Six consumers were responsible for a substantial 65 percent of total water usage, with the largest data centre in the study consuming over a quarter of the 1,066 million litres used by all 200 facilities combined.
This suggests there is a different narrative to explore about the sector’s future than what is promoted by TechUK. The WRc research indicates that if the two largest consumers were emblematic of the entire sector, it would lead to a two percent increase in non-household demand and a tenfold increase in sector usage.
Blaming
The data centre industry’s assertion that it is still expanding from a low base is misguided. Water companies have developed detailed plans for managing water demand from households, industries, and public services between 2024 and 2029, yet they are unsure about how many million litres are unaccounted for as data centres proliferate.
Industry insiders understand that the sheer number of facilities provides limited insights, though their concentration in certain areas is significant.
The critical data is the power needed to operate servers and other computing equipment (IT Load). When combined with details about the operator’s choice of cooling system configuration, it helps estimate water demand.
Accurate estimates are challenging, and water companies, data centre operators, and public authorities are blaming one another for the lack of usable data rather than disclosing what they know.
It is evident, however, that the choice of cooling system greatly impacts the water consumption of a data centre.
Water
The Environment Agency’s report indicates that hybrid cooling systems are responsible for seasonal spikes in demand.
These systems combine air-cooled and water-cooled methods, primarily using air cooling when temperatures are low, and switching to water-intensive chillers when temperatures rise.
According to TechUK’s survey, 44 percent of data centres used hybrid systems, with an additional 5 percent relying solely on water-based systems. The remaining 51 percent claimed they did not use water at all, except for domestic purposes like flushing toilets.
The Environment Agency’s analysis confirms that hybrid cooling systems remain “water intensive,” and the WRc report further emphasizes how the growing size of data centres increases the risks associated with water demand spikes during heatwaves.
Data centre operators often argue that new facilities entering the planning system will not require significant water on site.
Gigafactory
Reducing water usage, however, leads to increased power needs for fans and mechanical cooling systems, resulting in “waterless” data centres having a greater indirect water use at the electricity generation source.
Data centres currently under construction or being planned across the UK consume electricity comparable to many small power plants. Globally, major data centre hubs like those in northern Virginia and Ireland have significantly reshaped electricity grids to meet their demands.
Ireland imposed a ban on new data centre connections in the Dublin area from 2021 to late 2025 due to grid capacity issues. The increased power demand during heatwaves, as cooling systems work overtime to mitigate temperatures in homes, offices, and hospitals, makes the choice between less water and more power increasingly difficult.
The Environment Agency warns that their analysis has yet to consider the impact of AI hyperscale data centres. The issue of AI production is also not limited to “gigafactory” style sites located outside urban areas.
Efficiency
Certain AI workloads require proximity to the end-user, such as autonomous vehicle driver assistance systems. City-centre data centres are more likely to choose water-based cooling due to limited floor space for alternatives, as confirmed by major data centre operator Global Switch in the WRc research.
A UN report published last month emphasizes that there is no straightforward trade-off between power, water, and land use for data centre operators and the AI industry as climate change accelerates.
“Many believe that AI’s environmental footprint decreases as technology advances and processes become more efficient. However, this is only part of the overall problem,” said Professor Kaveh Madani, a co-author of the report.
“More efficient and affordable AI and energy lead to increased AI consumption, making the overall footprint significantly larger than what is saved through efficiency improvements.”
This Author
Anne Alexander is a journalist and researcher, currently engaged in a project with Watershed Investigations examining the impact of data centres and the AI industry on the planet.
