Our thinking

AI? Too hot to handle!

23 August, 2024

The “hidden footprint” of tech is increasingly visible, with the carbon impact of AI an increasing cause of concern. Today we’re talking about water. Water is increasingly front and centre of environmental discussions – the use (and abuse) of it affects nature, human wellbeing and climate, while climate change impacts where water is found, and where it is scarce. As we use more data, and turn to AI to solve our challenges, this puts a huge strain on water resources used in cooling processes. For tech optimists this is justified by an opportunity to increase efficiency in water intensive processes and ultimately reduce water stress globally. However, looking at current statistics, we’re struggling to come around to that view.

North Virginia has a tech hub known as “data centre alley”, home to the likes of Google, Amazon and Microsoft. In this region, water consumption by data centres has grown by two thirds since 2019 and continues to rise. In fact, US data centres consumed the same amount of water in 2023 as the whole of the UK does in 4 months, according to research group Dgtl Infra. While water extraction is not necessarily a huge concern everywhere, Virginia has a history of dry spells, particularly in recent years and most recently in 2023. To give further context, 42% of the water that Microsoft consumes globally is from “areas with water stress”. Microsoft does have a commitment for the data centres to become “water positive” by 2030, but is that good enough when droughts are happening now?

The reason we think it’s so important to shine a light on this issue is that the footprint of tech really does feel hidden compared to an industry selling, say, food or plastics. And as companies set targets to reduce their social and planetary impacts, are they thinking enough about the impacts of their data and technology?

By Anna Heis