citiesabc, first_page
University of Waterloo Develops New Cooling Technology That Reduces AI Chip Hotspot Temperatures by 17°C in 15 Seconds
10 Nov 2026
Researchers at the University of Waterloo have developed an experimental cooling method that uses tiny drops of boiling coolant to manage sudden temperature spikes in computer chips. The technology could help data centres reduce cooling costs and improve the reliability of AI infrastructure.
Waterloo, Canada, 10 November 2026 – Researchers at the University of Waterloo have developed a new cooling technique that could make artificial intelligence data centres more efficient and less expensive to operate by targeting overheated areas of computer chips with tiny drops of specialised coolant.
In laboratory experiments, the team demonstrated that a single drop of coolant could reduce the temperature of a simulated chip hotspot by up to 17°C in just 15 seconds, several times faster than conventional immersion cooling under the tested conditions.
The research introduces a hybrid approach that combines inexpensive deionised water for general cooling with small quantities of specialised coolant applied to areas experiencing sudden temperature increases.
The system is being developed by a team led by Dr. Sushanta Mitra, Professor of Mechanical and Mechatronics Engineering at the University of Waterloo and Executive Director of the Waterloo Institute for Nanotechnology.
The findings were published online in the Journal of Colloid and Interface Science, offering a potential new approach to managing the growing thermal demands of AI processors and high-performance computing infrastructure.
Why AI Data Centres Need More Efficient Cooling
The rapid expansion of artificial intelligence is increasing demand for powerful computer processors capable of handling complex workloads.
These processors generate substantial amounts of heat during operation, requiring data centres to invest in cooling infrastructure that maintains suitable operating temperatures.
When temperatures rise too quickly, computer chips may reduce their performance to prevent damage, a process commonly known as thermal throttling. Prolonged exposure to excessive heat can also affect component reliability.
Traditional air-cooling systems circulate chilled air around electronic equipment, while liquid-cooling technologies transfer heat using fluids.
Immersion cooling, in which electronic components are placed in specialised liquids, offers another way to manage heat generated by dense computing systems.
However, even liquid-based systems can struggle when small regions of a processor experience sudden temperature increases.
These localised hotspots can develop within seconds, making them difficult to address using cooling methods designed primarily for steady heat generation.
The Waterloo researchers focused on developing a system capable of responding directly to these temporary temperature spikes.
How the University of Waterloo's Hybrid Cooling System Works
The experimental method separates general cooling from the treatment of sudden hotspots.
Deionised water provides the primary cooling environment, while a specialised liquid with a lower boiling point is used in small quantities where additional cooling is required.
It's a division of labour,
said Dr. Sushanta Mitra.
Water is excellent at handling steady, everyday heat and the drops take care of sudden spikes.
The research team selected an electronics coolant that is denser than water and does not readily mix with it.
When the coolant drop encounters a sufficiently hot surface, it begins to boil, absorbing thermal energy as it changes state.
The resulting bubbles also disturb the surrounding water, improving heat movement away from the heated area.
This combination allows the liquid to respond rapidly to concentrated heat without requiring the entire cooling environment to use the more expensive specialised coolant.
In laboratory testing, the technique reduced the temperature at the centre of a simulated chip hotspot by approximately 17°C within 15 seconds.
The published research examined how the cooling response changed under different heat loads, water depths and droplet sizes.
Researchers Develop Coated Coolant Drops for More Precise Temperature Control
Beyond demonstrating the cooling effect, the researchers investigated how to control the behaviour of individual coolant droplets.
The team included postdoctoral researcher Dr. Tian-Yu Zhang and PhD student Arnav Banerjee, who worked with Mitra on the experimental system.
Building on earlier research into liquid encapsulation, they developed a method of surrounding coolant drops with a thin layer of silicone oil.
Tiny heat-conducting ceramic particles were added to the coating to help control how quickly the droplets begin boiling.
This approach could allow cooling systems to regulate when and where heat absorption occurs, making the technology more adaptable to different operating conditions.
The researchers envision two possible methods for future implementation.
One would involve circulating coated coolant droplets through water, allowing them to activate when they encounter overheated areas.
The second would use sensors to detect hotspots and deliver droplets directly to the affected locations.
Both approaches remain concepts for future development rather than systems currently operating in commercial data centres.
Could the Technology Reduce AI Data Centre Operating Costs?
Cooling represents an important operating expense for data centres, particularly facilities using large numbers of high-performance processors.
The University of Waterloo's research suggests that combining water-based immersion cooling with targeted coolant droplets could reduce reliance on specialised liquids.
Instead of filling an entire cooling system with expensive coolant, operators could potentially use smaller quantities only where rapid heat removal is required.
This could lower material costs while helping maintain processor performance during sudden changes in workload.
The technology may also contribute to improved equipment reliability by reducing the duration of localised overheating.
However, the researchers have not yet demonstrated a specific percentage reduction in commercial data centre operating costs or electricity consumption.
Those outcomes will require further testing in realistic computing environments, including assessments of equipment compatibility, droplet delivery, maintenance requirements and long-term performance.
The laboratory result therefore demonstrates the effectiveness of the cooling mechanism under controlled conditions rather than a confirmed financial saving for data centre operators.
Potential Applications Beyond AI Computing
Although the research focuses on heat management for advanced electronics, the same approach could have applications in other industries.
Electric vehicle batteries, power electronics, high-powered lasers and LED lighting systems can also experience localised temperature spikes.
In these environments, rapid cooling may help protect sensitive components or maintain performance during demanding operating conditions.
The researchers believe the ability to control boiling through encapsulated liquid droplets could create opportunities for designing more adaptable thermal management systems.
Further work will be required to determine whether the approach can be scaled for different devices and operating environments.
What the Research Means for Future AI Infrastructure
As investment in AI computing continues, data centre operators are examining cooling technologies capable of supporting processors that generate large amounts of heat.
The Waterloo study introduces a method for addressing sudden temperature spikes using a relatively small quantity of specialised fluid.
Its principal advantage is the separation of routine cooling from rapid hotspot suppression, allowing each task to be handled by a different liquid.
While commercial deployment remains some distance away, the findings provide researchers and equipment manufacturers with another approach to consider when designing future cooling systems.
For cities hosting expanding data centre infrastructure, improvements in cooling efficiency could eventually help operators manage energy demand and operating expenses.
Those wider benefits will depend on successful engineering development and testing beyond the laboratory.
The University of Waterloo's findings demonstrate that targeted cooling can respond rapidly to simulated chip hotspots, providing a foundation for further research into more efficient thermal management for AI infrastructure.
About the University of Waterloo
The University of Waterloo is a Canadian research university based in Waterloo, Ontario, recognised for its engineering, computing and applied science programmes.
The institution has more than 41,000 students and conducts research across areas including artificial intelligence, nanotechnology, advanced manufacturing and sustainable technologies.
The cooling research was conducted by Dr. Sushanta Mitra, Dr. Tian-Yu Zhang and Arnav Banerjee.
Their study, A Hybrid Immersion Cooling Assisted by Liquid-Liquid Encapsulation: A Game Changer for Thermal Management of Transient Hotspots, was published online in the Journal of Colloid and Interface Science.
Official website: https://uwaterloo.ca/
Share
Sara Srifi
Sara is a Software Engineering and Business student with a passion for astronomy, cultural studies, and human-centered storytelling. She explores the quiet intersections between science, identity, and imagination, reflecting on how space, art, and society shape the way we understand ourselves and the world around us. Her writing draws on curiosity and lived experience to bridge disciplines and spark dialogue across cultures.
