AI-Driven Data Center Energy Profile, Power Quality, Sustainable Sitting, and Energy Management: A Comprehensive Survey
Rouzbeh Reza Ahrabi, Alireza Mousavi, Ebrahim Mohammadi, Ryan Wu, Aoxia Kevin Chen · 2025
The rapid expansion of Artificial Intelligence (AI) workloads has significantly reshaped data center energy consumption, cooling requirements, and power quality management, with projections indicating that U.S. data centers alone could account for up to 12% of the nation’s electricity consumption in the coming years. This survey reviews the shift from conventional CPU-based servers to high-density AI-specialized systems, spotlighting the growing need for advanced cooling technologies like liquid and immersion cooling. We also address power quality issues introduced by GPUs, emphasizing adherence to IEEE 519-2022 standards for mitigating harmonic distortion. Key factors influencing data center siting—including power availability, climate considerations, and regulatory incentives—are examined, alongside the adoption of renewable energy solutions for greener operations. Furthermore, we explore emerging AI-driven Energy Management Systems (EMS) that enable real-time optimization of power distribution, workload scheduling, and energy storage. By offering a holistic view of current practices and challenges, this paper highlights actionable, sustainable strategies for meeting escalating AI-driven computational demands.