AI is on the rise. What started off as initial hype a few years back, has quickly developed into the generative AI revolution we’re seeing today. In fact, AI has taken off so rapidly that it is the key driver for billion-pound investments now being poured into AI infrastructure.
With AI workloads requiring unprecedented levels of compute density, these investments are necessary to be able to ensure there is enough computing capacity to handle the increasing power demands of AI, now, and in the future. As a result, the UK, like many other countries, is taking up its position in the race to become an AI superpower. Within the next five years, the number of UK data centres is predicted to grow by nearly 20%, with major AI players such as Google and Microsoft already having pledged to invest hundreds of millions of pounds to build new infrastructure.
But the need for more computing power comes at a price. More power means higher energy consumption, and AI is known for using a significant amount of both. In fact, recent research from Gartner revealed that energy demand for data centres is set to double by 2030. The resulting heat output in turn increases reliance on advanced cooling technologies, such as water-based systems, adding to the pressure on limited resources.
With data centres expanding and capacity increasing at pace, the industry has come under pressure from both inside and outside, raising concerns about the sector’s energy usage and carbon output. These concerns are valid. For example, Google’s proposed new Essex facility is expected to emit about 570,000 tonnes of CO2 a year, equivalent to 500 short-haul flights every week.
On a global scale, this figure is even higher. In 2026, data centres could be using a total of 1,000 terawatts hours. That figure roughly equals the energy consumption of Japan – that's a population of 125 million people.
Conflicting pressures
For data centre operators, this means being pulled into two different directions. On the one hand, they are under pressure to deliver the infrastructure needed to support AI-driven growth; on the other hand, they need to respond to stricter sustainability regulations and rising customer demand for better environmental performance.
These challenges didn’t start with AI. The data centre industry has been under constant scrutiny for years. As a result, regulatory oversight has been steadily increasing and rules
such as the Energy Efficiency Directive (EED) now require operators to be much more transparent about their sustainability goals and use of renewable energy sources.
Customer expectations have shifted as well, with sustainability now being a top priority for many. As such, sustainability credentials and certifications such as ISO 14001:2015 are often among the deciding factors in many procurement processes. This shift in mindset shows that compliance alone is no longer enough; operators must demonstrate continuous progress towards their set sustainability KPIs.
AI poses a challenge in this context. The United Nations Environment Programme (UNEP) has pointed out the heavy environmental impact of AI and data centres specifically, due to the significant power and cooling demands involved. The biggest challenge for operators lies in providing the necessary infrastructure to support higher performance requirements without putting unsustainable strain on their infrastructure or local energy networks.
Raising the bar on environmental responsibility
These pressures are serious, however, every effort is being made to address them. For example, new approaches to cooling are emerging, including water-cooled servers and fully water-cooled racks, which aim to manage heat more efficiently and reduce overall energy use.
Using renewable energy sources for power generation is another area of focus. Many operators are working hard to cut reliance on traditional sources and opt for greener options instead. While this sounds promising, the gap remains significant. The International Energy Agency estimates that data centres currently obtain around 27% of their electricity from renewable sources. With planned expansion, renewables are expected to account for roughly half of the sector’s demand by 2030.
Meanwhile, sustainability is becoming a point of competition. With public awareness and customer scrutiny increasing, operators must differentiate themselves through their environmental performance and that of their suppliers. For some providers, sustainability now carries as much weight as availability, resilience or security.
There are certainly no quick fixes to the challenges faced by the data centre industry. While things are pointing in the right direction, it will require sustained effort by the industry as a whole to balance growth with environmental responsibility. As demand for AI infrastructure continues to rise, the sector’s ability to meet both its commercial ambitions and environmental obligations will be closely watched.
