Green AI for Sustainable Big Data Platforms

Abstract

This article explores how AI-driven optimization can be applied to Big Data workloads in hyperscale cloud data centers to reduce energy consumption and carbon footprint without compromising performance. The article focuses on using machine learning models to intelligently manage compute resources, workload scheduling, data placement, and query optimization across large-scale distributed platforms such as Spark, Hadoop, and cloud-native analytics systems. By dynamically predicting workload demand, energy usage, and system efficiency, the proposed framework aims to minimize power waste, improve resource utilization, and enable carbon-aware computing. The proposed framework achieves up to 38% energy reduction and 44% carbon footprint reduction compared to baseline configurations while maintaining performance within 1.5% of service level agreements. The outcome of this research contributes to building environmentally sustainable, cost-efficient, and high-performance Big Data ecosystems for next-generation cloud platforms.