A report published in April 2026 found that 84 percent of proposed data centres in the UK are planned for areas projected to be water stressed by 2040. The finding lands as Whitehall accelerates its push for artificial intelligence capacity, with designated growth zones and a target to increase computing power twentyfold by 2030.
One hyperscale campus can consume around 1.5 million litres of water a day, roughly the daily requirement of 10,000 people. Nationally the picture is starker still. The UK faces a projected water deficit of nearly 5 billion litres per day by 2050 or 2055, the result of climate patterns, population increase and competing demands that already include these very facilities.
Global water demand linked to artificial intelligence is expected to reach between 4.2 billion and 6.6 billion cubic metres a year by 2027. That volume exceeds half the United Kingdom's total annual household water use. The majority of the planned sites sit in London and the South East, regions already among the most water stressed in Europe. Thames Water has projected that data centres could account for almost 30 percent of new water demand in its area by 2050.
Efficiency already embedded
Yet the data tell a more nuanced story than simple scarcity narratives allow. TechUK's report from August 2025, produced with the Environment Agency, showed that 51 percent of surveyed data centres rely on waterless cooling systems. More than half of UK facilities currently operate non-evaporative methods. Many consume less water annually than a typical leisure centre.
Baringa has reported that without clear frameworks for potable and non-potable water alongside proactive planning support, high-potential sectors such as data centres face unnecessary risk while communities deal with shortages. The caution is well placed. It underscores the case for intelligent allocation rather than blanket environmental vetoes that would erode competitive edge.
The Government Digital Sustainability Alliance has called for mandatory reporting of water and energy use by data centres and AI systems. Such transparency would sharpen national planning. It would not, however, justify layering regulatory hurdles atop an industry already demonstrating rapid technical progress in efficiency.
