With power demand rising sharply and pressure mounting on energy, water and other resources, Andre Gargi, Global Head of Data Centres at Hitachi Energy, explores how operators can balance the growth of AI with increasingly ambitious sustainability goals.
Data centres are some of the most resource-intensive facilities on the planet. Now, AI workloads are set to drive a 160% increase in data centre power demand by 2030, according to Goldman Sachs. At the same time, regulatory pressure is intensifying, with environmental impact now touching every part of the business – from operating costs and customer relationships to the social licence to operate. The data centre industry faces a sustainability reckoning.
But these aren’t just environmental problems. They’re business problems, too. A recent Accenture report found that organisations that prioritise ESG generate up to 2.6 times more value for shareholders than their peers. In the face of rising energy costs, stringent regulations and growing customer expectations around sustainability, data centre operators have been forced to rethink every aspect of how they design, build and operate facilities.
There are, however, a growing number of strategies and technologies available to help reduce resource consumption while improving performance – from circular operating models to liquid cooling and renewable energy integration.
Building in energy efficiency at the design stage
Energy efficiency starts with infrastructure design. Modern data centres should be designed with energy optimisation as a core principle, not an afterthought. This means everything from server selection and cooling architecture to power distribution must be evaluated through an energy lens.
Physical infrastructure complements these digital systems. For example, we’re seeing some data centre operators becoming energy generators by tapping into renewable energy sources. The suitability of these technologies depends on the natural resources available in each region – a hot climate is well suited to solar power, while Scandinavian data centres often make use of hydroelectric or geothermal energy. By deploying battery energy storage systems (BESS), operators can also help address the intermittent nature of renewable supply.
Energy solutions can also lend themselves to circular operating models. For example, batteries that no longer have sufficient capacity for use in electric vehicles can be redeployed in BESS systems.
Keep it cool
Cooling accounts for approximately 40% of data centre energy consumption. It’s also one of the areas where innovation is having a significant impact. Traditional air cooling is being supplemented and, in some cases, replaced by liquid cooling solutions that can handle the extreme thermal loads generated by AI accelerators.
Liquid cooling can offer efficiency benefits, while also enabling higher-density compute configurations that would be difficult to support with air cooling alone. This means more computing power per square metre, which can translate into better land utilisation and smaller physical footprints. However, liquid cooling can be resource-intensive in its own way. According to the IEA, data centres’ annual water use will reach 1,200 billion litres by 2030, while construction in water-scarce regions is projected to rise by 63%.
But cooling innovation goes beyond technology. It also involves rethinking cooling strategies holistically. This includes using outside air economisation when weather permits, optimising airflow to eliminate hot spots and deploying AI-driven controls that dynamically adjust cooling based on real-time thermal loads.
Embedding sustainability across the full lifecycle
Achieving sustainability goals requires operators to consider infrastructure across its entire lifecycle. This starts with procuring systems that have clear sustainability credentials. Providers should be looking to reduce the emissions associated with production, transportation and installation, as well as operations and end-of-life recycling.
Waste reduction at the start of the lifecycle can be as simple as reducing packaging. Prefabricated systems that are assembled directly in racks can help avoid excess material use. Equally, containerised solutions can simplify construction while reducing infrastructure’s impact on land, local communities and resources. These shifts might seem relatively minor, but if implemented across the entire value chain, marginal gains can contribute to meaningful reductions in waste and emissions.
End-of-life considerations matter, too. Equipment should be designed for disassembly and component reuse. Materials should be recyclable. And operators should have clear partnerships with recycling facilities that can properly handle e-waste.
The road ahead
Sustainability presents a complex challenge for data centre operators in the era of AI. But rather than an end destination, it should be viewed as a journey of continuous improvement. As new technologies emerge and best practices evolve, operators will need to keep exploring new approaches to resource efficiency.
The data centres set up for success over the next decade will be those that view sustainability not as a tick-box exercise, but as a business differentiator. From improved compliance and lower energy costs to stronger customer relationships, the benefits of sustainable operations are increasingly clear. Through collaboration between data centre operators, energy providers and other sectors, the industry can continue to develop new approaches to sustainable energy generation, storage and use.
