The AI boom has supercharged the computing sector’s energy use, prompting an increase in demand for data centres along the way. It’s a trend that shows no sign of abating, with more of the same expected for 2025 and new research showing that hyperscale data centre capacity is set to triple by 2030.
The increasing energy demands of data centres are raising significant concerns, particularly due to their high-power consumption, environmental impact, and the financial strain they impose on operators.
However, the cause of the problem could also be the cure: AI could be the very innovation that could lower consumption, using data to spotlight areas where processes could be improved.
In this article, I’ll explore how AI can help optimise data centre processes, what the data centre industry is currently doing, and other innovative pathways to data centre efficiency.
How the data centre industry is currently tackling the problem
Data centres require vast amounts of electricity to run and keep cool. This is costly to businesses, but it’s not just a concern from a financial point of view—it’s also an environmental one.
There have been strides made over the past decade to ensure data centres operate more effectively and efficiently, but innovative approaches will be required to ensure new technology doesn’t derail efforts to reduce IT infrastructure’s carbon footprint. Maintaining the 24/7 uptime we all expect and need creates huge operational energy costs.
Data centre operators have started to make inroads into this issue by optimising existing systems, investing in cutting-edge cooling technology, for example. They’re also opting for GPU hardware that reduces energy consumption, both of which can have a significant impact on overall usage.
There’s also been a move towards investing in renewable energy sources such as wind, solar, and geothermal. Not only does this reduce energy consumption by powering data centres through means other than fossil fuels, but it also removes reliance on the national grid. And it helps save costs in the long run, making it a win-win. Another possible approach is to build data centres in colder locations, where atmospheric cooling can be used to help control the centre’s temperature.
Could AI be the golden ticket?
The scale and growing demand of AI energy usage means that sustainability efforts alone won’t be enough. AI can play a role in helping improve the energy consumption and efficiency of data centres.
For example, AI can be used to analyse data centre energy use and suggest new ways to keep data centres cool. The technology could leverage historical data to predict workloads and monitor and adjust the cooling system in response. It can even help maintain equipment by allocating resources based on real-time demand. This helps extend infrastructure lifespan and minimise waste, reducing the carbon footprint and spending simultaneously.
AI is also helpful in predicting potential equipment failures. By analysing sensor data and historical patterns, AI can highlight where proactive maintenance is needed, which minimises downtime. This improves operational efficiency.
The efficiency of AI models themselves is also another key area of development. We know that the quicker and smarter the AI model’s operation is, the lower the AI’s power consumption will be. Therefore, AI developers and coders can play a significant role in controlling energy usage by developing more efficient AI models.
At its best, AI is intended to take some work off our plates and make our lives easier. It can also do this for data centres by cutting energy and financial costs.
Looking to the future
It’s likely that we’ll see many hyperscale data centre operators turning to higher-capacity innovations, including hydrogen fuel cells and nuclear power generation. Initially, this might sound like an outlandish solution, as hydrogen continues to be a niche power source. But it does have widespread availability.
Nuclear energy also offers a compelling solution, although its risk profile and political sensitivity do pose challenges. Google, Amazon, Microsoft, and Meta are among the most recognisable names exploring or investing in nuclear power projects. These different sources of energy should be considered as high-capacity, cleaner alternatives to oil or natural gas, and are worthy of serious attention.
As we move forward, addressing the energy demands of AI while also securing a sustainable future for the planet will require collaboration among not only the IT and energy industries, but with governments as well. AI use cannot come at the cost of increased global heating or deforestation; alternatives to fossil fuels must be prioritised, along with research and development on energy use reduction.
