Exclusive: How sustainability challenges are putting data centres under pressure

Data centres are expanding at breakneck speed across the globe, driven by the explosive rise of artificial intelligence (AI).

But experts are warning that unless sustainability and resilience are prioritised now, the long-term viability of these facilities, and the services they power could be at serious risk.

"AI is the single most disruptive technology that has the ability to have a positive impact on society in terms of bringing better services to everybody in a more sustainable way," explained David Mudd, Global Head of Digital Trust Assurance at BSI.

Mudd, who has worked across the electronics, telecoms, and digital trust sectors, has been with BSI for nearly eight years. His work focuses on ensuring digital infrastructure, including data centres, is robust, secure, and sustainable.

But while AI holds promise for society, it brings a sharp increase in demand for data storage and processing.

"Every time you take a photo or send a message, that's more data being stored," he told TechDay during a recent interview.

"Generative AI in particular sucks up vast quantities of data and requires huge processing power—often 1,000 times more than a standard internet search," he explained.

This surge is triggering an unprecedented boom in data centre construction, from massive hyperscale facilities to smaller, localised edge centres.

"We're seeing data centres like mini cities, consuming the power of a small town," Mudd said. "But we're also seeing more of them in more places, driven by demand and by national data sovereignty concerns."

That, he added, is where the sustainability challenges become urgent. "Before we even start on efficiency, we have to ask - is there power even available to build these facilities? In some regions, there's already a three-to-five-year waiting list for power connections."

Beyond electricity, water is another growing concern. "We've seen headlines calling AI 'thirsty', and it's not far off. Data centres use vast amounts of water for cooling - either directly or indirectly through the power stations that feed them," Mudd said.

Carbon emissions from energy use and construction materials such as concrete are also part of the equation. The risk, he warned, is that in the rush to meet AI demand, corners could be cut. "There's a real pressure to get data centres up fast. But what gets built today will be around for 30 or 40 years. If shortcuts are taken, we're risking their long-term availability and effectiveness."

That's especially important given how critical data centres have become to modern life.

"They're now part of our critical infrastructure. Banking, healthcare, utilities - all rely on them," he explained. "In the UK and Europe, data centres have been formally designated as such."

Climate change poses additional risks. "We've already seen sustained high temperatures in London leading to increased outages," said Mudd. "It's not just hotter places like Dubai where we need to worry. Even relatively temperate areas are seeing extreme events like floods and heatwaves that strain existing infrastructure."

Good location choices and careful design are essential. "You might need to build near a high-tech hub, but is there power? How will the local community respond? What are the climate risks - floods, heat, lightning?" he said.

International standards play a key role in getting this right. "Standards like the European EN 50600 series and its ISO equivalent, ISO/IEC 22237, provide globally agreed best practice across the entire data centre lifecycle - from design and construction to maintenance and eventual end-of-life," Mudd said.

"This isn't just one viewpoint - it's 100 countries agreeing on what good looks like."

While these standards don't solve everything, they help align stakeholders and enable trust across a global industry. "No one organisation has all the best answers," he added. "Having a common language helps everyone work together more effectively."

Reducing water and energy consumption, particularly for AI workloads, requires both conventional and advanced solutions. "From evaporative cooling towers and cold aisle containment to on-chip liquid cooling, there are options at every level," Mudd explained. "Even just having solar panels and wind turbines on site is something every organisation should consider."

Still, Mudd cautioned against focusing solely on the data centres themselves. "They're only one part of the picture. We also need to rethink software design, telecom infrastructure and societal expectations."

"There's been this idea that data creation, storage and processing are free - just like we used to think of energy. That has to change."

According to Mudd, talent shortages are "another hurdle."

"The data centre industry is facing the same workforce crunch as the broader ICT sector," he said. "New markets especially are struggling to build that critical mass of expertise. That's why it's vital to engage with universities and young engineers now."

His message to those considering a career in data centres is simple: "Without data centres, there would be no AI. If you're helping to design and operate them, you're enabling a smarter, more inclusive, and sustainable society."

So what should developers be doing right now?

"Look at industry best practice and understand the long-term risks of a short-term mindset," Mudd said.

"We've got a one-off chance to get this right for the next 30 years. The opportunity is now."