Data center infrastructure providers are exploring every avenue in search of the best way to handle data center cooling, whether it’s air, liquid, or a combination of both. Today, many are reaching a critical point of transition from one to the other.

As Airedale by Modine’s global product manager for liquid cooling, Richard Burcher leads the strategic direction, delivery, and execution of the company’s liquid cooling programs, and shares his perspective on the evolving data center liquid cooling and CDU landscape.

Burcher captures the current state of the market in one sharp phrase: high-level explosive growth.

At this point, and for the foreseeable future, no discussion about liquid cooling can take place without mentioning AI. With its variable loads and high-intensity surges, AI is driving a sharp rise in data center heat output – and with it, the need to remove that heat more efficiently from critical equipment.

While established colocation providers and hyperscalers are increasingly interested in liquid cooling, they must weigh adoption against the realities of their existing infrastructure and the skill sets of their service engineers. In contrast, a new generation of specialized cloud providers – ‘neoclouds’ – are rapidly building high-density, GPU-driven environments from the ground up, making liquid cooling a foundational requirement rather than a retrofit.

Whether these players have evolved organically or entered the sector through more inorganic means, the industry as a whole is in a high-growth phase that demands significant investment to keep that momentum going.

“Neoclouds, for example, are expanding rapidly – over 200 percent year-on-year – with heavy investment from REITs fueling an ecosystem projected to reach about $180 billion by 2030,” says Burcher, adding:

“Across the board, momentum is strong, and with liquid cooling growing at around 20 to 30 percent CAGR, it’s clear that innovation and efficiency are becoming central to how this industry evolves.”

Translating industry development into liquid cooling needs

Energy and sustainability pressures are making operators more considered in their approach to liquid cooling adoption, as retrofitting legacy facilities demands careful evaluation of cost, performance, and heat-reuse ambitions – particularly when higher coolant temperatures support sustainability targets but can limit the chip-level performance that GPU-driven workloads require.

The skills and integration expertise required for these upgrades – whether operators are incorporating liquid cooling or enhancing their existing air or hybrid systems for greater efficiency – are equally demanding. Microfiltration systems, for example, must be easy to clean and maintain to ensure continuous, reliable operation. Yet, it’s worth remembering that liquid systems are already commonplace in other industries, and learning from those established practices will be vital to mastering them in data centers.

“Fluids need to be managed carefully. You have to make sure they’re compatible with every part of the system, from the TCS loop to the CDU and facility water side. That’s why filtration is such an important part of reliable operation,” explains Burcher.

All these tangible challenges make liquid cooling highly specialized, underscoring the need to work with experienced suppliers who can simplify the process. But there are also intangible barriers:

“Trust is a big one. When you’re working with new technology, you need a trusted provider – and trust across the entire ecosystem. There are still questions around delineation of responsibility: who owns which part of the chain?” says Burcher.

Without large-scale, real-world examples of liquid cooling in operation, the technology can still appear to many operators as something of a dark art. Through greater knowledge sharing, the development of best practices, and open collaboration, the industry can simplify processes and demystify the technology. Burcher adds:

“Associations have a big role to play here. ASHRAE in North America is already working on standards, and the Open Compute Project is doing great work too. But we need to go further – to bring more openness, continue to break down barriers, and accelerate progress on best-fit solutions. That’s beginning to change.”

Leveraging trusted expertise to drive next-generation cooling

The pace of change and innovation in the data center industry can be overwhelming. In this landscape, successful cooling increasingly hinges on early engagement with thermal management experts.

For Airedale, Burcher explains that this expertise comes from more than 50 years of engineering experience and a global network of expertise to adapt as technologies evolve:

“That gives us an incredible breadth of knowledge in thermal management, combined with global execution, at scale that we can call upon. Being relationship-driven is pivotal. We’ve worked hand in glove with key customers, grown alongside them, and faced challenges together to develop the best-fit solutions – designing for today, but always with tomorrow in mind,” says Burcher.

By collaborating early, operators gain access to deep technical insight and a system-level perspective that looks beyond individual components to the impact of the entire facility.

“Our core values remain unchanged, and they are essential to our success as we grow into new markets and address diverse needs. While we must continue to evolve – since no single solution fits every situation – we do so with a clear commitment to the principles our customers rely on and trust us for,” Burcher adds.

“Our ability to engineer at scale and engage early allows us to deliver fully integrated, high-performing systems. In liquid cooling especially, that early collaboration is what ensures efficiency from the chip to the chiller.”

Turning a hindrance into a help

Densify, densify, densify – that’s the message being delivered to data center operators as they support the GPUs powering today’s compute-intensive workloads. AI-driven demand is reshaping where and how compute is deployed, prompting operators to rethink every aspect of infrastructure, from power distribution to cooling strategy. One approach gaining traction across the industry is skid-based systems, which offer modular, scalable solutions capable of meeting the requirements of high-density environments.

In the construction world, modular ‘skid’ systems are well-established, but in data centers, they’ve been slower to take hold. The industry’s focus on bespoke design is now showing its limits as the need for rapid scaling grows. The result is greater complexity, with facilities weighed down by legacy infrastructure and little room to adapt.

So, what exactly are skid-based systems in the context of liquid cooling, and why are they viewed as a future-proof solution?

These self-contained, modular units integrate key components of a cooling process, such as pumps, heat exchangers, and controls, onto a single frame or base. Designed for efficient layering, optimized footprint, ease of service and maintenance with centralized connection points, and can be built off-site to avoid disrupting ongoing operations – a critical advantage for uptime and SLA compliance.

This approach replaces traditional onsite construction with the delivery of complete, ready-to-install units.

“As providers look to densify, they’re also demanding higher capacities. At the same time, they want to maintain clear delineation and limit access to IT spaces. Skid-based applications align perfectly with that – offering greater kilowatt capacity and modularity to scale up efficiently,” Burcher explains, adding:

“As GPUs evolve and generate more heat at the chip level, overall server densities increase – and that’s driving greater demand for skid-based solutions. We’re seeing a lot of RFIs and active development in that space.”

Don’t just make it – make it matter

The next step is being intentional in designing and deploying AI-ready environments that deliver real, measurable impact. As adoption deepens, the industry is starting to see genuinely high-value applications, from material innovations to advanced design tools, as we strive to realise the art of the possible. Burcher highlights how these trends are shaping the industry:

“There are lots of things keeping me awake at night – but in a good way. In particular, the convergence of AI technologies and their increasingly impactful applications will drive greater demand around latency, more Edge-based solutions, and developments in immersion cooling, especially in scenarios where compute is both produced and consumed at the same location.”

Burcher also anticipates heightened competition – not only from established chip developers like Nvidia and AMD, but from hyperscalers now designing their own silicon. He sees this growing diversity as a strength, not a threat.

“Competition drives progress. Every region is developing differently, and that’s a good thing – it means we can learn faster by sharing what works and what doesn’t. It’s about learning in the right way – protecting your systems, managing your data, and ensuring what you share is intentional and secure,” he explains.

Stressing the importance of responsibility, he brings the conversation full circle with a powerful reflection:

“In our manufacturing processes, we ensure everyone in the value chain understands their impact. Every stakeholder and employee knows what we’re building, the critical role it plays, where these systems go, and how they connect with them every day.”

With local teams attuned to the operational context, every deployment is planned to maximize efficiency and reliability. By embracing a holistic approach to system integration, air and liquid cooling technologies – including CDUs, immersion systems, chillers, and CRAHs – are seamlessly coordinated through intelligent controls. This orchestration ensures the data center operates as a unified, efficient ecosystem rather than a collection of isolated components. Ultimately, the whole delivers more value than the sum of its parts.

The message is clear: embrace the hypergrowth, but do it responsibly – with repeatability, precision, and purpose.

To learn more about Airedale by Modine’s approach to data center cooling, please visit airedale.com.