Disclaimer: Views in this blog do not promote, and are not directly connected to any Legal & General Investment Management (LGIM) product or service. Views are from a range of LGIM investment professionals and do not necessarily reflect the views of LGIM. For investment professionals only.

16 Jan 2025
3 min read

Why is AI so thirsty?

In the first part of a new series on AI and water, we explain why these systems could soon be consuming around five times as much water as Denmark every year.

clean water wave

It would be hard to miss the rapid deployment of artificial intelligence (AI) that’s taking place today. An estimated 82% of large enterprises have either deployed AI or are experimenting with doing so.[1] 

The rapid expansion of the digital sector reflects how our society and economy are evolving. As well as being essential to so many of our everyday activities, we believe the infrastructure underpinning digitalisation is critical to key societal goals such as increased financial inclusion and creating a more environmentally sustainable world.

An aspect of the rise of AI might be less obvious is that every ChatGPT query triggers an ocean of calculations, necessitating clean water to cool the servers that crunch the numbers.

Generative AI (GenAI), which uses large language models (LLMs) to create text, images and other outputs, requires the use of huge server farms that use chilled water to cool equipment by absorbing heat from the air. Once water has been used to cool data centres, some of it is recirculated to the cooling system several times before being discharged, while some evaporates in the cooling process.[2] 

Datacentre water consumption in numbers 

As Google parent Alphabet* and Microsoft* prepared their LLMs in 2021 and 2022, both companies saw major spikes in water consumption, with annual increases of 20% and 34%, respectively.[3] 

Google data centres consumed around 20 billion litres of water in 2022, roughly equivalent to the annual water consumption of 2.5 million Europeans, or 1.2 million Americans.[4] 

On average, depending on weather conditions and operational settings, data centres can evaporate about 1-9 litres per kWh of server energy for cooling purposes.[5]

The supply chain effect

AI’s thirst for water isn’t confined to data centre on-site cooling. 

Off-site water usage during power generation is another demand driver. In 2023, US data centres had an indirect water footprint of nearly 800 billion litres due to electricity use.[6] The electric power sector accounts for about 40% of total water withdrawals in the US, as thermoelectric plants require significant cooling and hydroelectric plants lose water through surface evaporation. 

There is also water consumption associated with AI supply chains. Semiconductors and microchips require large volumes of water in the manufacturing stage, with a single 12-inch wafer layer produced by TSMC* consuming around 60% of the average person’s daily domestic water use in Taiwan.[7]

A roadmap to 2030

The total water consumption by data centres (including water used on-site for cooling, and off-site for power generation) globally rose 6% per annum from 2017 to 2022 and is estimated to reach 450 million gallons a day by 2030, according to Bluefield Research. 

This makes data centres one of the fastest-growing water verticals. Academics suggest that AI demand will drive up water withdrawal, where water is removed from the ground or surface sources, to between 4.2 billion and 6.6 billion cubic metres by 2027, equivalent to half of the UK's annual water consumption, or around five times Denmark's yearly water usage.[8]

In the next part of this blog, we’ll examine some innovative water technologies and the offerings of three real-life companies to improve the water usage efficiency of data centres. 

* For illustrative purposes only. Reference to a particular security is on a historic basis and does not mean that the security is currently held or will be held within an LGIM portfolio. The above information does not constitute a recommendation to buy or sell any security.


 
[1] Source: https://newsroom.ibm.com/2024-01-10-Data-Suggests-Growth-in-Enterprise-Adoption-of-AI-is-Due-to-Widespread-Deployment-by-Early-Adopters
[2] Source: Data Center Water Usage: A Comprehensive Guide - Dgtl Infra
[3] Source: Generative AI’s environmental costs are soaring — and mostly secret
[4] Source: Data Center Water Usage: A Comprehensive Guide - Dgtl Infra
[5] Source: AI's Challenging Waters | Civil & Environmental Engineering | Illinois
[6] Source: https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report.pdfUnited
[7] Source: https://esg.tsmc.com/file/public/e-APractitionerofGreenPower_2.pdf ; per capita water consumption via Statistica
[8] Source: AI boom sparks concern over Big Tech’s water consumption

Index equity Index thematics Responsible investing Investment stewardship ETF equity ETF thematics Index ESG Technology Environment, Social and Governance
Shichen Zhao

Shichen Zhao

Thematic Research Analyst

Shichen is a Thematic Research Analyst in the ETF team, contributing to growing the thematic equity ETF range from an investments, research and analytics perspective.…

More about Shichen

Recommended content for you

Learn more about our business

Legal & General Investment Management is one of the world's largest asset managers, with capabilities across asset classes to meet our clients' objectives and a longstanding commitment to responsible investing.

Image of London skyscrapers

Sign up for blog email alerts

Receive the latest articles in a weekly digest by registering via the email preference centre