Data centres and AI are straining natural resources
What’s happening? The US-China tech trade war and the rise of AI have spotlighted the immense resource consumption of data centres. Gallium prices have surged due to the trade conflict, affecting semiconductor manufacturing. AI models such as ChatGPT drive significant water and energy use, with data centres consuming up to 760 million litres of water annually, an issue exacerbated by severe global water stress. To mitigate the impacts, the data centre industry must adopt efficient cooling strategies, renewable energy with lower water footprints, and extend product life to reduce supply chain resource usage. (Techerati)
Why does this matter? The environmental impact of AI technologies is a pressing issue as the world grapples with the urgency of the climate crisis. Data centres, essential for operating AI models such as ChatGPT, contribute significantly to greenhouse gas emissions, surpassing even the aviation industry in their carbon footprint. The training of ChatGPT-3 alone consumed an estimated 700,000 litres of water for cooling. A Cornell preprint suggests that water withdrawal for AI could reach between 4.2 – 6.6 billion cubic metres by 2027, roughly half the UK’s annual water consumption.
Microsoft’s massive emissions – Since 2020, Microsoft’s carbon emissions have increased by nearly 30%, primarily due to the construction of more data centres to support its cloud services. The company aims to mitigate this impact by sourcing 100% carbon-free electricity by 2030, improving data centre efficiency, and investing in renewable energy and carbon removal projects. Additionally, Microsoft is focusing on using greener building materials and advocating for policy changes to support its goal of becoming carbon-negative by 2030.
Mineral demand – The extraction of minerals essential for data centres, including gallium for semiconductors and lithium and cobalt for batteries, poses additional environmental and ethical dilemmas. Mining processes often lead to significant water pollution and are frequently associated with human rights violations whilst also straining the finite supply of natural resources. Such activities undermine efforts to achieve sustainable development goals and compromise water security for vulnerable communities.
Grid challenges – Moreover, the growing demand for electricity by data centres is straining outdated power grids, hindering the development of affordable housing and other essential infrastructure. The issue is exemplified in the UK, where the electricity network’s capacity constraints have paused the construction of thousands of new homes. As households increasingly rely on renewable electricity over fossil fuels, the pressure on national grids will only intensify, highlighting the need for more sustainable and efficient energy solutions.
EU regulation – To combat this ongoing trend, the European Commission has adopted a new delegated regulation to establish an EU-wide scheme for rating the sustainability of data centres, requiring operators to report key performance indicators starting 15 September 2024. The initiative aims to increase transparency and promote efficient designs to reduce energy and water consumption, enhance renewable energy use and improve grid efficiency, following extensive consultations and public feedback. The regulation is part of the broader effort to achieve the EU’s new binding target of reducing energy consumption by 11.7% by 2030, with data centres currently accounting for nearly 3% of EU electricity demand.