As AI clusters scale rapidly, data center cooling architectures are evolving beyond traditional air-cooled systems toward liquid cooling, immersion, hybrid adiabatic systems, and advanced facility water loops. Existing sustainability metrics such as WUE and PUE provide useful directional guidance but fail to capture the full operational, geographic, and lifecycle complexity of water consumption in modern AI infrastructure.

This panel brings together contributors from the OCP water efficiency workstream to discuss emerging best practices for holistic water accounting across cooling technologies, facility loops, and regional water conditions. Topics include stress-weighted water metrics, cooling loop boundaries, reuse and replenishment strategies, embodied water in power generation, operational resilience, and normalization across diverse cooling architectures.