Water Consumption by Data Centers
資料中心水資源消耗
Inception
Risk Description
When data centers required for AI inference and training consume substantial cooling water; due to unmitigated control gaps, in regions where water is already scarce, heavy consumption can intensify supply pressure and create competition with residential and agricultural use, triggering external stakeholder impacts and causing this environmental impact has historically been overlooked in assessments and has only recently drawn attention.
Framework Mappings
NIST AI 600-1Environmental Impacts
ISO/IEC 42005影響評估流程
ISO/IEC 5338啟動
MIT AI Risk RepositoryDomain 6
Risk Treatment & Implementation Guidance
Track data-center water consumption and factor local water stress into siting; Improve cooling efficiency and assess alternatives to reduce dry-season withdrawal impact; Establish community communication and water-use coordination