From dorms to trout ponds: Leveraging waste heat from AI computing

This episode of The Future Built Smarter examines modern high performance computing facilities and the opportunity to use their waste heat as an energy source.

“High performance facilities for computing are different than what most people would consider a data center,” says Brandon Fortier, IMEG's Director of Science and Technology. “Data centers serve general purpose IT needs for a building or a company, things like web hosting, cloud storage capabilities, et cetera. High performance facilities for computing serve more of a scientific or other research need. They use significantly more computing power because they're doing things like climate modeling or machine learning for AI.”

Such energy-intensive AI deep learning computing requires an intense amount of cooling, which generates a significant amount of waste heat. In the right locations—such as on a healthcare, higher education, or federal campus, as well as private research campuses—this heat can be recovered and pushed into district energy systems for use in the heating of neighboring buildings, greatly supporting overall sustainability and greenhouse gas reduction goals in the process.

“These high performance data centers definitely produce a lot of heat,” says Ken Urbanek, a client executive in IMEG’s Denver office. “On the other hand, that's a great energy source. Let's grab that source energy from that data center and push that into the campus.”

Ken provides several real-life examples of leveraging such an energy source, from an East Coast university that is planning to condition millions of square feet across its campus to more niche examples such as conditioning greenhouses in colder climates and even warming a pond at a trout farm.

“There are a lot of synergies out there,” Ken says. For a campus, he adds, “as you're looking to decarbonize and you're looking at various options, part of that exercise should be if a data center makes sense… There's not a research facility, whether it's on a healthcare, academic, or other campus, that probably can't find a business case to say, ‘How can we utilize advanced AI to help further our research?’

Now you just have to say, ‘Where on this campus can we find 10, 15, 20 megawatts of power?’ In your decarbonization plan, you might be pulling that energy anyway. And better to put it into AI and utilize it for research than to put it into heat pumps or some sort of electric generation system.”