Sector
Artificial intelligence workloads
AI training and inference concentrate very high power density in individual facilities, which increases cooling demand. Published per-query water figures circulate widely and almost always omit the assumptions that determine the answer.
Key facts
- Part of
- Data centres
What drives water use in this sector
- Facility cooling design and location
- Whether the figure covers training amortised over queries, inference only, or both
- Whether indirect water from electricity generation is included
- Hardware generation and utilisation
- Model size and query length
Relationships
- Part of: Data centres
Related intelligence
Sources
- Food and Agriculture Organization of the United Nations — AQUASTAT — Global Information System on Water and Agriculture. Public but restricted · CC BY-NC-SA 3.0 IGO
- United States Geological Survey — Estimated Use of Water in the United States. Open · US Government work — public domain
- International Energy Agency — Water–energy nexus analysis. Public but restricted · Publisher terms — public access, reuse not clearly granted