The water use of data center workloads: A review and assessment of key determinants

Overview

This article is a review and assessment by LBNL and UCSB authors of water use per computing workload. It was published in Resources, Conservation and Recycling, volume 219 (2025). The study treats water use per workload as two components: on-site cooling water and grid-related water. The study matters for treating data centre water use at workload level and in connection with electricity generation, not only as a cooling matter. This wider view carries water consumption beyond the facility boundary.

The article ranks the factors that drive litres per workload and estimates achievable minima under site constraints. The ranked determinants are server efficiency, grid water factor, utilization, cooling type, infrastructure efficiency, climate zone, inactive servers and refresh cycle. The headline findings show how wide the variability is: workload-level water use varies by more than 10,000-fold; water per kWh of server electricity varies over 1,000-fold, and server workload efficiency about 10-fold. The conclusion is that there is no single recipe.

The article is a paid journal publication and only the record's abstract-level information was used here; the full text was not read. Method details and the numeric weights of the determinants are therefore not given here, and the original article should be consulted for them. This ranking helps set design and operating priorities by showing which factors affect water per workload most. Achievable minima, meanwhile, vary with site constraints.

Key points

1. Workload-level water use varies by more than 10,000-fold.
2. Water per kWh of server electricity varies over 1,000-fold; server workload efficiency about 10-fold.
3. Ranked determinants: server efficiency, grid water factor, utilization, cooling type, infrastructure efficiency, climate zone, inactive servers, refresh cycle.
4. Water use combines on-site cooling water and grid-related water.
5. The conclusion is that there is no single recipe.
6. Published in Resources, Conservation and Recycling vol. 219 (2025).

Using it in design

• In the water module, report both on-site WUE and grid (source) water using regional factors.
• Let users see the sensitivity to cooling type and climate.
• Avoid presenting a single best cooling choice.

Who should read it

Designers and researchers weighing water use, sustainability and cooling type together.

Related standards

• ISO/IEC 30134-9

Publication details

Authors

Nuoa Lei; Jun Lu; Arman Shehabi; Eric R. Masanet

Publisher

Resources, Conservation and Recycling (Elsevier)

Year

2025

Document language

English

Source type

Scientific

Access

Citation only

DOI

10.1016/j.resconrec.2025.108310

Licence

Copyright retained; citation only — the publisher’s terms apply to the original document

Citation