Cooling the future of AI with a high-density vision

Share this article

The relentless rise of artificial intelligence (AI) and accelerated computing has forced the data centre industry to confront an unavoidable truth: traditional cooling methods are reaching their limits. The exponential increase in computing power, driven by GPUs and high-performance architectures, is creating heat densities that air cooling alone cannot manage efficiently. Schneider Electric’s recent acquisition of a majority stake in Motivair is a strategic response to this challenge, positioning both companies at the forefront of next-generation cooling solutions.

High density applications

Schneider Electric had been monitoring Motivair for over a year before the company entered the market for acquisition. Several factors made them an attractive prospect, beginning with their long-standing expertise in high-density applications. “They were well known in high-performance computing and supercomputers,” Andrew Bradner, Senior Vice President Cooling Business at Schneider Electric, says. “Their broad and deep portfolio complemented gaps in our existing offerings, and as we progressed, the cultural alignment between our companies became clear.”

Sustainability commitments played a crucial role in the deal. Motivair shared Schneider’s focus on deploying high-density applications responsibly, particularly as generative AI and large language models increased demand. Geographically, Motivair’s US presence filled a strategic gap for Schneider, whose cooling business had been largely international. “Their manufacturing and competency base in Buffalo, New York, was another advantage,” Bradner adds. “They needed a partner with industrial and commercial scale to expand globally, and our footprint in over 130 countries made that an ideal fit.”

The acquisition also received strong validation from customers. “One particularly influential customer described it as a ‘marriage made in heaven,’” Bradner says. “The combination of Motivair’s domain knowledge with Schneider’s industrial scale gave customers confidence in our ability to expand Motivair’s solutions worldwide.”

Managing the Motivair brand within Schneider

Schneider Electric acquired a 75 per cent stake in Motivair, integrating it financially while retaining its existing leadership. “Rich Whitmore will continue as President and CEO, and we are keeping Motivair as an associated brand,” Bradner continues. “We plan to complete the full acquisition by 2028, but maintaining Motivair’s expertise and identity is critical.”

A similar approach was taken with APC, which Schneider acquired 17 years ago and continues to operate under the ‘APC by Schneider Electric’ brand. “Motivair has strong recognition, especially in high-performance computing, and deep partnerships with silicon leaders like Nvidia and AMD,” Bradner says. “These relationships, combined with their reputation, provide significant value to customers.”

AI’s impact on cooling requirements

The shift toward AI and accelerated computing has fundamentally changed data centre cooling. “The main driver is densification,” Bradner explains. “GPU-driven computing has significantly increased rack densities. While air cooling still plays a role, at these higher densities, liquid cooling is the only responsible way to deploy compute at scale.

“Industry estimates suggest around 90 gigawatts of new data centre capacity will be deployed over the next five years, primarily for AI applications. If this were all air-cooled, energy demand could rise to 120 gigawatts or more, which would be inefficient and environmentally irresponsible. Liquid cooling is far more sustainable.”

Liquid cooling is not new. It was common in mainframes 30 or 40 years ago and has been central to Motivair’s work with Cray supercomputers, which required 100 per cent liquid cooling. “More than a decade ago, Motivair built 400-kilowatt racks,” Bradner says. “Today, cloud computing designs are at 20 to 30 kilowatts, but Nvidia’s NVL 72 is already at 130 kilowatts per rack. Motivair has long been operating at these densities, making them well-positioned for AI-driven infrastructure.”

The acquisition allows Schneider and Motivair to consolidate engineering efforts and accelerate innovation. “We now have the broadest high-density cooling portfolio and can focus on next-generation solutions rather than duplicating existing efforts. Innovation is central to Schneider’s DNA,” Bradner continues.

The role of air cooling in a liquid-cooled future

Despite the growing dominance of liquid cooling, air cooling will not disappear. “Direct-to-chip liquid cooling still requires air cooling for 10 to 30 per cent of the heat load, depending on the server design,” Bradner says. “Many cloud data centres still use 25 to 30-kilowatt racks. If you have an NVL 72 at 130 kilowatts per rack and 20 per cent of that heat needs to be rejected by air, that’s still 26 kilowatts that must be managed through air cooling.”

Hybrid systems are likely to be the norm. “Liquid cooling actually complements air cooling rather than replacing it entirely,” Bradner explains. “The key is an end-to-end solution, from power input to heat rejection, with efficient chillers, indoor air systems, and liquid cooling expertise. Customers increasingly seek partners who can provide a holistic approach.”

Motivair’s portfolio is already certified for Nvidia’s latest hardware, and its experience with exascale supercomputing means it is well-equipped for further increases in rack density. “For now, the portfolio is fit for purpose, but once densities exceed 200 kilowatts per rack, power distribution will become far more complex,” says Bradner. “A full-system approach, integrating power and cooling, will be critical.”

New cooling methods such as two-phase and immersion cooling are gaining interest, but each presents challenges. According to Bradner immersion cooling can struggle with high densities, leading some providers to add cold plates in addition to submersion. It also presents operational difficulties, particularly for IT maintenance.

Regulatory and environmental concerns are another factor. Depending on the liquid used – whether mineral oils, two-phase refrigerants, or dielectrics – there can be issues around PFAS substances and other chemicals. “Our approach remains technology-agnostic, ensuring we stay ahead of all cooling developments,” Bradner explains.

Nanofluids are emerging as another efficiency driver, improving both internal system performance and outdoor heat rejection. Software automation and autonomous cooling control are also critical areas of innovation. Integrating intelligent cooling controls with building management systems offers significant efficiency gains, and optimising these end-to-end systems through software will be a key focus in the coming years.

Regional differences in cooling adoption

Cooling strategies vary by region, shaped by regulatory and economic factors. “Europe has historically been more focused on sustainability, driven by regulations such as F-gas restrictions,” says Bradner. “The US has recently introduced similar refrigerant regulations, though policy direction may shift with changes in administration.”

In regions with abundant, cheap power, efficiency is sometimes deprioritised in favour of speed and scale. The US has traditionally relied more on air handler economisation, while Europe has favoured chilled water systems. This has made the transition to liquid cooling easier in some European markets where facility water loops were already in place. In the US, the retrofit challenge is greater, leading to rising demand for chillers to support liquid cooling.

Despite regional differences, the overarching trend is clear: cooling must become more efficient and sustainable as AI-driven infrastructure scales. “The demand for high-density compute is accelerating, and cooling technology must evolve in parallel,” Bradner concludes. “Our partnership with Motivair ensures we remain at the forefront of this transformation, delivering solutions that balance performance, efficiency, and sustainability.”

As AI reshapes the digital landscape, the industry’s ability to manage heat will define the pace of innovation. For Schneider Electric and Motivair, the challenge is clear, but so is the opportunity.

Related Posts
Others have also viewed

Do not skip the hard part of AI deployment

Data quality, real-time context, and production infrastructure will define the future of enterprise AI success. ...

Platform disruption is coming faster than your next upgrade

AI is redefining how enterprises design, deploy, and interact with software, leaving traditional digital experience ...

Growth of AI will push data centre power demand thirtyfold by 2035

Artificial intelligence is poised to become the dominant force shaping the future of global infrastructure, ...

AI needs quantum more than you think

As AI models grow larger and classical infrastructure reaches its limits, the spotlight is turning ...