<p>An optimal design methodology for high-speed angular contact ball bearings (ACBBs) is presented that integrates a quasi-static load distribution model with a thermal network analysis within a multi-objective genetic algorithm (MOGA) framework. The primary objective functions are the maximization of stiffness and fatigue life, while thermal expansion and the resulting clearance variations are explicitly incorporated to improve prediction accuracy. Experimental validations confirmed the reliability of both the stiffness and thermal models by comparing numerical results with available literature data. Simulations under constant loading conditions (radial and axial loads of 500 N each) revealed that, without thermal effects, stiffness and fatigue life change gradually with increasing rotational speed. However, when thermal expansion is considered, a critical speed of 24000 rpm was identified, beyond which interference fit occurs due to negative clearance, leading to abrupt changes in performance. The optimization process, conducted over 200 generations with 3000 individuals per generation, yielded a Pareto front from which a knee-point solution was selected. Compared to the reference bearing, the optimized design achieved up to a 2-fold increase in stiffness and a 55-fold improvement in fatigue life, with the knee-point solution showing improvements of approximately 1.76-fold and 4.86-fold, respectively. Sensitivity analysis further indicated that inner ring conformity and the number of balls are key factors influencing fatigue life.</p>