Multi-speed transmissions for electric vehicles (EVs) can better improve EV efficiency and performance in comparison with conventional single-speed transmissions. To achieve a superior multi-speed transmission design, gear ratios and shifting patterns should be optimized by considering the electrical and mechanical efficiencies. Because most powertrain losses occur in the inverter, motor, and transmission, loss analysis was conducted according to the characteristics of each component. To confirm the effects of these component efficiencies on EVs, an EV efficiency analysis was performed based on three cases of powertrain efficiency by considering the variable efficiency of each component combination: i) motor only, ii) motor and transmission, and iii) inverter, motor, and transmission. An optimization problem was formulated by adopting design variables, such as gear ratios and shifting patterns, and the objectives of energy consumption and acceleration ability. An artificial neural network (ANN)-based multi-objective optimization process was proposed as an alternative to the excessive effort for calculation in the optimization process. To verify the importance of considering the efficiency of various powertrain components, transmission optimization was performed for the variable efficiency of each component combination. When comparing the highest efficiency solutions of the cases considering only part of component efficiency with the case considering all components, the energy consumption and acceleration ability of the latter case were found to be superior by approximately 1.7% and 5.1%, respectively. Furthermore, since the computational time for the EV and ANN models required approximately nine months and one day, respectively, this result demonstrates the effectiveness of the ANN-based multi-objective optimization process.