Chapter 09 · Sustainability & AISustainability & AI
AI water footprint
Definition
The freshwater consumed in support of AI systems — for cooling data centres directly, and indirectly through the water used in electricity generation and chip manufacture.
References
This source provides part of the technical or institutional basis for the definition.
This source supports the explanation of how the term is applied, measured or governed in practice.
This source supports the wider sustainability significance and context described in the entry.
Overview
What it means
Water demand is largely hidden: evaporative cooling on site, plus water intensity of the power supply. Research quantifying models' water use (Li et al. 2023, 'Making AI Less Thirsty') showed consumption varies sharply with location, season and energy mix. Facilities in water-stressed regions raise siting questions.
How it is used
Data-centre water use is now reported alongside energy in the EU's mandatory reporting scheme and tracked through metrics such as WUE. Tech companies publish water commitments and, increasingly, site-level consumption data.
Why it matters
AI's resource debate has been electricity-first; water is the less visible twin. In stressed basins, data-centre withdrawals compete with communities and ecosystems — making water a material siting and disclosure issue, not a footnote.