Global Futures: Cheriton School of Computer Science researchers develop a process to reduce the amount of energy needed to run data centres

Friday, January 23, 2026

Considering the ubiquity of smart devices and the increasing data demands of artificial intelligence, the amount of energy required to operate data centres continues to grow dramatically. Depending on the scale of the data centre, its power supply requirements can range from five to 100 megawatts.

Cheriton School of Computer Science Professor Martin Karsten and his former graduate student Peter Cai (MMath ’23) uncovered a subtle but powerful way to make data centres more efficient. Their research could reduce energy consumption by as much as 30 per cent by rethinking how network traffic is processed.

“It’s not just about being green,” Cai says. “Companies save money by using less energy. We ask: how can we do more with the hardware we already have?”