Recent months have seen a growing backlash against AI technologies as they develop and are deployed at scale. Water use in data centres and the stress that use is putting on local water resources has been part of this backlash. A recent survey found that most Americans would rather have a nuclear power plant in their area than a data centre.
Globally, communities are now facing competition over their water from AI-driven data centre operations. Many of these communities were already feeling the effects of longstanding water management challenges exacerbated by climate impacts.
As countries including the UK embrace the rapid build out of AI infrastructure, governments and companies must ensure that water use is managed sustainably and transparently or risk further backlash against AI on a wider scale.
Data centres and local supply
The connections between AI and water are wide-ranging, spanning from local impacts that are intertwined with national politics through to geopolitical risks related to water use in global supply chains.
Most visibly at the local level, technology companies that are building and operating large-scale digital infrastructure platforms are facing scrutiny on how they use water, especially in some of the world’s driest areas. Data centre water use is closely connected to the enormous electricity consumption required for computation. That energy use generates heat that must be dissipated, and evaporative water cooling systems are currently a common way to do that.
Despite major water use efficiency improvements and more waterless innovations being deployed in data centres, AI’s rapid growth means that data centres are still becoming a fast-growing driver of water demand.
In the UK, although data centres currently account for a very small proportion of water demand, there are reported plans to build around 100 new centres by the early 2030s. These are expected to become a significant new source of demand.
The UK government has positioned AI as central to its growth plans, pledging £68 billion in investment since January 2025 and designating five AI Growth Zones. This embrace of AI implies an assessment that water allocated to grow the digital economy will, over time, lead to a higher tax revenue and stronger growth.
The government is also planning to build the first new reservoirs in 30 years to keep up with increased demand. But despite these well-meaning plans, there are still concerns over water: 84 per cent of proposed UK data centres are planned in areas that are projected to be water stressed by 2040.
How much water is used for AI, and the extent to which water for AI should be prioritized over uses in other sectors, is a complicated issue and subject to debate. AI is evolving rapidly; it is difficult to quantify exactly how much water it consumes for different purposes, such as using a chatbot or processing a prompt. Simply quantifying water in data centres and then comparing that figure to water use in another sector, such as agriculture, fails to capture the full scope of the footprint.
Geopolitical risks and impacts
Local considerations on AI water use are also connected to geopolitical risks and impacts further down the supply chain. Governments should take these into account when calculating the impact of AI water use.
A data centre might look like an isolated piece of industrial infrastructure in a local community, but the servers inside it connect it to global mining and manufacturing supply chains. These servers rely on complex components such as high-powered semiconductor chips, which are tied to global supply chains that have their own intense water impacts.
Taiwan produces over 90 per cent of the world’s advanced semiconductors. Semiconductor manufacturing is water-intensive, due to the high consumption of ultrapure water (UPW) required to maintain extreme purity levels in manufacturing processes.
But Taiwan’s hydrological balance relies on seasonal typhoons to replenish groundwater, and climate change has made typhoons less predictable, increasing the risk of drought. This water-based risk is compounded by other geopolitical risks such as shifting tariff policies and the potential of military conflict with China, leaving the global supply chain vulnerable to shocks that should be factored into water-related strategic decision making.
A shared challenge
Given that water is a shared resource, and any water challenges are essentially shared across society, collective action from governments, investors and companies – in collaboration with communities – is necessary. System-wide improvements are needed.
Some technology companies are already taking circularity solutions seriously, and are scaling advanced cooling technologies. Water recycling in data centres has been implemented in some places such as the Netherlands, where closed loop systems are starting to be put into use.
These solutions are encouraging and will go a long way. But they will not fully address the fundamental water challenges that are currently inherent in scaling AI. Governments committed to the digital economy will need to fund broader solutions, which means greater investment in public water services.
They will also need to scale those solutions that support good stewardship of water. These include developing practical actions to protect shared water resources, including equitable access to public water services that prioritize domestic water use and more vulnerable water users.
