The environmental impact of AI data centres

Many who frequently use generative AI might not know that this technology is causing significant environmental damage.
Artificial Intelligence (AI) is rapidly transforming industries around the world, but what you might not know is it’s also having a significant environmental impact. Most large-scale AI systems run on data centres, many operated by cloud providers. While these data centres are crucial for AI, they come with serious environmental costs.
Raw materials for AI hardware
One major issue is the large amount of raw materials required to create the hardware that powers AI. To make just a 2kg computer, 800kg of raw materials are needed. Additionally, microchips used in AI often rely on rare earth elements, which are mined in ways that can harm the environment, according to the United Nations (UN).
E-waste problems
Another concern is electronic waste (e-waste). Data centres house a lot of electronic equipment, and as technology advances, older systems are discarded. Many of these electronics contain harmful substances like mercury and lead, which can be dangerous if not disposed of properly.
Water usage in data centres
Data centres also consume large amounts of water. Water is needed both for construction and for cooling of the electrical components of the systems when they are running. In fact, AI-related infrastructure might soon use six times more water than the entire country of Denmark, home to six million people. This is particularly worrying because, according to the UN, a quarter of the world’s population lacks access to clean water.
High energy consumption
Data centres require a lot of energy to run, and in many places, this energy still comes from fossil fuels. This results in high greenhouse gas emissions. For example, a single request made through an AI-based assistant like ChatGPT uses 10 times the electricity of a Google search, according to the International Energy Agency (IEA). IEA has also estimated that in Ireland, AI-driven data centres could use nearly 35 per cent of the country’s energy by 2026.
The growing number of data centres
The number of data centres has grown dramatically, rising from 500,000 in 2012 to eight million today. With AI demand continuing to grow, experts expect the environmental impact of these centres to increase even more, according to the UN.
The need for sustainable solutions
While over 190 countries have adopted ethical AI guidelines that include environmental concerns, and both the EU and the US are working on laws to reduce AI’s environmental impact, such policies remain rare.
“Governments are racing to develop national AI strategies, but rarely do they take the environment and sustainability into account. The lack of environmental guardrails is no less dangerous than the lack of other AI-related safeguards,” said Golestan Radwan, the Chief Digital Officer of the United Nations Environment Programme (UNEP).