by  

Dino Foderaro

Last month, a flurry of rumors began to circulate in the world of Formula 1. (Hang on; we promise this’ll be relevant.) The FIA has just concluded investigations of McLaren Racing Limited, the second-most successful Formula 1 team of all time, which was recently accused of cheating by Red Bull. McLaren’s cars have seemingly gained “a massive performance advantage over their rivals,” which some assume is due to the company using water to cool their tires—a practice that’s forbidden in Formula 1.

Turns out? It isn’t water—it’s something even better: the art of phase change. Allegedly, McLaren has relied on phase change materials (or PCMs) to maintain perfect tire temperatures while its cars achieve staggering speeds. Martin Buchan, who wrote his doctoral thesis at McLaren, explains the brilliant simplicity of phase change, and why any material that undergoes phase change has become known in F1 as a “magic temperature sponge”:

If materials don’t change their phase, they just keep getting hotter and hotter. But when reaching the melting point of one material, they need to absorb a lot of heat for the melting process, which means they keep the same melting temperature for longer and absorb a lot of heat.

This ability to capture greater levels of heat via phase change is what has most likely enabled McLaren to reach new heights of dominance on the track. “This is a completely new quality of superiority,” explained Auto Motor Sport. “They demonstrate their superiority on every [tire] type,” commented Toto Wolff, a former driver and current CEO of Mercedes’ F1 team. “We have a similarly fast car to them, but only on one lap. McLaren can reproduce that [speed] every lap.”

Two-phase liquid cooling, in which a coolant undergoes phase change by boiling from a liquid to a gas, can unlock these same benefits for your data center. Not only can it deliver superior heat capture when it’s needed most, but offers the same consistency of stellar performance throughout your data center’s lifecycle.

While water-based single-phase cooling may be the most familiar liquid cooling solution on the market today, that familiarity comes with unseen costs: issues such as higher maintenance requirements, or profound risks of biofouling or corrosion. And it won’t be long before single-phase hits its cooling limits, while next-gen AI workloads continue to draw more and more heat.

Simply put: the commoditization of single-phase cooling has become a race to the bottom. With adoption of single-phase, your data center risks being stuck in the slow lane.

Two-phase, direct-to-chip cooling, however? Its performance will provide an onramp for your data center to achieve what’s needed for the next era of AI. It’s the McLaren approach: smarter, faster, and engineered for victory. In other words: it’s a race towards excellence.

Advanced Cooling: It Isn’t A Grey Area

Single-phase’s sprint towards commoditization is largely due to an industry focus on the grey space (i.e., the back-end infrastructure designed to support a data center’s servers and storage). In the grey space, standardization is king: “tried,” “competent,” and “cheap” are terms commonly used to praise preferred back-end infrastructure. And while this short-term approach may nominally drive down costs, it also drives down innovation.

In the white space, however, where AI inference, next-gen compute, and other revenue-generating IT equipment resides, standardization is a trap. Here, differentiation is the point. Critical infrastructure in the white space should emphasize maximizing uptime, protecting multi-million-dollar AI investments from potential (and entirely avoidable) catastrophe, and optimizing thermal performance for the long haul. This emphasis towards excellence consequently requires adopting solutions that are robust, resilient, and able to be modified to activate your data center’s ideal architecture—not just manufactured at scale.

NeuCool™: Engineered for Excellence

As such, while data center cooling is typically relegated to the grey space, Accelsius is committed to providing you with a solution that has the fit, form, function, and finish of a white-space product. We deliver cooling to white-space engineers who treat their data halls like a McLaren: those who want their data center infrastructure to be “sleek,” “powerful,” and “built to last.” We couldn’t think of a better way to describe our NeuCool products.

We’ve thoroughly analyzed our product to eliminate any roadblocks to unparalleled cooling performance. We don’t use plastic or multi-component systems, which save a few cents while sacrificing reliability. Instead, we utilize enterprise-grade materials throughout every facet of our design, including:

  • Multi-layer braided hoses for durability
  • Automotive-grade fittings and tubing
  • All-metal vaporators designed from the ground up

We’re not here to cut corners. We’re here to provide cutting-edge cooling to empower cutting-edge compute.

Excellence Doesn’t Mean Excess

Of course, this means it’s time to dispel another myth. Some may assume that “excellence” always equals “expensive”; or, rather, that “runaway performance” can only be enabled by “runaway costs.” This couldn’t be further from the truth.

Intentionality is the best cure for excess—and we’ve been intentional with our design since our company’s inception. Unlike our competitors, whose products cost extra thousands of dollars for needless complexity and “component overkill,” we’ve ensured every part of our product is necessary, scalable, and easy to maintain.

If anything, this approach to excellence prevents those pesky runaway costs from occurring. We’ve written elsewhere about how NeuCool™ leads to a lower TCO over time vs. single-phase cooling. That’s because our non-conductive refrigerant, which won’t cause any equipment failure if leaks arise, maximizes your ROI by maximizing your uptime. Plus, recent testing has also shown that we’re able to operate with facility water that’s 6-8°C warmer than other liquid cooling technologies, which leads to over 25% energy savings.

In short, we’ve dedicated ourselves to ensuring there’s zero waste in our product design. In turn, for years to come, you won’t have to suffer from wasteful spending.

Our Multi-Rack McLaren

Our upcoming multi-rack CDU, the NeuCool MR250, exemplifies our approach. It’s built to cool 250kW+ workloads across up to four racks, scaling fluidly with current and future industry releases:

  • 8-way GPU rack configurations, for example? The MR250 can support those.
  • NVIDIA’s NVL72? The MR250 is perfectly sized for two racks, and even offers additional capacity for the GB300.
  • The first set of upcoming Rubin racks, slated to arrive in 2027? That’s right—the MR250 can cool those, too.

The MR250, like any of our other products, is modular, hot-swappable, and beautifully finished. Beyond that, however, it sets you up for future success. Excellence shouldn’t only exist in the short term, after all. Instead, we’ve thought ahead to where the market is headed for the next few years—while still providing what’s needed to excel today.

The Final Lap

By now, it should be clear: at best, a trend-chasing approach to advanced cooling will leave you squarely in the middle of the pack. In a world where chip development cycles are measured in months, not years, you can’t afford to wait for commoditized cooling to catch up.

Instead, choose the solution that’s raced to the top. Choose NeuCool.