XXXXXXXXX ↓
XXXXXXXXX ↓
XXXXXXXXX ↓
XXXXXXXXX ↓
TRENDS
AI adoption accelerates shift to high-density data centres
As consumers’ and businesses’ digital requirements evolve and AI becomes more embedded in everyday life, the data centre market is innovating to keep up with the pace of technological change. One element of this is the fast-growing demand for the development of high-density, liquid-cooled data centres designed to support AI workloads.
Our report shows that 75 percent of those surveyed are already working on AI data centre projects, to varying degrees across regions. 47 percent of survey respondents expected AI data centres to form more than half of the workload within the next 12 to 24 months. This trend is supported by recent analysis which forecasts AI data centres will comprise 28 percent of the overall data centre market by 2027.
While traditional fan cooling remains the main model in use according to our survey, liquid cooling is emerging as the dominant cooling solution for data centres with AI workloads. The majority (53 percent) of respondents to our survey expect that in the coming years the majority of their high-density projects will include liquid cooling.
SUSTAINABILITY
Water challenges
This shift brings with it another potential reputational challenge for the sector to manage when it comes to water use. Many liquid-cooling systems are ‘closed loop’ – requiring only an initial injection of water for recirculation.
However, there are still many designs and sites that do use larger amounts of water, and clients should assess markets where water shortages exist or may exist when deciding on their designs. As well as the lower environmental impact, more efficient, closed-loop systems may make applications for new schemes more attractive to regional planning authorities – often enabling faster approval of development pipelines.
DEMAND
Supply chain readiness
New technological advances in turn bring pressure on supply chains to meet the demands of industry. Our survey indicates signs of improving supply chain stability for data centres. 80 percent of respondents said delays for critical components such as switchgears and generators were occasional, rare or non-existent over the past 12 months. However, the outlook for 2026 is less positive, mostly due to the ongoing geopolitical and economic instability globally. Only 19 percent of respondents express confidence in the supply chain’s ability to meet manufacturing delivery dates in the next 12 months. When looking at cooling expertise specifically, our respondents raised concerns. Only 17 percent reported a view that the industry was well prepared or fully prepared to meet demand, while 47 percent believed bottlenecks were likely, and a further 31 percent raised questions about available expertise. For clients to successfully adopt these new technologies, they need to plan proactively to secure their supply chains – looking comprehensively at procurement to make sure they have the right relationships in place to make the shift towards higher-powered and more complex AI assets.
Figure 9:
How prepared is your local supply chain to support the increased adoption of the above cooling technologies over the next 12-24 months?
MOMENTUM
Hunger to succeed
Ultimately, our index and survey show an industry that is eager and ready for the potential challenges ahead. Look at how far the sector has come over the past decade. The power density of data centre server racks has increased by one hundred times – and we are still only at the start of the true AI revolution.
Meeting the requirements of critical infrastructure to support this change will need unparalleled focus and innovation from every discipline – solving power, engineering and cooling challenges, including those we haven’t even thought of yet. An open and creative mind to explore new and different delivery approaches will be the crucial point of differentiation in the years to come – essential to unlocking the vast benefits of the AI world at our fingertips.
Contents
Follow us
