The Permanent Magnet Synchronous Motor (PMSM) drives are needed to have high dynamic accuracy and robustness under parameter uncertainty and load disturbances. Typical Model Predictive Direct Speed Control (MPDSC) techniques tend to have poor performance since the prediction accuracy is heavily dependent on the motor model. This is the main impact of this paper, which concerns the development of a real-time parameter mismatch and load disturbance estimation strategy based on the adaptive Sliding-Mode Observer (SMO) to enhance the predictive control prediction accuracy without increasing the predictive control horizon or the need for iterative optimization. The proposed adaptive SMO adaptively tunes the observer gains depending on the operating condition of the system, which enhances disturbance rejection gracefully and lessens chattering as compared with the conventional observer-based MPDSC, which uses fixed gains. The 2-level 3-phase voltage-source inverter was utilized to implement the suggested controllerin MATLAB/Simulink. The predictive controller was operated with Ts = 50μs, and the current harmonic distortion was estimated by Fast Fourier Transform (FFT) analysis as per IEEE harmonic measurement practice at 1500 rpm of the steady state operating speedof the machine and 10kHz switching frequency of the inverter. The simulation results show that the performance is better under varying speed and variable-load operating conditions. The settling time of the proposed controller was reduced to 38ms as compared to 40ms for the conventional MPDSC, the current THD was reduced from 13.95% to 2.41%, the speed overshoot and torque ripple were reduced, and the disturbance rejection and tracking accuracy were improved significantly. Moreover, the adaptive observer minimized prediction errors and kept the computational structure the same as that of the conventional finite-control-set MPDSC. The results obtained in this study prove that the suggested approach is an effective and practical scheme for strong drive of PMSM.