Data Centres vs. Drinking Water: What Matters To Us Most?

Summary of Data Centres vs. Drinking Water: What Matters To Us Most?

by Goalhanger

16m•July 30, 2026
Listen to Episode

Overview of Data Centres vs. Drinking Water: What Matters To Us Most?

This episode of Goalhanger’s When Nature Becomes a Security Threat argues that water scarcity is no longer just an environmental issue — it is becoming a national security, economic, and geopolitical threat. Host Caroline Lucas examines how global “water bankruptcy,” UK water stress, and the rapid expansion of AI data centres are colliding, raising a stark question: should scarce water be reserved for essential human needs, or for powering the digital infrastructure behind AI growth?

The Big Argument

  • The UN has warned that the world has entered an era of “global water bankruptcy”, meaning some water systems are suffering irreversible damage.
  • Water scarcity is described as a “threat multiplier” because it worsens:
    • food insecurity
    • economic instability
    • social tension
    • geopolitical conflict
  • For the UK, water insecurity is not distant:
    • 15% of surface water
    • 27% of groundwater aquifers are already being depleted faster than nature can replenish them.

Why Data Centres Are Central to the Problem

AI infrastructure is water-intensive

The episode focuses on the growing water demand from AI data centres, which need water in two ways:

  • Direct cooling of servers
  • Indirect water use through electricity generation

A guest from Global Action Plan UK, Ollie Hayes, argues that the true scale is underestimated because the industry is not transparent about its water use.

Key claims about data centre water use

  • A mid-sized AI data centre can use as much water as about 10,000 people.
  • Many proposed sites are being placed in water-stressed regions.
  • Around 84% of proposed data centre sites may be in areas projected to be water-stressed by 2040.
  • The UK’s southeast, already the driest and most water-stressed region of the country, is seeing heavy concentration of existing and planned facilities.

Why the Numbers Are So Hard to Pin Down

The episode criticizes the lack of public data and regulation:

  • There are no strong requirements for data centres to disclose water consumption.
  • Some water companies reportedly do not even know how much they are supplying to data centres.
  • Industry advocates often cite misleading averages, such as water used per ChatGPT query, which can hide the real impact of large, clustered facilities.

A Critical Policy Concern: Data Centres as National Infrastructure

A major revelation discussed in the episode is that AI data centres were designated critical national infrastructure in 2024.

Why that matters

  • Water companies may be unable to disconnect or restrict supply to them during drought or severe water stress.
  • This creates a troubling hierarchy where data centres could effectively be protected ahead of:
    • households
    • farms
    • hospitals and the NHS
    • other essential users

Polling cited in the episode suggests the public strongly disagrees with that priority order.

The Irony of Public Water-Saving Campaigns

The episode highlights the contradiction between:

  • government campaigns urging the public to take shorter showers and use less water
  • and the growing tolerance for water-hungry AI infrastructure

This is framed as politically and morally difficult, especially as the UK faces an estimated 5 billion litre per day water shortfall in coming decades, even before factoring in the full impact of data centres.

Can Technology Solve It?

The guest says some mitigation is possible, but only with strong rules:

Potential solutions

  • Require least-water-intensive cooling methods
  • Use wastewater where possible
  • Deploy closed-loop cooling systems
  • Ensure data centres are powered by renewable electricity

But there are limits

  • Data centres prefer evaporative cooling because it is cheaper and more efficient for their business model.
  • Even if direct cooling water use is reduced, most of their water footprint comes from electricity generation.
  • The episode notes that three-quarters or more of their total water footprint may come indirectly from powering the facilities — potentially up to 90% in some studies.

The Climate and Growth Collision

The episode concludes that the AI boom and climate goals are on a collision course:

  • Big tech firms have increasingly abandoned earlier climate commitments.
  • The industry’s messaging has shifted from:
    • “AI doesn’t use much energy”
    • to “we can power it all with renewables”
    • to “it’s climate goals or AI — choose one”
  • Caroline Lucas argues this is not a realistic or responsible trade-off.

Main Takeaways

  • Water scarcity is already a serious security issue, not a future one.
  • Data centres are becoming a major new source of water demand in already stressed areas.
  • The UK lacks strong transparency and regulation around AI infrastructure’s environmental impact.
  • Prioritizing AI water needs over homes and essential services would be politically and socially explosive.
  • The only sustainable path forward is likely:
    • strict reporting requirements
    • environmental impact assessments
    • limits on siting in water-stressed regions
    • renewable-powered operations
    • reduced reliance on water-intensive cooling

What the Episode Sets Up Next

The host closes by previewing future topics in the series, including:

  • food security
  • climate migration
  • the myths behind public narratives on environmental collapse