Short #88 - Science vs. What Works

BACK

Short #88 – Science vs. What Works

March 10, 2020

In this episode, Bryan talks about how to balance practice and experience with the “why” behind what we do and how either one can cause an error if you aren’t careful

Check out our most recent video

How to Evacuate an AC system, Full Vacuum Procedure

Our latest Podcast

HVAC School – For Techs, By Techs

Real training for HVAC ( Heating, Ventilation, Air Conditioning and Refrigeration) Technicians. Including recorded tech training, interviews, diagnostics and general conversations about the trade.

In this episode of the HVAC School Podcast, host Bryan sits down with Jeff Staub, Director of OEM Sales for Danfoss North America, to explore one of the most rapidly evolving frontiers in the HVAC and refrigeration world: thermal management for AI data centers. With nearly 30 years of industry experience spanning technical support, application engineering, and product development, Jeff brings deep expertise on how the explosive growth of AI chip technology is reshaping data center cooling architecture — and creating major new opportunities for HVAC professionals, contractors, and facility managers alike.

A central theme of the conversation is heat recovery — specifically, how the enormous amounts of heat generated by high-density GPU chips in modern data centers can be captured and repurposed rather than simply rejected into the atmosphere. Jeff explains that while heat recovery itself is not a new concept (supermarkets have used reheat coils and heat reclaim for decades), its application in AI data centers presents fresh challenges and possibilities. The heat coming off liquid-cooled server chips typically runs around 90 to 100 degrees Fahrenheit — useful, but not immediately at the temperature needed for most end applications like domestic hot water or space heating. Boosting that heat using heat pumps or feeding it into district energy systems, boiler pre-heat loops, vertical farms, or multifamily housing developments are among the most promising strategies being explored around the world.

Jeff highlights a significant contrast between Europe and the United States in how heat recovery is being adopted. In Europe, where district energy networks are widespread, data centers can plug directly into community heating infrastructure — and projections suggest that 80% of European data centers will incorporate heat recovery in the near future. In the US, the picture is more fragmented: while opportunities exist at universities, hospitals, urban mixed-use developments, and facilities co-located with nuclear power plants, the economics are trickier. Key sticking points include who owns the capital expenditure for heat recovery modules and heat pumps, and who ultimately benefits from the recovered heat. Bryan and Jeff discuss how innovative ownership models — with landlords, municipalities, or co-tenants sharing infrastructure — are beginning to unlock these opportunities, and how co-generation arrangements with power stations present exciting long-term potential.

The episode wraps up with highly practical guidance for HVAC contractors and facility managers looking to break into the data center space. Jeff encourages technicians not to be intimidated: the fundamentals of vapor compression, chiller systems, and fluid flow that HVAC professionals already know transfer directly to data center work. The key additions are familiarity with large centrifugal and screw compressors, variable frequency drives on pumps, glycol loop management, and central distribution unit (CDU) architectures. Bryan emphasizes that the boundary between HVAC and plumbing will continue to blur as secondary fluid pumping becomes more prevalent — and that staying curious and investing in ongoing training (through manufacturer programs like Danfoss Learning, Carrier University, and others) is the best way to ride this wave rather than get left behind. Both hosts agree: AI data centers are not going away, and the technicians who keep them cool will be indispensable.

Topics Covered

The evolution of data center cooling — from direct vapor compression on chips, to air-conditioned server rooms (CRAC units), to today’s liquid cooling and chiller-loop architectures
Why AI GPU chips generate unprecedented heat densities, with individual server racks approaching 250 kW to 1 MW of heat output
What heat recovery means in the data center context: capturing hot water (90–100°F) off chip cooling loops instead of rejecting it to outdoor air
The concept of ‘heat quality’ — why low-temperature waste heat is abundant but difficult to use directly, and how heat pumps solve the temperature-lift challenge
Real-world heat recovery applications: district energy systems, boiler pre-heat, vertical farms, multifamily housing, hospitals, and universities
Europe vs. the US: why district energy adoption makes heat recovery far more common in European data centers, and what the US can learn
Business model challenges: who pays for heat recovery infrastructure, and how co-location, municipal incentives, and landlord ownership models can unlock value
Co-generation opportunities: feeding recovered heat back into steam turbines at co-located nuclear or power plants
How heat recovery makes heat pump technology more viable by raising the source temperature and reducing compression ratio
Danfoss’s role in data center thermal management — from compressors and drives to plate heat exchangers, CDU flow control, and prepackaged heat recovery modules
Refrigerant transitions and what they mean for data center cooling (R-410A to R-454B, CO2 transcritical systems, potential two-phase refrigerant direct-to-chip cooling)
The convergence of HVAC and plumbing trades in a world of secondary fluid pumping and isolated refrigerant charges
Absorption chiller technology as a potential future use case for low-grade waste heat
Advice for contractors: how existing chiller and refrigeration skills translate to data center work, and what new competencies to build
Career and training resources: Danfoss Learning, manufacturer universities (Carrier, Trane, McQuay), and leveraging AI tools for self-education
The importance of redundancy and uptime in mission-critical data center environments — and what that means for service response expectations

Learn more about Danfoss at danfoss.com/learning

Have a question that you want us to answer on the podcast? Submit your questions at https://www.speakpipe.com/hvacschool.

Purchase your tickets or learn more about the 7th Annual HVACR Training Symposium at https://hvacrschool.com/symposium.

Subscribe to our podcast on your iPhone or Android.

Subscribe to our YouTube channel.

Check out our handy calculators here or on the HVAC School Mobile App for Apple and Android.

Search Episodes

Heat Recovery from Data Center w/ Jeff Staub

03/19/2026

Bryan Orr

All About ESCO with Renee Tomlinson and MeasureQuick’s Latest Features with Jim Bergmann

03/17/2026