Data centres powering artificial intelligence (AI) in Scotland are consuming a staggering amount of water, equivalent to filling 27 million half-litre bottles each year. This significant water usage, primarily for cooling the powerful computer equipment required for AI systems like ChatGPT and Google's Gemini, has quadrupled since 2021, raising concerns about the sustainability of Scotland's burgeoning AI ambitions.
Key Takeaways
AI data centres in Scotland use enough tap water annually to fill 27 million 500ml bottles.
Water consumption by these centres has quadrupled since 2021.
The majority of centres use 'open loop' cooling systems, which require constant mains water supply.
Scottish Water is urging the sector to explore sustainable alternatives like wastewater systems.
The carbon footprint of these data centres is also a significant concern.
Escalating Water Demand
Data centres, essential for powering everything from online banking to movie streaming, have seen their water and energy demands surge due to the boom in generative AI. While Scotland currently has 16 operational data centres, this number is expected to grow. Freedom of Information data obtained by BBC News reveals the dramatic increase in tap water usage, prompting Scottish Water to describe the trend as "significant."
Despite this increase, Scottish Water notes that data centre water consumption still represents a small fraction, approximately 0.005%, of the total water supply. However, experts from the University of Glasgow have highlighted the broader implications, calculating that the water consumed is equivalent to every person in Scotland drinking an extra 2.48 litres per year. Furthermore, the carbon footprint associated with these data centres is estimated to be equivalent to every person in the country driving an additional 90 miles annually.
Seeking Sustainable Solutions
Colin Lindsay, operations manager at Scottish Water, expressed concern over the reliance on "precious tap water" for industrial cooling. "We would like to try to look for other alternative solutions rather than using precious tap water," he stated, advocating for the exploration of sustainable alternatives such as wastewater systems.
Currently, most data centres in Scotland employ "open loop" cooling systems, which necessitate a continuous supply of mains water. These systems draw in fresh water, use it to cool the servers, and then discharge the warmed water. The industry is gradually transitioning towards more efficient "closed loop" systems, which recirculate a fixed amount of water. However, this shift can sometimes lead to increased energy consumption, presenting a trade-off between water conservation and power usage.
Scottish Water is encouraging developers to consider locating open-loop facilities near wastewater treatment works, where treated effluent could be used as a sustainable water source, thereby minimising demand on public drinking water and potentially reducing energy use.
The Broader Environmental Impact
Experts also point to the substantial carbon footprint associated with AI data centres. Professor Ana Basiri from the University of Glasgow's Centre for Data Science and AI noted that the energy consumption of AI tools like ChatGPT is significantly higher than that of a simple Google search. She stressed the importance of transparency, stating that the lack of mandated reporting on energy and water usage by data centres poses a significant challenge.
The UK government considers data centres critical national infrastructure and is investing in infrastructure to meet future demands, viewing AI development as crucial for economic growth. However, the rapid expansion of AI necessitates a sustainable approach to ensure that growth does not compromise Scotland's natural resources or its net-zero ambitions. The Scottish government encourages developers to adopt water-saving measures and sustainable alternatives.