This paper investigates the performance of Neural Network (NN) based PI Controller (NNC) and Fractional Order Sliding Mode Controller (FOSMC) for sensorless speed control of Interior Permanent Magnet Synchronous Motor (IPMSM). It proposed new method of NN based PI sensorless speed control based on offline learning using look up table obtained from analysis of the PI controller. The FOSMC was designed, analysed and its stability was guaranteed using Lyapunov stability theory to validate its higher performance. This paper proposes novel speed observer as low pass filter of motor currents and load torque in the time domain to increase the reliability of the closed loop system. Simulations results using MATLAB/SIMULINK proved the improved performance of the two controllers and the strong robust performance of FOSMC compared to Neural Network based PI sensorless speed Controller (NNC) against large ranges of uncertainties and external load disturbances in field-oriented Vector Control (VC) scheme. Analysis study of the effect of fractional order differentiator was presented to show that as its value decreases, the transient and steady state performance of the motor is improved.