High-temperature electrolysis (HTE) systems are expected to operate with renewable sources as energy storage devices due to their high efficiency, reversibility and eco-friendliness. However, the high capital costs of stacks and heat management devices make capacity optimization of HTE systems necessary in the design phase. The current literature formulates the problem merely under constant loading conditions, which ignores the fact that dynamic operation is essential for renewable energy storage. Under this background, we propose a novel model to incorporate the capacity optimization of HTE systems under volatile loading conditions. Specially, this model is formulated in the form of two-stage stochastic programming in which operation optimization is treated as a subproblem of capacity optimization. The advantages include its ability to analyze the influence of different operating strategies on device capacity and its “min–max” transformation, which is convenient to solve by commercial code. In a case study, we discuss the influences of power source volatility and operating strategy on the optimal capacities. Different changing trends of the optimal device capacities with power source volatility are observed around one particular hydrogen price of approximately 4.2$/kg, above which the capacities increase with power input volatility; at prices below 4.2$/kg, the opposite effect occurs due to renewable spillage. In addition, considering the operating strategy, we find that an optimal stack inlet temperature of 1200K obtains economic results comparable to those in a variant temperature operation and provides useful guidance on the controller design. Compared to previous capacity optimization studies under constant operation, this proposed method shows better system economy; e.g., the net revenue increases by approximately 29.54% and 71.52% for the HTE system under hydro and wind power inputs respectively.