Optimizing Performance and Cost Efficiency in AI-Driven Cloud Infrastructures: A Multi-Objective Approach
Sai Krishna Khanday · 2025
Natural Language Processing (NLP) in cloud computing, combined with artificial intelligence (AI), has significantly enhanced system capabilities by maintaining high performance. However, achieving optimal performance and cost efficiency in AI-driven cloud infrastructures remains a complex challenge. AI workloads, due to their heterogeneous and dynamic nature, present unique difficulties in simultaneously addressing performance and cost objectives. This paper introduces a multi-objective optimization model designed to optimize both performance and cost efficiency in AI-driven cloud infrastructures. Leveraging machine learning techniques, the model analyzes historical workload data to predict future resource demands, enabling a dynamic resource allocation strategy for efficient resource provisioning and management in cloud environments. Additionally, a price-aware scheduling algorithm has been developed to allocate resources appropriately based on the computational requirements and cost constraints of various AI workloads. By continuously monitoring resource usage and dynamically adjusting allocations, the proposed algorithm achieves an efficient balance between performance targets and cost objectives. Simulation and experimental results validate the superiority of the proposed approach over traditional methods, demonstrating significant improvements in cost efficiency and performance metrics. This research underscores the potential of multi-objective optimization in addressing the challenges of AI-based cloud infrastructures. The findings highlight the effectiveness of dynamic resource allocation and price-aware scheduling in improving overall system efficiency, paving the way for more advanced and cost-effective AI-driven solutions in cloud computing.