The increasing adoption of cloud computing has led to a significant rise in energy consumption within data centers, posing both economic and environmental challenges. A specific problem faced by cloud data centers is the inefficient allocation of computational resources, leading to excessive energy usage and operational costs. Traditional resource management techniques often fail to adapt dynamically to varying workloads, resulting in suboptimal energy utilization. To address this issue, we propose the implementation of an adaptive machine learning-based resource allocation algorithm that optimizes energy consumption in real time. Our solution leverages a combination of reinforcement learning and predictive analytics to dynamically adjust resource allocation based on workload predictions and current system states. Experimental results demonstrate a substantial reduction in energy consumption while maintaining high performance and service quality. This approach not only enhances the energy efficiency of cloud data centers but also contributes to sustainable computing practices. By addressing the specific problem of inefficient resource allocation, our proposed solution offers a viable pathway for data centers to achieve significant energy savings and operational efficiency, thereby supporting the growing demand for eco-friendly cloud computing solutions.