News
Why Is Water Use In Focus With AI Data Centres
When you think about the advances we’ve seen in artificial intelligence (AI) technology in the past year, what likely comes to mind are tools like ChatGPT and an opportunity to automate a range of tasks.
However, AI data centres are increasingly under the spotlight for the significant amount of resources they use and, more specifically, water.
Scotland is one location in the UK where data centres to power AI services are popping up and a recent BBC report revealed that these data centres are using enough water to fill 27 million bottles of water every year.
A Freedom of Information request by the broadcaster revealed that the amount of water being consumed by Scotland’s data centres has quadrupled in just four years. Scottish Water has already signalled that it is working with data centre providers to help them find ways to reduce their water use.
Why do AI data centres use so much water?
One of the main reasons why AI data centres use so much water is because the servers and other computing equipment needs to be kept within a specific temperature range to prevent damage to the equipment and ensure it functions optimally.
In many cases, AI data centres use an open-loop system, which means they cool the air in the server rooms through a process of water evaporation. This increases the amount of water they use as it doesn’t stay in the system.
If you also factor in the water consumption that occurs to produce the power needed to support these data centres, the amount of this resource being used by this growing technology is even higher.
Why do we need to reduce the amount of water being used?
Although it might be easy to think that the UK, and particularly Scotland, has a plentiful supply of water, this isn’t necessarily true. Due to climate change, our weather patterns are becoming less predictable and that is having an impact on the reliability of our water supply.
As the Energy Saving Trust points out, many parts of the UK are already suffering from what’s known as water stress, which means the supply of fresh water is struggling to keep up with demand from households and businesses.
Businesses like AI data centres, which use a considerable amount of this precious resource, therefore have a duty to find more efficient ways of operating for the good of everyone.
The good news is that, overall, industrial and commercial water use has been declining in the UK because of more efficient processes and technology. There’s no reason why the same trend can’t happen in AI data centres if their owners and operators take responsibility for their water use.
How can AI data centres reduce the amount of water they use?
There are several alternative options to cooling the servers in AI data centres, such as closed-loop systems that target their cooling efforts on the server racks alone. This means that they use less water as they are trying to cool a much smaller area.
Of course, there are a multitude of other ways in which businesses, and not just AI data centres, can reduce their water consumption. Working with experienced water consultants is a good place to start so that you can identify all the places in which water is being wasted and then seek solutions.
Another option for AI data centres could be to source their water through processes like rainwater harvesting or water recycling. This would mean they wouldn’t be taking water directly from the mains system and instead would be recycling or collecting water on their own sites.
Which of these options is appropriate will depend on a range of factors, which is why working with experts in water systems is essential to ensure that resource use is maximised.
This brings benefits beyond simply the environmental ones of reducing the strain on a sought-after resource. For a business, reducing water consumption from the mains supply will reduce bills, which in the case of AI data centres can be significant.
So, while it can be more costly to install more efficient cooling systems in the first instance, this will be recouped through lower water bills once the centre is operational. In addition, we all – as individuals and businesses – have a responsibility to manage such resources responsibly.
All of us rely on having access to fresh water to survive, so firms setting up AI data centres need to consider how they can minimise their impact on both the planet and the communities in which their facilities are located.