Green giants: Engineering sustainable and resilient data centres

The explosive growth of AI is driving an unprecedented demand for data centre capacity across the UK, pushing power density and consumption to their limit. This rapid expansion presents a crucial dilemma: how do we meet this energy-intensive demand while simultaneously achieving rigorous net zero and sustainability targets? Stephen Phimister, director of engineering at independent MEP, sustainability & ESG consultancy Wallace Whittle discusses how to engineer data centres that are both designed for longevity and a greener world.  

Earlier this year, construction researchers Barber ABI revealed that the number of data centres in the UK was set to increase by almost 20%. With a current estimate of around 500 already in the UK, this could mean around 100 of these giant warehouses built over the next five years – mostly driven by the growth in AI and the need for more processing power.

Of course, as a nation that is also committed to reaching net zero in the next 25 years (sooner for some local authorities) this explosion of growth has to be balanced by sustainable measures – both for environmental reasons, and for resilience.

AI demands

With the UK AI market predicted to grow to £1 trillion by 2035 (International Trade Administration, 2023) this has a direct impact on data centre demand. One of the most talked about challenges within AI data centres is the high levels of heat generated by the energy-hungry beasts, and exactly how to implement sophisticated cooling strategies that can handle it. These solutions must balance technical performance with sustainability, cost, and long-term planning, making cooling one of the most critical aspects of modern data centre design.

Traditional air cooling on its own is no longer sufficient to manage the intense heat generated by high-density AI GPU racks. Neither is open loop cooling systems that result in high water demand. This drives a shift toward liquid cooling solutions, or a hybrid of air and liquid solutions. Critically these systems are closed loop and therefore counter the inaccurate media hysteria around water consumption.

This industry transition is evident in the widespread adoption of technologies like direct-to-chip cooling and immersion cooling, often preferred for larger AI deployments. These advanced cooling techniques are vital for maintaining operational efficiency, reducing energy waste, and making data centres truly "AI-ready." AI workloads are not suited to deployment in data centres which use traditional cooling techniques. Furthermore, the increased cooling demand makes strategic location in cooler climates a significant factor, offering opportunities for substantial energy savings.

Given that AI data centres operate 24/7, the nature of these critical workloads demands equally robust power infrastructure. Resilient energy strategies are essential to prevent downtime, meaning that designs must prioritise power reliability, redundancy, and sustainable energy solutions. This focus on uninterrupted, reliable, and efficient power delivery, alongside the implementation of cutting-edge liquid cooling, forms the foundation for supporting the massive growth of AI computing while striving for environmental accountability.

Grid stress

Another concern for large data centres, is the energy usage and impact on electricity infrastructure in surrounding areas. In Ireland, data centres account for more than a fifth of all electricity consumption (CSO 2024), and in the UK, the National Grid anticipates a six-fold increase in commercial data centre demand by 2034 (Data Centre Review). In conjunction with the need for connecother sustainable connections it led the National Energy System Operator (NESO) to entirely reform the grid connection process.

A key decision in locating new data centre development is to consider the carbon emission factors for the electricity grid that they are connecting to, preferring connections to grids that have significant electricity generation from renewable sources.

This lack of available grid connections is forcing developers to think not only about off-grid or part load connection, but whether there are alternative power sources. Micro-grid solutions are seen as they way forward, incorporating a mix of renewable sources including wind, solar PV and battery storage. Gas-powered systems are making a comeback, and although there are environmental questions to be answered a gas solution has significantly lower emissions than the traditional diesel engines used for standby applications. These micro-grid solutions can also offer support back to the transmission systems by running the data centre independently at times of peak grid demand.

The imperative for sustainable data centre design is also pushing the sector towards renewable energy integration and the circular economy, with global business leaders looking to pursue large-scale, resilient power solutions. While large-scale solar (PV) integration alone is insufficient for high-density AI data centres, a more promising approach involves direct links to renewable generation sources, such as connecting facilities to nearby wind farms. Regions with high renewable energy resources, such as Scotland with its extensive wind power, are uniquely positioned to adopt this model and ensure that energy consumption is directly matched with green generation.

Beyond sourcing clean power, the opportunity lies in waste heat recovery; capturing and reusing the significant heat byproduct from server operations is a vital circular, efficient solution that can reduce energy consumption, lower carbon emissions, and even provide heating for nearby communities. This practice, already mandated for large facilities in the EU under the Energy Efficiency Directive (EED), is likely to see wider adoption, including eventually in the UK.

Looking ahead

With such a changing climate and more challenges to resilience, the AI evolution demands consistent enhancements and adaptation. Not only from teams with MEP and sustainability, but also looking at how to support energy infrastructure and the rising threat of cyber attacks. Data centres are particularly critical making it essential for both physical and digital long-term security.

In order to support the sustainability drive within data centre design, to support all the challenges that are on the horizon, it will require more collaboration and expertise from a range of sectors. Changes to regulations will also be important to guide these transformations. And as AI demands continue to shape the future of data centres, MEP and ESG considerations are key to ensuring efficient, sustainable, and resilient infrastructure.