A modern numerical stochastic optimization method, namely the evolution strategy (ES), was applied to an ultra-low aspect ratio transonic turbine stator blade in order to seek a new aerodynamic design concept for lower secondary flow losses. The low stator blade count is selected to avoid the direct viscous interaction of the stator wake with the downstream rotor blade. This led to the ultra-low aspect ratio stator blade. In the optimization, two kinds of objective functions were used, that is, (1) minimization of the “aerodynamic loss” (a single objective), (2) minimization of the “aerodynamic loss” and of the “variation of circumferential static pressure distribution” downstream of the stator blade (multi-objective optimization). In the case of the single objective, the aerodynamic loss is improved by an extreme aft-loaded airfoil with a noticeable bent part near the trailing edge, although the circumferential static distribution is slightly worse than that of the baseline. In the case of the multi-objective optimization, we observe a trade-off relation between aerodynamic loss and variation of static pressure distribution which is not easily resolved. A new design concept to achieve lower aerodynamic loss for ultra-low aspect ratio transonic turbine stator blades is discussed.