As AI continues to reshape the data center industry and put strain on existing infrastructure, one topic that’s heating up more than most, is cooling for the AI revolution. With traditional systems under equipped to meet the surging power demand of AI workloads, new strategies for meeting energy requirements while addressing subsequent cooling challenges, are vital to the future of the industry.

Cooling isn’t simply about removing heat. It’s about enabling and supporting performance, contributing to uptime, and managing sustainability. In the AI era, cooling has emerged as one of the defining challenges, and opportunities, for the next generation of data centers.

In this episode of DCD>Talks, Jim Simonelli, CTO at Schneider Electric shares his insights on why data centers are approaching a crucial tipping point when it comes to cooling, and how the company is addressing this key challenge.

The end of air cooling?

“Air cooling can only go so far. We’re getting to the point now where chips are exceeding the capability of this system,” says Simonelli.

For decades, data centers have relied on air to dissipate heat from servers and computing equipment. But as AI and high-performance computing (HPC) rise in popularity, power and thermal loads are quickly exceeding the capacity of standard air cooling methods.

“When you get to 100kW per rack and you try to air cool it, you’re going to need to supply so much air at such a low temperature and high flow rate, it’s going to become very inefficient and ineffective,” he explains.

In practical terms, this means that traditional cooling systems would have to work overtime utilizing significant airflow and sub-optimal temperatures, resulting in increased energy usage and diminishing returns. As such, the industry has been forced to rethink operations and seek out new solutions to help keep up with demand.

The AI boom and cooling

It isn’t just rack densities that are changing. The way in which power must be deployed, clustered and scaled has introduced entirely new infrastructure and management requirements. Simonelli explains: “The power and thermal densities are getting very high and they’re clustered in zones.”

This shift is driving the need for liquid cooling solutions, with water for example, carrying a thermal conductivity around 23 times higher than that of air, meaning it can transfer heat away from components more efficiently. However, despite growing awareness of the need to adapt cooling systems for the AI boom, adoption across the industry is varied:

“There are customers that are on their second or third AI cluster, and then there are those that are just beginning to explore it,” says Simonelli. But with the growth of AI-driven workloads only expected to rise, the thermal requirements of data centers shows no sign of slowing down

Staying ahead of the curve

In a constantly shifting landscape, it’s vital that operators prioritize future-ready solutions by adapting and advancing alongside industry demands. As part of its mission to stay ahead of the curve, Schneider Electric is taking a proactive approach by expanding its partnerships and, crucially, acquiring valuable capabilities in liquid cooling.

“The acquisition of Motivair was a really important one for us. They’re very focused on liquid cooling solutions, cooling distribution units (CDUs), rear door heat exchangers, direct-to-chip, and more,” explains Simonelli.

An ideal addition to Schnieder’s extensive experience in modular and scalable power and cooling architecture, Motivair’s portfolio is helping the company to move quickly and seamlessly into the next-generation of cooling.

“We have a customer base today that is demanding solutions. They’re building clusters today and they need answers today,” continues Simonelli.

Hybrid cooling for a transitional world

Even while liquid cooling gains ground, most data centers cannot convert overnight. In fact, many operators are opting for hybrid cooling solutions that utilize both traditional air systems and new liquid cooling setups. Simonelli points out:

“There’s a whole lot of infrastructure out there that’s already built. So, how do we make sure it’s future-ready?”

In response to this question, Schnieder has designed modular, prefabricated infrastructure that supports fast yet phased upgrades. Operators can start with air-cooled configurations and gradually introduce liquid cooling as and when required – whether via rear-door heat exchangers, CDUs, or direct-to-chip loops.

“People want to know, ‘can I build a facility today that’s going to be ready for liquid cooling, even if I’m not deploying it right now?’”

As well as reducing risk and cost during the transition to new cooling solutions, this approach offers operators the chance to adapt at their own pace, on their own terms.

Supply chain challenges and the people problem

Scaling liquid cooling involves more than just finding the right technology – it brings key manufacturing, logistical and global readiness challenges into question. In response, Schneider is expanding its production capabilities across multiple regions to meet the growing demand, efficiently and effectively. Simonelli explains:

“We’re scaling up manufacturing capacity to be able to deliver CDUs, rear door heat exchangers, and integrated systems, not just in one or two factories, but in multiple locations globally.”

Supply chain alignment is another key piece of the puzzle to ensure that components are readily available in the right areas to be deployed rapidly. But fundamentally, having the right people on board to install and maintain these new systems is key to successful implementation and scaling.

“There’s definitely a skills gap,” says Simonelli. “Liquid cooling is a different animal than air cooling. It involves fluid dynamics, pressure, flow, sometimes even chemical interactions.”

Investing in the right talent at the right times is another talked-about industry challenge that must be addressed sooner rather than later. Schneider has launched new training initiatives and technical centers focused on upskilling technicians, engineers, and service partners, in a bid to feed the new generation of data center personnel.

The investment in people is as crucial as investment in technology. The two must go hand in hand in order to meet rising power demand and address the key challenges that come with it. Simonelli notes:

“If you don’t have the right people to install, maintain, and troubleshoot new cooling systems, it’s not going to be successful.”

With AI driving the evolution of data center operations and infrastructure, key industry players must step up to meet it. Schneider's innovative approach and continued investment in innovation is paving the way for operators to successfully scale up, and cool down.

To explore more insights from Simonelli on the future of data center cooling and the company's key strategies for staying ahead of the curve, watch the full DCD>Talks episode here.