AI INFRASTRUCTURE
The bigger picture for AI infrastructure What makes this story particularly inspiring for the AI industry is how it demonstrates that the environmental challenges of AI might not be insurmountable obstacles but design problems waiting for clever solutions.
The computational demands of AI aren’ t going to decrease, but projects like this show how infrastructure can evolve to handle these demands more intelligently.
The integration of computing hardware with building systems also points toward a future where data centres become community assets rather than just necessary evils that consume resources and generate noise.
Jonathan says:“ 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.”
The financial case for heat recovery also becomes increasingly compelling as energy costs rise and carbon pricing mechanisms make waste heat increasingly expensive to simply throw away.
For universities and other organisations with substantial heating demands located near computing infrastructure, the QMUL model offers a template that addresses both operational costs and environmental responsibilities.
“ This approach also created a platform to support its sustainability objectives via heat reuse, while enabling the University to act proactively and preventatively to intercept and remediate potential future issues,” John concludes.
“THE NEW DATA CENTRE IS MORE RELIABLE AND EFFICIENT THAN EVER”
Jonathan Hays, Professor, QMUL
142 September 2025