Data Centre Water Use Could Nearly Triple by 2030

Global data centre water consumption could nearly triple by 2030 as artificial intelligence (AI) drives a rapid expansion in computing capacity, raising concerns over the impact of the technology boom on already water-stressed regions.

Rystad Energy estimates that data centres consumed about 222 billion litres of water directly for cooling in 2025. Without significant water-saving measures, annual consumption could climb to almost 644 billion litres by 2030.

However, stronger efficiency measures could substantially reduce the increase. Rystad estimates consumption could reach about 543 billion litres under a moderate water-efficiency scenario, while aggressive adoption of water-saving technologies could bring it down to around 388 billion litres.

“Forecasting the water impact of data centres is far from straightforward,” said Minh Khoi Le, Global Head of Data Center & Hydrogen Research at Rystad Energy, noting that consumption varies according to cooling technology, climate and how water use is measured.

Data Centre Water

The rapid adoption of AI is a major factor behind rising demand. AI servers generate considerably more heat than conventional computing equipment, increasing cooling requirements. However, higher AI workloads do not necessarily translate into a proportional increase in water consumption.

Rack-level liquid cooling can remove heat more efficiently and reduce the burden on facility-level cooling systems. This can allow data centres to make greater use of dry cooling technologies, potentially reducing their reliance on water.

The effectiveness of such measures depends heavily on location. Dry cooling is generally more effective in colder climates, while warmer regions can face greater challenges in managing heat without water-intensive systems.

Rystad also highlighted the importance of water-use effectiveness (WUE), an industry measure that helps assess water consumption relative to IT energy use. The overall water footprint, however, extends beyond water used directly at a data centre because electricity generation can itself require substantial amounts of water.

In the United States, for example, the indirect water consumption associated with electricity can be more than twice the direct water consumption of data centres, illustrating why focusing only on on-site cooling can underestimate their total water footprint.

Water efficiency also varies considerably between operators and locations. Amazon Web Services reported an average direct site WUE of 0.12 litres per kilowatt-hour in 2025, compared with 0.19 for Meta, 0.27 for Microsoft, 0.59 for Digital Realty and 0.91 for Equinix.

different methodologies

But Rystad cautioned that the figures should not be treated as a straightforward ranking because companies use different methodologies and have different mixes of owned, leased and colocation facilities.

AWS’s regional figures demonstrate the importance of geography, ranging from 0.02 litres per kilowatt-hour in Stockholm to 2.85 litres in Jakarta.

Major technology companies are also committing to becoming “water positive” by 2030 by replenishing more water than they consume. AWS, Google, Microsoft and Meta are pursuing water restoration and efficiency projects, although the benefits will differ depending on local watersheds.

Regulation is beginning to emerge. Singapore’s Green Data Center Roadmap targets water-use effectiveness below two litres per kilowatt-hour, while the European Union is developing data centre energy-efficiency measures. US regulation remains fragmented, with states taking different approaches.

The challenge is particularly acute in water-stressed regions. Areas facing high or extremely high water stress are projected to account for 34% of global direct data centre water consumption by 2030.

Rystad estimates that requiring the least water-intensive cooling technologies in these areas could cut consumption by 45%, underscoring the growing need to make water availability a central consideration in the next phase of the global AI and data centre expansion.

#businessnews