Every time someone uses ChatGPT to write a 100-word email, roughly 519 millilitres of water is consumed, almost equal to a standard water bottle.
That estimate comes from a peer-reviewed 2025 paper published in Communications of the ACM by Pengfei Li, Shaolei Ren and their colleagues at the University of California, Riverside.
The study accounts for both the direct water used to cool data centre servers and the indirect water required to generate the electricity that powers them.
When this is scaled to millions of users making several queries every day, the numbers become staggering.
By 2027, global AI infrastructure is projected to consume between 4.2 billion and 6.6 billion cubic metres of water every year. That is nearly equal to half of the United Kingdom’s total annual water withdrawal.
More worryingly, much of this water is being drawn from regions that are already facing water stress.
Why AI Data Centres Consume So Much Water
Data centres generate enormous heat. The powerful chips that run modern AI systems, especially high-end graphics processing units, can each consume and release between 300 and 700 watts of heat while operating under heavy load.
To control this heat, many data centres rely on evaporative cooling. In this system, water is pumped through the facility to absorb heat from servers. A portion of that water then evaporates into the atmosphere.
Around 80 percent of the water drawn into an evaporative cooling system is permanently lost through evaporation.
The remaining water may return to the system, but often at higher temperatures and with chemical residues.
The new generation of AI-focused hyperscale data centres is larger, denser and far more heat-intensive than the general cloud infrastructure built in the 2010s.
A single large data centre campus can now consume more water in a day than a town of 10,000 people uses for drinking, sanitation, cooking and agriculture combined.
According to the 2024 US Data Centre Energy Usage Report by Lawrence Berkeley National Laboratory, prepared for the US Department of Energy, data centres consumed around 17.4 billion gallons of water directly for cooling in 2023.
Another 211 billion gallons were consumed indirectly through electricity generation.
The report also noted that data centre load growth has tripled over the past decade and is projected to double or triple again by 2028.
Google, Microsoft And Meta’s Water Use
Major technology companies have started disclosing their water consumption in annual sustainability reports. The trend is clear: water use is rising.
Google’s 2024 Environmental Report said the company consumed around 8.1 billion gallons of water during the year, with about 95 percent used at data centres. This marked an 8 percent increase from 2023.
The previous years had also seen sharp increases, with Google’s water consumption nearly doubling in three years.
Microsoft has reported smaller numbers, but the pattern is similar. The company consumed around 1.7 billion gallons of water in 2022, a 34 percent increase from the previous year.
Independent reporting on Microsoft’s data centre cluster in West Des Moines, Iowa, where GPT-4 training runs were conducted in 2022, found that a single training run consumed 11.5 million gallons of water in July 2022 and 13.4 million gallons in August.
The same cluster has since expanded to five facilities, drawing 68.5 million gallons annually from the local municipal water system.
Meta consumed around 813 million gallons of water globally in 2023. Amazon, which operates the world’s largest cloud infrastructure, does not publish aggregate water consumption figures.
AI Is Expanding In Water-Stressed Regions
The Li and Ren paper projects that by 2027, global AI demand could lead to water withdrawal equivalent to more than four times Denmark’s annual usage, or nearly half of the United Kingdom’s total yearly withdrawal.
The problem is not only the volume of water used, but also where that water is coming from.
Microsoft acknowledged in its 2023 sustainability report that around 42 percent of its water consumption came from areas classified as water-stressed under the World Resources Institute’s rating system.
Google reported that 15 percent of its freshwater withdrawals in 2023 came from areas facing high water scarcity.
The consequences are already visible. In Chile, Google paused a planned $200 million data centre near Santiago after an environmental court ruled that the company had not adequately assessed the impact on the Central Santiago Aquifer.
The country had been facing drought for 15 years and had begun rationing residential water in 2022.
In Querétaro, Mexico, where 32 new data centres are planned, the state experienced its worst drought in a century in 2024.
Microsoft has secured rights to around 25 million litres of water annually from a local aquifer that is already running a 60-million-litre annual deficit.
In Arizona, a $14 billion data centre project was withdrawn in 2024 after local residents opposed the rezoning.
What Companies Are Not Disclosing
The available figures are based only on what companies have chosen to disclose. The actual water footprint of the AI industry is likely much larger.
There are three major disclosure gaps.
The first is the difference between water withdrawal and water consumption. Withdrawal refers to the total volume of water taken from a source, while consumption refers to the water permanently lost, mostly through evaporation.
Companies often report only one of these figures, which can make their footprint appear much smaller.
The second gap is the difference between direct cooling water and indirect water used for electricity generation.
The Li and Ren research estimates that indirect water use could be around 12 times higher than direct cooling water use. Very few corporate reports include this figure.
The third gap is the absence of facility-level data. A company-wide annual total does not tell local communities whether their own aquifer or water system is under pressure.
The UC Riverside paper is important because it uses publicly available information to estimate these hidden gaps.
As independent researchers produce credible estimates, technology companies may face growing pressure to disclose more detailed water data.
The Big Question For AI
The global infrastructure for AI is being built at an extraordinary pace. Behind the excitement over artificial intelligence are massive industrial facilities that depend heavily on cooling systems and electricity.
Each individual AI query may seem small. But when multiplied by billions of interactions, the environmental cost becomes significant.
Supporters argue that AI could help solve climate and water problems through better climate modelling, improved irrigation systems and more accurate drought prediction.
But the real question is whether AI can deliver those benefits faster than its own water consumption grows.
For now, that question remains unanswered.
