📊 Key Data
  • Power Density Surge: AI server racks now draw up to 120 kW, far exceeding traditional 5-10 kW limits.
  • Market Growth: Global data center liquid cooling market projected to hit $30B by 2034, up from $4B last year.
  • Thermal Efficiency: Immersion cooling achieves Power Usage Effectiveness (PUE) near 1.05, the most efficient method demonstrated.
🎯 Expert Consensus

Experts agree that liquid cooling is now essential for AI data centers, requiring holistic infrastructure planning to manage thermal, structural, and spatial challenges.

about 7 hours ago
Beyond Air: How Panduit's New AI Cooling Lab is Reshaping Data Centers

Beyond Air: How Panduit's New AI Cooling Lab is Reshaping Data Centers

TINLEY PARK, IL – November 17, 2026 — The generative artificial intelligence boom is frequently discussed as a software revolution, but inside the sprawling facilities that power the modern internet, it has triggered a thermodynamic crisis. As advanced AI accelerators from chipmakers like NVIDIA and AMD push server rack power requirements to unprecedented extremes, traditional air cooling systems are hitting hard physical limits.

Recognizing this critical infrastructure bottleneck, Panduit, a global manufacturer of electrical and network infrastructure solutions, announced yesterday the expansion and rebranding of its Tinley Park thermal testing facility into the Panduit AI Cooling Lab.

The multi-technology demonstration center is a direct response to a stark new reality in enterprise computing: liquid cooling is no longer an exotic luxury for supercomputers. It is a mandatory requirement for the future of business. By bringing four distinct advanced cooling architectures into a single comparative environment, the Illinois-based company is highlighting a crucial, often-overlooked narrative in the AI gold rush. Selecting a cooling technology is no longer a localized IT decision; it is a foundational choice that dictates the structural, electrical, and spatial design of the entire data center.

A Menu of Thermal Solutions for the AI Era

The physics of the current computing landscape are unforgiving. A standard enterprise server rack historically drew between 5 and 10 kilowatts (kW) of power. Today, racks packed with next-generation AI accelerators—such as NVIDIA's GB200 NVL72 or clusters of AMD Instinct MI350s—are routinely drawing 50 kW, with some specialized deployments exceeding 120 kW. Air simply cannot absorb and transport heat fast enough to keep these densely packed components from melting down. Liquid, however, is approximately 23 times more thermally conductive than air.

Panduit's new lab provides a physical space to evaluate the four primary liquid-based methodologies battling for dominance in this transition. The first, rear-door heat exchangers (RDHx), acts as a bridge technology. By circulating chilled water through a radiator-like door attached to the back of a server rack, facilities can cool exhaust air before it re-enters the room. It is a highly effective retrofit strategy for racks pushing 20 to 30 kW, neutralizing the heat load without requiring liquid to enter the server chassis itself.

For denser workloads, the lab showcases single-phase direct-to-chip (D2C) cooling. This approach circulates a liquid coolant through cold plates attached directly to CPUs and GPUs, capturing heat at the source. It is rapidly becoming the workhorse of the AI era, capable of handling 40 kW and above.

Pushing the boundaries further, the facility also demonstrates two-phase direct-to-chip cooling, which utilizes a specially engineered dielectric fluid that boils and condenses within the cold plate. By leveraging the latent heat of vaporization, this method targets the extreme heat fluxes of next-generation silicon, where power densities can exceed 200 watts per square centimeter.

Finally, the lab features immersion cooling, where entire servers are submerged in non-conductive dielectric fluid. While requiring a radical redesign of traditional rack layouts, immersion offers unparalleled thermal efficiency, often achieving a Power Usage Effectiveness (PUE) near 1.05, making it the undisputed champion of high-density energy conservation.

The Hidden Domino Effect of Liquid Cooling

While the thermal performance of these systems is impressive, the true strategic value of Panduit's new lab lies in its focus on infrastructure interdependencies. Introducing liquid into a data center triggers a massive domino effect across the facility's physical footprint.

"Cooling is not an isolated decision. As computing densities increase, organizations need to understand how a cooling approach interacts with the infrastructure around it," explained Tom Kelly, Panduit's Chief Technology Officer. "The AI Cooling Lab gives customers a hands-on opportunity to explore those relationships, compare approaches, and have informed conversations before making important deployment decisions."

Industry engineers note that the sheer weight of liquid cooling systems, particularly immersion tanks, can necessitate structural reinforcements to data center floors. Furthermore, the introduction of coolant manifolds and hoses creates severe spatial conflicts. These fluid pathways must be meticulously routed alongside high-voltage power distribution units (PDUs) and delicate, high-bandwidth fiber-optic cabling.

Leak prevention is another paramount concern. Manufacturers are increasingly adopting aerospace-grade design principles and 100 percent leak testing for their manifolds. Panduit’s approach frames thermal management not just as a plumbing challenge, but as an integrated systems engineering problem. If a cooling manifold blocks access to a critical network switch, or if a minor fluid drip threatens a primary power line, the operational risk to the enterprise skyrockets.

Strategic Positioning in a Surging Market

The financial stakes underlying these infrastructure decisions are staggering. The global data center liquid cooling market, valued at roughly $4 billion just last year, is projected by industry analysts to approach $30 billion by 2034, expanding at a compound annual growth rate exceeding 25 percent.

This explosive growth has triggered an arms race among infrastructure providers. Competitors like Vertiv and Schneider Electric have aggressively expanded their liquid cooling portfolios, while specialized players like nVent are forging deep partnerships with chipmakers like AMD to integrate cooling directly into rack-scale AI platforms. CoolIT Systems, recently acquired by Ecolab, is leveraging chemical expertise to dominate the coolant and monitoring space.

Panduit’s strategy with the AI Cooling Lab is to differentiate itself by refusing to be boxed in as a pure cooling component vendor. By offering "Discovery Sessions" where clients can map out their specific legacy constraints against multiple cooling topologies simultaneously, the company is positioning itself as a holistic infrastructure consultant. This is particularly vital as the industry shifts toward hybrid environments, where legacy air-cooled servers for standard enterprise applications must coexist safely alongside liquid-cooled AI clusters on the same data center floor.

Simulating the Inevitable at SC26

The launch of the facility perfectly times with the broader industry's convergence at the SC26 supercomputing conference, where Panduit is currently exhibiting its lab's capabilities. The trade show floor has transformed from a showcase of pure processing power into a frantic marketplace for thermal management solutions, reflecting the new reality that computation is now fundamentally bottlenecked by heat rejection.

For enterprise CIOs, colocation providers, and facility architects, the transition from air to liquid is fraught with capital risk. A miscalculated deployment can lead to stranded capacity, catastrophic leaks, or an inability to scale as AI models grow larger and more power-hungry. By providing a sandbox to test the physical realities of rack weight, cable routing, and power distribution alongside thermal efficiency, initiatives like the AI Cooling Lab are becoming critical de-risking tools.

As the physical footprint of the internet undergoes its most radical transformation in two decades, the underlying message from Tinley Park is unambiguous: in the era of artificial intelligence, the companies that master the physical realities of infrastructure integration will be the ones that ultimately control the cloud.

Topics & Related

Event:
Expansion
Rebranding
Theme:
Data Centers
Generative AI
Metric:
CAGR
Sector:
Cloud & Infrastructure

📝 This article is still being updated

Are you a relevant expert who could contribute your opinion or insights to this article? We'd love to hear from you. We will give you full credit for your contribution.

Contribute Your Expertise →
UAID: 55149