tap, sink, dirty water, desire, ecology, contamination, environment, problem, pollution, crisis, water, drought, the greenhouse effect, closeup, drinking water, water quality, plumber, fittings, bathroom, sanitary, hygiene, rust, iron, corrosion

AI data centres and its strain on Britain’s water supply

From the Guardian article: Not enough water for UK’s datacentre plans, trade body says

AI data centres could become a real pressure point for the UK’s water system unless planning keeps pace with expansion. The core issue is simple: the sector needs water for cooling, while the UK already faces a growing supply gap and hotter, drier summers are increasing stress on the network.

sever, digitization, mainframe, computer, digital, binary, cable, network, data center, internet, frame, router, ethernet, server, switch, data center, data center, data center, data center, server, server, server, server, server

What the article says

The Guardian reports that Water UK warned the government’s AI growth plans are “fatally flawed” because official water forecasts do not properly include data centres. It also notes that England could face a 5 billion litre a day public water shortfall by 2055, which makes large new industrial demand harder to absorb. In practice, the concern is not only total demand, but when and where that demand appears, especially in already water-stressed regions.

Impact of data centres on the water supply

Data centres are often described as low-impact compared with heavy industry, but their cooling systems can still draw meaningful volumes of potable water. The challenge grows with AI because high-density computing tends to increase heat output, and that can raise cooling needs. Water demand is also concentrated in a relatively small number of large facilities, so a few sites can create local strain even if the sector’s national share looks modest.

Ways to expand safely without risking water security

There are several ways the UK could support more AI data centres without undermining household water security.

First, planning should require operators to disclose expected water use, including direct cooling demand and the indirect water linked to electricity generation.

Second, new sites should be directed away from the most water-stressed areas wherever possible, with AI growth zones aligned to actual water availability rather than just grid capacity.

Third, the sector should be pushed towards non-potable and recycled water, such as treated wastewater or closed-loop systems, so drinking water is not the default cooling source.

Fourth, operators should be required to use the most water-efficient cooling designs available, including air cooling, dry cooling, and other low-water technologies where feasible.

Fifth, the wider water system needs faster investment in leakage reduction, new storage, and treatment infrastructure, because the UK cannot meet rising industrial and domestic demand by efficiency alone.

A workable balance

The best path is not to block AI infrastructure outright, but to make water a hard planning constraint from the start. That means better disclosure, stronger siting rules, non-potable cooling, and pressure on the wider water industry to cut losses and expand supply. Done properly, the UK could grow AI capacity while keeping water security for households as the priority.


Sources