Industry News

The relentless march of artificial intelligence (AI) is pushing data centre cooling systems to their absolute limits.

Inside these massive computing facilities, densely packed servers generate enough heat to require industrial-scale cooling solutions, with some areas reaching critical temperatures exceeding 100°F (37.8°C). As AI workloads continue to multiply exponentially, traditional cooling methods are struggling to keep pace with the escalating thermal demands.

The challenge is set to become even more daunting. Industry analysts at Goldman Sachs project an extraordinary 160% surge in data centre power requirements by 2030. This impending energy crisis has sent engineers and researchers scrambling to develop more efficient cooling solutions before the current infrastructure reaches its breaking point.

Enter a groundbreaking innovation from the University of Texas at Austin. Their research team has engineered an advanced data centre cooling solution that could revolutionise how we manage heat in these digital powerhouses. This novel thermal interface material doesn’t just marginally improve upon existing solutions – it shatters previous performance benchmarks, delivering up to 72% better cooling efficiency than current commercial technologies.

The secret lies in an ingenious combination of liquid metal Galinstan and ceramic aluminium nitride, brought together through a sophisticated mechanochemical process. This innovative approach to data centre cooling could reduce overall facility energy consumption by 5%, representing a significant breakthrough in operational efficiency and environmental sustainability.

“The power consumption of cooling infrastructure for energy-intensive data centres and other large electronic systems is skyrocketing,” explained Guihua Yu, professor in the Cockrell School of Engineering’s Walker Department of Mechanical Engineering and Texas Materials Institute. 

“That trend isn’t dissipating anytime soon, so it’s critical to develop new ways, like the material we’ve created, for efficient and sustainable cooling of devices operating at kilowatt levels and even higher power.”

The timing of this breakthrough couldn’t be more critical. Goldman Sachs also estimated that AI applications alone are expected to drive an additional 200 terawatt-hours per year in data centre power consumption between 2023 and 2030. With cooling currently accounting for approximately 40% of data centre energy usage – equivalent to eight terawatt-hours annually – the need for more efficient cooling solutions has never been more pressing.

The new thermal interface material’s performance is particularly impressive. It can remove 2,760 watts of heat from a mere 16 square centimetres of area. This exceptional capability could reduce cooling pump energy requirements by 65%, addressing a significant component of the overall electronics cooling challenge. 

When implemented across the industry, this innovation could reduce total data centre energy usage by 5% – substantially improving both environmental impact and operational costs.

Lead author Kai Wu emphasises the broader implications of this development: “This breakthrough brings us closer to achieving the ideal performance predicted by theory, enabling more sustainable cooling solutions for high-power electronics. Our material can enable sustainable cooling in energy-intensive applications, from data centres to aerospace, paving the way for more efficient and eco-friendly technologies.”

The research team achieved this breakthrough by utilising a specialised mechanochemistry process that enables the liquid metal and aluminium nitride to mix in a highly controlled manner. This precise engineering creates gradient interfaces that significantly enhance heat transfer efficiency, bridging the long-standing gap between theoretical cooling potential and real-world performance.

While the current tests have been conducted on small lab-scale devices, the research team is actively working on scaling up material synthesis and preparing samples for testing with data centre partners. This next phase will be crucial in validating the technology’s effectiveness in real-world applications and its potential to address the growing cooling demands of AI and high-performance computing infrastructure.

The implications of this thermal interface material extend beyond just cooling efficiency. As data centres expand their AI capabilities and processing power, this innovation could enable the development of more compact, energy-efficient facilities. This could lead to significant cost savings while supporting the sustainable growth of digital infrastructure necessary for advancing AI technologies and other computational innovations.