Confronting the AI Data Centre Sustainability Dilemma

Whilst digital technologies and artificial intelligence (AI) are contributing to an increase in energy consumption and emissions for data centres, they could also work to offer significant potential for improving sustainability across a wide range of business sectors. 

Steven Brown, Vice President, Digital Energy Solutions at Schneider Electric spoke on this subject at Sustainability LIVE London 2024. He touched on the complex relationship between digital technology and sustainability and how businesses could use the technologies to drive significant improvements in sustainability.

Data Centre Magazine speaks exclusively with Steve about how the world increasingly relies on digital infrastructure, including data centres and AI, along with how this impacts our global footprint.

AI is ‘hungry for data’

From carbon costs to the role of edge computing in building management, Steven explains how there are both challenges and opportunities presented by our global digital future.

“Everything has a footprint, or an energy-intensive use case to store and sync information,” he says. “Researchers at IET have estimated that there is 10.5kg of carbon dioxide per-year-per-person for unwanted photos. Digital has a link that feels a little bit nebulous with sustainability.”

When it comes to single-use data, the volume continues to increase, with 68% of stored data never being used again after being created. This is significant, as this unused data continues to consume energy - with examples including spreadsheets, word documents and recorded telephone conversations.

“Photos also illustrate the impact of this data, given that AI is going to be powerfully important. A photo generated by generative AI (Gen AI) could use much more energy,” Steven comments. “AI could even use as much energy as charging film for a single image depending on the model and how efficient it is. 

“So now we're in this world of AI where suddenly we need to deliver a lot of capacity to service this need for digital compounded with AI and against constraints certainly in the market. It requires a lot of compute, a lot of data centres.”

AI and Sustainability: Striking a balance

The impact of AI on global sustainability is already huge. Already in September 2024, it was reported that data centre emissions from the leading tech giants may be 7.62 times (662%) higher than official figures. This is largely on account of their AI-led data centres which consume much more energy.

However, as Steven explains, there could be a solution that sees these disruptive technologies being used to mitigate this impact.

“Sustainability is a clear use case for AI, especially as we get into a more complex energy management environment. Yes, AI requires compute, but at the same time, how can we apply this to make our society much more decarbonised?” He says. “Around 45 gigatons of carbon dioxide are released each year, in addition to methane and other gases making up around 80% of emissions. That's cause for concern, especially when you see the dramatic growth in the last 50 years

“Through the various commitments of different countries and companies, we are on a path to reduce emissions by four gigatons of carbon dioxide per year, which by 2030 would have us on a much better trajectory, but it’s still not fast enough.”

Steven suggests that it is the power of digital technology that will help the world on its path to sustainability sooner.

“At Schneider Electric, we call it Electricity 4.0, but it's using digital infrastructure, digital technology to become more efficient and then it's electrifying where we haven't yet to decarbonise,” he says.

“Electrical continues to be the most efficient form of secondary energy use. There are so many opportunities to remove heat related engines to electrical, so digital has become more efficient. Electrify, whether it's EV vehicles, boilers to heat pumps, plenty of opportunities to electrify those two together are, or electricity four, sustainable green energy that's going to help us decarbonise and go three times faster.”

With data centres around the world using more electricity than ever before, companies are having to quickly prioritise ways to be more energy efficient. According to the International Energy Agency (IEA), global demand for electricity grew 2.2% in 2023, highlighting that it could double as soon as 2026.

With such demand, Steven explains that companies are having to find ways to make their electrical loads carbon-free.

“Again, we have to look at the use cases for digital. These have major applications for decarbonising buildings, helping microgrid infrastructure and smart grid infrastructure,” he explains. “We need that digital infrastructure to help us be more efficient and reach net zero.”

Decarbonising data via its infrastructure 

In recent times, there has been a rapid acceleration of cloud-based technologies, with cloud compute technologies now being in very high demand. This movement is in part led by hyperscale data centres that seek to manage the technology.

With the emergence of AI, Steven suggests that there will be a re-emergence of edge infrastructure where data centres will need to deploy workloads close to where they are being used.

“Once a model is trained, that has to be deployed as an inference model where you get much more agility,” he says. “We've seen cloud AI, a lot of concentration, and now we're going to have to deploy that.”

In order to continue decarbonising, Steven suggests that buildings and infrastructure could need to be updated in order to innovate in a cleaner way.

“When we think about connecting a building today, many of them aren't particularly digitised,” he notes. “Especially in this modern world of working from home and hybrid environments, it all changes the dynamic of how you can make a building more efficient. Building at the edge requires the kind of critical infrastructure to protect it that you might expect in a data centre environment.

“Some of the biggest gains come from just deploying automation and digital technology. You need that edge infrastructure to connect the data. You need cloud to make those analytics and AI decisions. And then you need back on-premises for that control and automation, applying AI insights to make a building run more efficiently. Through this, you can be on the fast track towards a decarbonised building.

Steven describes his ethos at Schneider Electric: “Let's use digital to decarbonise buildings and let's check in to make sure that you're using efficient cooling, that you're monitoring and measuring a data centre and as much as possible controlling that to be as efficient as it can be.”

******

Make sure you check out the latest edition of Data Centre Magazine and also sign up to our global conference series - Tech & AI LIVE 2024

******

Data Centre Magazine is a BizClik brand