AI-driven cooling technologies for high-performance data centres: state-of-the-art review and future directions
Dlzar Al Kez, Aoife Foley, Fadhli Wong B.M. Hasan Wong, Andrea Dolfi, G. Srinivasan
Abstract
The increasing computational demands of artificial intelligence (AI), high-performance computing (HPC), and hyperscale cloud platforms are placing significant thermal and energy pressures on data centre infrastructure. Traditional air-based cooling systems are increasingly inadequate for managing these loads, prompting a transition toward more efficient, scalable, and sustainable alternatives. This study presents a comprehensive, system-wide review of next-generation cooling technologies, including direct liquid cooling, immersion cooling, two-phase systems, spray and jet impingement cooling, and heat pipe-based solutions. Unlike previous reviews focused on component-level or single-technology evaluations, this study integrates technical performance, commercial readiness, and environmental impact across diverse deployment conditions. A detailed comparative framework synthesises thermal efficiency, scalability, and water usage across air, liquid, and hybrid systems. Special attention is given to commercially mature solutions such as RDHx and cold plate DLC, while the feasibility of emerging methods like AI-driven cooling, phase-change materials, and thermoelectric technologies is evaluated. The review further explores heat reuse potential and ESG-aligned design strategies critical to decarbonising digital infrastructure. By mapping trade-offs across performance, cost, and sustainability, this study offers actionable insights for data centre operators, designers, and policy stakeholders navigating the path to high-efficiency, AI-ready cooling.