Powering data centres in the age of AI

The rise of artificial intelligence (AI) is reshaping the data centre landscape, introducing significant challenges related to power consumption and availability. As data centres are increasingly relied upon to support AI services and applications, the resilience of their power infrastructure is being put to the test.

In Australia, local data centre operators have spoken out about how the artificial intelligence boom has accelerated land and energy challenges. With the increase in energy prices and discussions around moving to renewable energy, the increasing use of AI will put pressure on energy grids in Australia immediately.

In this climate, ensuring an uninterrupted power supply becomes paramount, especially as the demand for data continues to grow.

Understanding edge data centres

Edge data centres (EDCs) are located on the edges of a larger network to help process data closer to users. This set-up speeds up response times, which is especially important in areas like health care, where quick access to information is essential. Despite their importance, many people still don’t fully understand what edge data centres are and how they can be used.

Types of EDCs

EDCs can be classified into three main categories: on-premise edge, network edge and regional edge. Each type serves distinct purposes, but collectively, they aim to enhance the speed and reliability of data processing. Interestingly, many organisations building large enterprise data centres for operational efficiency may not even recognise that they are, in effect, creating regional edge data centres due to their strategic locations.

Future considerations for EDCs

The ongoing evolution of edge data centres is marked by the integration of AI technologies, which demand substantial power and prompt responses from power systems. Data centres, traditionally operating at 60–80% of their designed capacity, are now facing immediate, full-capacity demands at any time of day. This shift towards fluctuating power loads — referred to as step loads — requires a re-evaluation of power distribution and backup systems.

A reliable uninterruptible power supply (UPS) system becomes critical to manage these step loads effectively. As AI applications rapidly increase, ensuring that the infrastructure can respond instantly to changes in power demand is essential for maintaining operational continuity.

The importance of selecting the right power infrastructure

Choosing the appropriate UPS is vital for safeguarding the critical infrastructure of data centres. Power system designers must consider how their equipment will cope with the increased demands brought about by AI. High-quality UPS systems with advanced control mechanisms are essential for managing fluctuating loads and minimising potential disruptions.

The industry is also witnessing advancements in battery technologies, aimed at supporting the growing demands of AI workloads. AI is poised to drive a 160% increase in data centre power demand globally¹, increasing the need for reliable, efficient power systems.

Conclusion

As the data centre industry navigates the complexities introduced by AI and edge computing, the future holds both challenges and opportunities. The growing importance of EDCs emphasises the necessity for robust power solutions capable of adapting to fluctuating demands.

The journey towards optimising power infrastructure for AI applications is just beginning. Data centre operators must remain vigilant, continuously assessing their power systems to ensure reliability and performance as they embrace this new technology. The evolving landscape of edge data centres is not merely a trend; it is a fundamental shift that will shape the future of digital infrastructure, both locally and globally.

_{1. https://www.goldmansachs.com/insights/articles/AI-poised-to-drive-160-increase-in-power-demand}

Image credit: iStock.com/Дмитрий Ларичев