In this study, a turbine cascade squealer tip is optimized by a multi-objective genetic algorithm (MOGA) with varying squealer heights and tip cooling configurations. The three objectives selected are the aerodynamic efficiency, the film cooling effectiveness, and the surface fluid temperature variance. The multi-scale methodology is implemented to reduce the computational cost and to skip the meshing of cooling holes. Two optimization approaches are compared: (a) a conventional method that optimizes an uncooled shape and the cooling configuration sequentially and (b) a method that optimizes shaping and cooling concurrently. The concurrent method is found to obtain better aerodynamic efficiency and heat transfer performance than the conventional optimization. Moreover, the aerodynamic efficiency ranking is changed by adding cooling to the uncooled blades. These observations are due to the strong interaction between the coolant and the tip leakage flow. They indicate that the coolant injected at the tip is not passive as expected in the conventional film cooling designs. By blocking the over tip leakage flow or forming a layer of air to level up the equivalent squealer cavity floor, the coolant can reduce the tip leakage loss, which contradicts the conventional wisdom that the added coolant should always lead to extra losses due to the extra mixing. The detailed observations of the flow field indicate that the influence of the squealer height towards the aerodynamic efficiency is caused by two competing effects: the blockage effect to reduce the tip leakage mass flowrate and the sudden expansion/contraction loss effect to generate additional loss.