Data centres: can innovations reduce the environmental impacts of this asset class?

As the demand for data centres grows, how can their environmental impacts be mitigated? In the second of this two-part series, Tarrant Parsons, RICS head of market analytics, examines the innovations that could help reduce the carbon and water intensity of this real estate sector.

Data centres, used to store the physical infrastructure needed for delivering digital applications, are forecast to outpace other commercial real estate sectors in the coming few years. Driven by the rising use of artificial intelligence (AI) and machine learning across industries, the rapidly increasing demand for this sector raises understandable concerns connected to global sustainability targets.

Dominance of the digital economy will increase demand for data centres

The OECD estimates that growth in information and communication technology averaged 6.3% per annum from 2013 to 2023 (across 27 of the member nations analysed), outpacing total economic growth threefold during that period. Going forward, momentum behind the digital economy is not expected to dissipate anytime soon, with Forrester’s global forecast pointing to a near 7% compound annual growth rate through to 2028. The World Economic Forum, meanwhile, estimates that 70% of all new economic activity created over the coming decade will be based on digitally-enabled platform business models.

Aligned with this increasingly digital world, the demand for data centres will likely be propelled by numerous industries in the future. As such, from a commercial real estate perspective, data centres as an asset class display a strong medium- to long-term outlook. At a global level, projections made by McKinsey point to global data centre demand increasing by 19% (CAGR) between now and 2030.

Impact of energy demand from data centres

As data centre usage grows, so too does the vast energy consumption associated with these buildings. This is due to the intensive requirements of both processing power and the cooling mechanisms in place to deal with the heat generated by servers.

Accordingly, under their ‘medium scenario’, McKinsey projects that electricity demand from data centres based in the US (measured in Terawatt-hours) will more than triple between 2024 and 2030. That would equate to data centres accounting for almost 12% of the total US power demand by that point compared to around 4% currently.

As surprising as it may seem, electricity consumption in the US has hardly increased since 2007, with improvements in energy efficiency largely offsetting the rise in demand. Looking ahead though, amplified by the surge in AI adoption alongside continued migration to cloud services, electricity usage is set to grow significantly through to the end of the decade.

For context, the energy required for a generative AI model to respond to a prompt is typically around ten times as much as that needed when using a search engine. Linked to the energy consumption of these new technologies, Mckinsey estimates that, in the US, ‘data centre load may make up 30% of all net new demand for electricity between any two consecutive years until 2030’. Given the current global energy mix is still heavily reliant on fossil fuels (approximately 80% of energy comes from such sources), this expected escalation in electricity consumption is difficult to balance with sustainability targets.

How can data centres be more energy and water efficient?

It will be vital, where possible, to find innovative approaches to curb energy usage or reduce the carbon footprint of data centres. A critical area of focus will be around cooling technologies, given these systems are one of the largest energy sinks of data centres (normally accounting for at least 40% of electricity usage). Liquid cooling has emerged as a solution that offers several efficiency advantages over air cooling since the amount of energy required for forced convection across the latter is several times greater than that for moving liquid (for any given amount of cooling).

Data centres with high-density servers stand to benefit the most from liquid cooling, and there are several different liquid cooling techniques. These include direct-to chip cooling (where coolant is directly applied to the surface of the processor), and rear-door cooling (where coolant is circulated through a plate on the back of the server). If more widely adopted in the future, these alternative cooling techniques could play an important role in reducing the negative carbon impact of growth in data centre demand.

Still, certain types of liquid cooling methods come with pitfalls, one being the significant draw on the local water supply. This not only concerns on-site cooling, but also where water is used in powerplants generating the electricity to power data centres. Data centre location plays a crucial role, particularly for those located in areas of drought or degraded local infrastructure.

Immersion cooling, which directly submerges components in dielectric liquids that cool by absorbing heat, could reduce the water-intensity of data centres to a degree. However, it can be difficult to implement and will not be appropriate for all types of data centres. With Microsoft’s announcing plans for its operations to be ‘water positive’ by 2030, solutions to reduce water-intensity will also be a central part of data centre designs in the future.

Alternatively, another environmentally preferential option would be to take the heat byproduct from data centres and redirect this towards secondary uses, creating a more circular energy system. For instance, with the help of heat pump technology, the heat from data centres could be utilised either for nearby purposes, or to provide hot water to homes/commercial buildings situated further away via heat networks connected to cities. Additionally, excess heat could also be used for greenhouse farming in some locations.

Examples of green data centres

The ultimate ambition would be to create widespread and replicable blueprints for data centres to operate at net zero, running mostly or entirely on renewable energy, with minimal negative impacts on the environment.

At present, some of the most utilised renewable energy sources for data centres include solar power, wind energy and hydrogen power, all of which provide a cleaner energy supply. In fact, there are already many leading facilities that are considered ‘green data centres’ and, as much as possible, future projects should seek to incorporate the most practical sustainable elements of their design:

• Meta's Luleå data centre, for instance, boasts being entirely run on renewables, with natural cooling enabled by the cold climate in Northern Sweden.
• Scala Data Centers is a hyperscale data centre operator in Latin America. It is the first in the region to use 100% certified renewable energy and shows that such data centres can operate even in areas that do not have the advantage of cold temperatures.
• Virtus (based in London) has achieved 100% renewable electricity usage across all its sites, it has establishments that are in the top 10% of UK commercial buildings by energy efficiency and aims to be net zero by 2030.
• Aruba’s Global Cloud Data Center (IT3) is another impressive example. It is self-powered and zero-impact, making use of a hydroelectric plant, solar panels and chilling underground water.

There are many obstacles that prevent these exemplary models from simply being reproduced across all locations, but, there are still plenty of lessons can be taken to at least strive towards a more sustainable approach.

There will be no low carbon economy without decarbonised data centres

Data centres will be integral to our digital future, but their intensive power and water demands present a huge challenge to global sustainability goals. Addressing these will require a multi-pronged approach encompassing further technological innovation, renewable energy integration and water management, along with well thought-out government policy, regulation and guidance. As the world transitions to a low-carbon economy, data centres will need to both adapt and lead by example given their ever-increasing importance.

Related articles