Skip to main content
Cloud

How AI is changing the design of data centers

High-density facilities, more liquid cooling, and locations near power sources are some of the qualities experts tell IT Brew that AI data centers will have.
article cover

Jonathan Nackstrand/Getty Images

3 min read

Top insights for IT pros

From cybersecurity and big data to cloud computing, IT Brew covers the latest trends shaping business tech in our 4x weekly newsletter, virtual events with industry experts, and digital guides.

AI-friendly innovations are coming soon to a data center near you.

From new wooden construction models to out-of-this-world locations, data centers worldwide are no stranger to innovation. However, while a number of these changes have been spurred by sustainability goals, some cloud computing experts say a new motivator will begin to factor into the design of these facilities: AI.

What’s ahead? IT Brew spoke with several cloud computing experts to discuss what types of changes are in store for data centers across the country as AI becomes more mainstream. Nathan Blom, co-CEO at Iceotope Technologies, told IT Brew that for many up-and-coming AI data centers, density will be a key differentiator from traditional data centers as new locations look to “maximize” their footprint.

“The more you can pack into each rack, and the more racks you can pack into each row, the more efficient you become,” Blom said.

Andrew A. Chien, William Eckhardt professor in computer science at the University of Chicago, told IT Brew that the need for denser data centers will also bring on changes to the physical build of the infrastructures.

“If you have a house…[and] you put really heavy stuff in one place in the house, you have to build stronger beams and floors and all of that kind of stuff there,” Chien said. “So, that’s all at the data center level.”

Chien added that the new data centers will likely be placed in locations where there is a large power source available.

Kent Draper, CCO at IREN, a bitcoin mining and AI infrastructure company, told us that other changes data centers will undergo to support AI is to rely heavily on liquid cooling to accommodate the hefty cooling demand associated with AI.

“There has been a very small amount of liquid-cooled data center capacity built historically…I think we’re going to see a lot more liquid cooling as we move forward here just as the GPUs continue to become more powerful and those cooling requirements become greater and greater,” Draper said.

Out with the old? As some companies set off to build AI data centers from scratch, IT Brew asked the cloud computing experts if they believe companies will look to knock down traditional data centers to rebuild new and improved facilities that support AI innovation. Several, including Blom, had some optimism about the longevity of traditional centers.

“There’s a general movement towards not wanting to spend a ton on the retrofit, on the brownfield they call it, and do more greenfield and white space design, but that’s just simply because the market is still immature. It’s still new,” Blom said. “Ask me again in three years, and we’ll probably have a different answer.”

Top insights for IT pros

From cybersecurity and big data to cloud computing, IT Brew covers the latest trends shaping business tech in our 4x weekly newsletter, virtual events with industry experts, and digital guides.