Description

Data centers and their resource demands on surrounding communities have become national political news as AI infrastructure grows. Meanwhile, the industry's focus is about to shift again, from training frontier AI models to AI inference, which calls for smaller, more distributed facilities and raises a different set of potential concerns.

Henry Cubillan is senior vice president of Americas Commercial for Kelvion, a manufacturer of cooling and thermal management technology. In this episode, he sheds light on emerging data center cooling approaches that demand less water, along with their trade-offs. He also discusses how the industry should respond to public controversy about AI data centers.

In today’s episode, we’ll cover…

• New buildouts vs retrofitting existing data center facilities
• Evaporative cooling, closed-loop liquid cooling and dry cooling
• Striking a balance between data center power and water usage
• The outlook for enterprises with growing AI inference workloads

And more!

References:

• Data center builders confront public AI backlash
• AI data centers can run hot to stay cool: report
• A shadow water problem: AI's power demand shifts consumption upstream

To learn more about AI and IT Operations, check out TechTarget Infrastructure.

To watch clips from our podcast, subscribe to our YouTube channel, @EyeOnTech.