This is a Preprint and has not been peer reviewed. The published version of this Preprint is available: https://doi.org/10.1016/j.jclepro.2025.146528. This is version 2 of this Preprint.
Sustainable AI infrastructure: A scenario-based forecast of water footprint under uncertainty
Downloads
Supplementary Files
Authors
Manuel Herrera 
,
Xiang Xie,
Andrea Menapace,
Ariele Zanfei,
Bruno Melo Brentan
Abstract
The rapid expansion of artificial intelligence (AI) and cloud computing is creating a significant but often overlooked impact on global water resources. This paper presents a global assessment of water consumption in AI-driven data centres, distinguishing between water consumption for operational use at the facility, off-site water consumption related to electricity generation, and embodied water consumption associated with hardware manufacturing and supply chains. To anticipate future demand, a scenario-based probabilistic forecasting framework inspired by Bayesian methods is developed, combining sparse empirical data with expert-informed assumptions and policy-relevant growth trajectories for the years 2030 and 2050. Results suggest that, without mitigation, global water consumption associated with data centres could increase more than seven times by mid-century, with cooling-related operational consumption accounting for the majority of demand. Several mitigation pathways are identified, including improvements in cooling efficiency, adoption of alternative technologies, and infrastructure planning that takes into account regional water availability. A sensitivity analysis highlights the strong influence of compute growth and efficiency trends on future outcomes. The findings offer a transparent and adaptable basis for aligning AI infrastructure development with long-term water sustainability goals.
DOI
https://doi.org/10.31223/X55M86
Subjects
Engineering
Keywords
Artificial Intelligence, Water footprint, probabilistic forecasting, Data centres, Digital Sustainability
Dates
Published: 2025-04-24 13:10
Last Updated: 2025-07-29 13:19
Older Versions
License
CC BY Attribution 4.0 International
Additional Metadata
Conflict of interest statement:
None
Metrics
Views: 2058
Downloads: 867
Comment #208 Libor Ansorge @ 2025-05-05 05:40
Dear authors,
I find a fundamental contradiction with the Water Footprint Assessment Manual (Hoekstra et al., 2011) in your paper. Water footprint is defined as water consumption, not water use. I encounter this confusion very often with scientists who do not study the water footprint but use this tool in their own research. Equation 1 is generally correct, but the definition of the individual parts is not. In particular, Wo(t) is not about water use, but about water consumption. This error is particularly evident in cooling systems using evaporative water (Section 3.2). Your statement that blowdown represents additional water loss is not always correct, because in terms of the water footprint it represents mainly the so-called grey water footprint. The grey water footprint is a qualitative component of the water footprint, the size of which depends on the amount of pollution discharged. Under certain conditions, even a blowdown can represent a blue water footprint, i.e. a loss of water, but only if the blowdown is discharged into a different catchment than the one from which the water was taken. A theoretical analysis of this problem of misapplication of the water footprint in the energy sector is discussed in Water footprint which is not the water footprint: Critical review of the article by Müller et al. (2024) - https:doi.org/10.1016/j.jenvman.2025.124038. Of course, the principles on which this theoretical analysis is based can be found in the already mentioned Water Footprint Assessment Manual.
Sincerely
Libor Ansorge (libor.ansorge@vuv.cz)