally. After considering the benefit of auxiliary service, the cost recovery period is shortened by 2 years, while the improved NSGA-II markedly enhances Pareto solution quality and search capability. This method provides an effective approach to addressing cost recovery and profitability for pumped storage, thereby promoting investment in such projects.