This paper introduces a novel approach to optimizing cloud resource allocation using genetic algorithms and machine learning. The goal is to reduce costs while meeting service-level agreement (SLA) requirements. It combines genetic algorithms for real-time optimization and neural networks for predicting future resource needs. Testing on a cloud simulation platform shows it outperforms existing methods in cost-effectiveness and SLA compliance. The hybrid approach combines “NSGA-II and SPEA2” with supervised learning, leveraging their strengths to overcome individual weaknesses. These algorithms enable efficient solution selection, and supervised learning predicts fitness values for new solutions, enhancing optimization speed and accuracy for complex multi-objective challenges. This approach is versatile and can be applied in various cloud computing applications, improving resource allocation efficiency and reducing expenses. It consistently outperforms traditional resource allocation strategies.