While we’re ordered to not use hosepipes, a single artificial intelligence data centre can use the same amount of H2O as 3,000 homes. That’s why the drive to turn Britain into an AI superpower could leave us gasping for water
The article highlights concerns that data centres in water-stressed parts of southern England may consume substantial volumes of water while households face hosepipe restrictions. It argues that Britain’s expansion of artificial intelligence infrastructure could intensify pressure on local water supplies, particularly during heatwaves when cooling demand rises.
Slough, which has 32 data centres and is served by Thames Water, is presented as an example: reservoirs were reported to be 74% full rather than the usual 87% for the time of year. TechUK data cited by the article suggests one in 25 surveyed data centres uses more than 100,000 cubic metres of water annually—roughly the use of 3,000 households—though the survey covered only 72 sites. Larger facilities often use water-based cooling to manage heat from high electricity consumption, and US research cited suggests daily water use can be far higher in summer.
- Data centres may strain supplies in water-stressed areas.
- Slough households face restrictions while commercial sites remain exempt.
- Water use rises sharply when hot weather increases cooling needs.
New here? Start with this
Data centres are large buildings filled with computers that store data and run online services. Artificial intelligence systems need especially powerful computers, which produce a great deal of heat and must be kept cool.
Some centres use water as part of their cooling systems, although the amount varies by design, weather and location. This can become more significant in areas where water supplies are already under pressure, particularly in dry summers or during heatwaves.
The government wants to expand Britain’s computing capacity to support AI and other digital services. Water companies, councils, technology firms and regulators may therefore need to balance new development with the needs of homes, businesses and the wider local environment.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Supporters of tighter scrutiny argue that essential household water needs and environmental resilience should take priority over rapid data-centre expansion in already stressed catchments. They contend that large facilities can add demand precisely during heatwaves and drought restrictions, so transparent reporting, robust planning limits and investment in low-water cooling are needed before further capacity is approved. The underlying concern is that the economic benefits of AI should not shift scarcity costs onto residents or local ecosystems.
The case against
Supporters of AI infrastructure argue that data centres can deliver substantial economic, scientific and public-service benefits, and that isolated figures should not be treated as representative of an entire sector. They note that the cited survey covers a limited sample, water use varies greatly by design and location, and operators can reduce demand through recycled water, alternative cooling and more efficient equipment. On this view, the appropriate response is targeted water planning and enforceable efficiency standards, rather than impeding strategically important investment.