AI INFRASTRUCTURE
Machine learning algorithms – powered by the same AI infrastructure they’ re managing – predict when the servers will be working hardest and coordinate this with when the campus buildings will need the most heat.
From computational burden to campus asset Traditional data centres spend enormous amounts of money removing heat that serves no purpose beyond keeping expensive equipment from melting.
Yet QMUL has turned this cost centre into something approaching a profit centre by making waste heat into a campus utility.
The heat recovery system works through underground pipes that distribute thermal energy captured from server cooling systems to buildings across the campus. It’ s essentially district heating powered by AI research.
Professor Jonathan Hays from QMUL says:“ The support we’ ve had from APT and Schneider Electric has been unparalleled. Both companies came together to help us develop an exciting and innovative project which would enable us to provision for the future.
“ The biggest impact is that we were able to deliver on what we promised while improving our sustainability,” he says.
“ The new data centre is more reliable and efficient than ever and through the heat recovery, we have significantly reduced our spending on heating and hot water, while gaining enhanced reputational benefits from taking a lead on sustainability within our data centre operations.”
138 September 2025