In this Q&A, Gordon Johnson, Senior CFD Manager at Subzero Engineering, explores the pivotal role CDUs play in transforming data center operations. From maintaining system integrity and adapting to variable AI workloads to driving energy efficiency and future scalability.

An Interview with Gordon Johnson, Senior CFD Manager at Subzero Engineering

Overview

As data centers evolve to meet the exponential power and cooling demands of AI and high-performance computing (HPC), traditional air-cooling methods are reaching their limits. Direct Liquid Cooling (DLC) has emerged as the next frontier in thermal management, offering the precision and efficiency needed for today’s high-density workloads. At the heart of this innovation lies the Coolant Distribution Unit (CDU) – the intelligent control layer that ensures safety, stability, and sustainability in liquid-cooled environments.

QUESTION: What initiated the transition towards Direct Liquid Cooling, and what makes CDUs crucial for development?

The increase in AI and high-performance computing (HPC) demands has pushed rack power densities beyond the capabilities of conventional air-cooling strategies. High-end server CPUs and GPUs are approaching or even exceeding 700–1000 watts per socket, especially in AI and HPC deployments, and air-cooling can’t remove the generated heat amounts effectively. Direct Liquid Cooling (DLC), especially Direct-to-Chip (DTC) cooling, delivers accuracy and thermal capacity needed at these densities. However, DLC works effectively with a control layer, which is the CDU, allowing for regulations on pressure, flow, and temperature to guarantee consistent thermal performance. It transforms liquid cooling from a mechanical method into an adaptive system to accommodate high-density workloads.

QUESTION: In what ways does the CDU maintain safety and stability in a liquid-cooled data center?

The CDU acts as a safeguard between the facility’s water system (FWS) and the IT equipment. It separates the technology cooling system (TCS) loop from pollutants, pressure variations, and chemical substances present in the FWS loop, guaranteeing that the fluid supplied to the cold plates attached directly to the CPUs and GPUs is pure, regulated, and non-conductive.

By utilizing integrated sensors, pumps, and heat exchangers, the CDU regulates the ideal coolant temperature and pressure, avoiding condensation, and safeguarding hardware from thermal fluctuations. This separation protects IT resources as well as improving predictability and operational performance throughout the cooling system.

QUESTION: AI and HPC tasks are recognized for quick changes in power consumption. How do CDUs handle this uncertainty?

AI training and inference tasks are highly variable. GPUs can either increase or decrease performance quickly, resulting in immediate temperature surges. CDUs address these changes by adjusting pump speeds, flow rates, temperature, and valve positions to evenly distribute the load.

By continuously adapting to the workflow demand, CDUs maintain stable pressure and temperature throughout all racks. This degree of regulation guarantees that even in unstable environments, operations can run with limited risks and maintain a prolonged equipment lifecycle.

QUESTION: Are there different types of CDUs for varying facility sizes and requirements?

CDUs are usually classified into two primary groups:

• Liquid-to-Liquid CDUs: These utilize heat exchangers to move heat from the TCS (IT coolant loop) to the FWS and are best deployed for large-scale or HPC data centers with existing chilled water infrastructure.
• Liquid-to-Air CDUs: These expel heat straight into the ambient air within the data center via an internal exchanger, rendering them ideal for smaller or edge facilities, and where a FWS or chilled water is unavailable.

Both offer similar control, isolation, and safety advantages, but the decision typically relies on the site’s existing infrastructure and cooling capability.

QUESTION: In what ways do CDUs help achieve sustainability and energy efficiency goals?

CDUs play a crucial role in sustainable thermal management as they help avoid excessive pumping and cooling, which are significant causes of energy wastage in data centers. Smart flow and temperature regulation enhance Power Usage Effectiveness (PUE) and Water Usage Effectiveness (WUE), with PUE figures at times even reaching below 1.1.

Also, CDUs facilitate the integration of waste heat recovery systems, enabling operators to utilize surplus heat for district heating or industrial applications. This converts a byproduct into a sustainability resource, decreasing overall energy use and promoting circular energy approaches.

QUESTION: How do CDUs enhance scalability for future liquid cooling installations?

A significant benefit of the CDU is its flexibility, employing mixing and bypass control to adjust liquid coolant for particular IT loads. This adaptability allows operators to gradually implement liquid cooling, without the need for complete infrastructure overhauls. Therefore, CDUs render liquid cooling attainable, expandable, and secure, enabling data centers to progress alongside workload requirements and sustainability objectives.

QUESTION: What part do CDUs play in influencing the future of thermal management?

CDUs serve as the thermal regulation system in contemporary data centers. They provide intelligence, isolation, and efficiency to Direct Liquid Cooling, ensuring scalability, and sustainable functioning of high-density settings, and create an accurate framework that is purposefully designed for the AI era.

About Subzero Engineering:

Subzero Engineering specializes in providing turnkey, precision-engineered data center containment solutions that are designed for industry-leading functionality and scalable fast deployments. With a focus on sustainability, our solutions are built to provide maximum efficiency, minimize downtime, and reduce energy consumption. Our team of experts works closely with clients to understand their unique needs and requirements, and we pride ourselves on delivering customized solutions that exceed expectations. Whether you’re looking for a new data center build or an upgrade to an existing facility, our turnkey solutions are designed to deliver exceptional results. Since 2005, we have been containing critical environments and providing exceptional solutions to the world’s most demanding technology companies.

www.subzeroeng.com

As AI and HPC workloads push rack densities beyond the limits of air cooling, Coolant Distribution Units have emerged as the intelligent control layer that makes direct liquid cooling scalable, stable, and safe.

By Subzero Engineering

Why Direct Liquid Cooling Requires an Intelligent Control Layer

Traditional air cooling has hit its limits as rack power densities surpass 100 kW due to the relentless growth of AI and high-performance computing (HPC) workloads. Already, CPUs and GPUs exceed 700–1000 W per socket, while projections estimate 1500 W+ going forward. Fans and heat sinks are just unable to handle these thermal loads at scale; air cooling has hit its limits.

Hybrid cooling strategies are becoming the only scalable, sustainable path forward.

Single-phase direct-to-chip (DTC) liquid cooling has emerged as the most practical and serviceable solution, delivering coolant directly to cold plates attached to processors and accelerators. However, Direct Liquid Cooling (DLC) cannot be scaled safely or efficiently with plumbing alone. The key enabler is the Coolant Distribution Unit (CDU) – a system that integrates pumps, heat exchangers, sensors, and control logic into a coordinated package.

CDUs are often mistaken for passive infrastructure. But far from being a passive subsystem, they act as the brains of DLC, orchestrating isolation, stability, adaptability, and efficiency to make DTC viable at data center scale. They serve as the intelligent control layer for the entire thermal management system.

Intelligent Orchestration

CDUs do a lot more than just transport fluid around the cooling system. They think, adapt, and protect the liquid cooling portion of the hybrid cooling system. They maintain redundancy to ensure continuous operation, control flow, and pressure using automated valves and variable speed pumps, filter particulates to protect cold plates, and maintain coolant temperature above the dew point to prevent condensation. They contribute to the precise, intelligent, and flexible coordination of the complete thermal management system.

Because of their greater cooling capacity, CDUs are ideal for large HPC data centers. However, because they must be connected to the facility’s chilled water supply or other heat rejection source to continuously provide liquid to the cold plates for cooling, they can be complicated.

CDUs typically fall into two categories:

• Liquid to Liquid (L2L): Large HPC facilities are well-suited for high-capacity CDUs known as L2L. Through heat exchangers, they move chip heat into the isolated chilled water loop, such as the facility water system (FWS). 
• Liquid to Air (L2A): For smaller deployments, L2A CDUs are simpler but have a lower cooling capacity. By utilizing conventional HVAC systems, they transfer heat from the returning liquid coolant from the cold plates to the surrounding data center air by using liquid to air heat exchangers rather than a chilled water supply or FWS.

Isolation: Safeguarding IT from Facility Water

Acting as the bridge between the FWS and the dedicated technology cooling system (TCS), which provides filtered liquid coolant directly to the chips via cold plate, CDUs isolate sensitive server cold plates from external variability, ensuring a safe and stable environment while constantly adjusting to shifting workloads.

One of L2L CDU’s primary functions is to create a dual-loop architecture:

• Primary loop (facility side): connects to building chilled water, district cooling, or dry coolers.
• Secondary loop (IT side): delivers conditioned coolant directly to IT racks.

CDUs isolate the primary loop (which may carry contaminants, particulates, scaling agents or chemical treatments like biocides and corrosion inhibitors – chemistry that is incompatible with IT gear) from the secondary loop. As well as preventing corrosion and fouling, this isolation offers operators the safety margin that operators need for board-level confidence in liquid.

The integrity of the server cold plates is safeguarded by the CDU, which uses a heat exchanger to separate the two environments and maintain a clean, controlled fluid in the IT loop. Because CDUs are fitted with variable speed pumps, automated valves, and sensors, they can dynamically adjust the flow rate and pressure of the TCS to ensure optimal cooling even when HPC workloads change.

Stability: Balancing Thermal Predictability with Unpredictable Loads

HPC and AI workloads are not only high power, they are also volatile. GPU-intensive training jobs or changeable CPU workloads can cause high-frequency power swings, which without regulation, would translate into thermal instability. The CDU mitigates this risk by controlling temperature, pressure, and flow across all racks and nodes, absorbing dynamic changes and delivering predictable thermal conditions.

The CDU absorbs fluctuations by stabilizing temperature, pressure, and flow across all racks and nodes, regardless of how erratic the workload is. Sensor arrays ensure the cooling loop remains in accordance with specifications, while variable speed pumps modify flow to fit demand, and heat exchangers are calibrated to maintain an established approach temperature.

Adaptability: Bridging Facility Constraints with IT Requirements

The thermal architecture of data centers varies widely; some use warm-water loops that operate at temperatures between 20 and 40°C. By adjusting secondary loop conditions to align IT requirements with the facility, the CDU adjusts to these fluctuations. The CDU uses mixing or bypass control to temper supply water. It can alternate between tower-assisted cooling, free cooling, or dry cooler rejection depending on the environmental conditions, and it can adjust flow distribution among racks to align with real-time demand.

This adaptability makes DTC deployable in a variety of infrastructures without requiring extensive facility renovations. It also makes it possible for liquid cooling to be phased in gradually – ideal for operators who need to make incremental upgrades.

Efficiency: Enabling Sustainable Scale

Beyond risk and reliability, CDUs unlock possibilities that make liquid cooling a sustainable option.

By managing flow and temperature, CDUs eliminate the inefficiencies of over-pumping and over-cooling. They also maximize scope for free cooling and heat recovery integration such as connecting to district heating networks and reclaiming waste heat as a revenue stream or sustainability benefit. This allows operators to simultaneously lower PUE (Power Usage Effectiveness) to values below 1.1 while simultaneously reducing WUE (Water Usage Effectiveness) by minimizing evaporative cooling. All this, while meeting the extreme thermal demands of AI and HPC workloads.

CDUs as the Thermal Control Plane

Viewed holistically, CDUs are far more than pumps and pipes. They are the thermal control plane for thermal management, orchestrating safe isolation, dynamic stability, infrastructure adaptability, and operational efficiency.

They translate unpredictable IT loads into manageable facility-side conditions, ensuring that single-phase DTC can be deployed at scale, enabling HPC and AI data centers to evolve into multi-hundred kW racks without thermal failure.

Without CDUs, direct-to-chip cooling would be risky, uncoordinated, and inefficient. With CDUs, it becomes an intelligent and resilient architecture capable of supporting 100 kW and higher racks and the escalating thermal demands of AI and HPC clusters.

As workloads continue to climb and rack power densities surge,  the industry’s ability to scale hinges on this intelligence. CDUs are not a supporting component. They are the enabler of single-phase DTC at scale and a cornerstone of the future data center.

While direct-to-chip cooling dominates today’s data centers, immersion cooling may ultimately define how the industry scales performance, efficiency, and sustainability tomorrow.

By Subzero Engineering

The Safe Choice Today vs. the Scalable Choice Tomorrow

For many large-scale deployments, direct-to-chip (DTC) single-phase cooling has emerged as the market’s preferred direct liquid cooling (DLC) technique.

It is easy to see why. DTC is dependable, well-established, and reasonably simple to incorporate into the current data center infrastructure. For risk-sensitive facilities that are wary of operational disruption or retrofit headaches, DTC is the logical choice, and that’s why it has become the dominant DLC standard.

But the “logical choice today” is not the same as the “best choice for the future.” Technically, the superior solution is immersion cooling. With the ability to support denser racks than air cooling or DTC, immersion cooling offers higher heat removal capacity by immersing entire servers in dielectric fluids. Right now, immersion cooling is mostly used in specific areas such as crypto mining, experimental high-performance computing, and some edge computing setups. However, it hasn’t gained much popularity in the data center market yet, mainly due to high initial costs, the need for special infrastructure, and the challenges of training people to use it.

But with the consistent trajectory of compute, energy economics, and environmental pressures, is immersion cooling simply waiting for its moment to shine?

Why DTC Leads Today

Today, DTC dominates the market due to its ease of use. DTC solutions can be deployed in standard racks with minimal retrofits, data center teams don’t need to undergo extensive retraining, and maintenance procedures stay somewhat familiar. More than anything else, DTC is a straightforward solution for organizations that are unable or unwilling to re-architect their environments for immersion cooling in the face of increasing power densities.

Why Immersion Lags

Immersion Cooling is still in the early adoption and growth phase rather than being fully mature however, its efficiency improvements indicate that it will soon move from being a specialized solution to a vital component of hyperscalers’, HPC operators’, and edge deployments’ thermal management arsenal.

So why isn’t immersion adopted more often if it is a technically superior solution?

• High CAPEX: Immersion costs are significantly more than DTC retrofits because it requires specialized tanks, dielectric fluids, pumps, and monitoring systems.
• Specialized Infrastructure: Racks replaced with tank-based designs require redesign of cabling, power distribution, and architectural layouts.
• Learning Curve: Hardware compatibility, fluid handling, and maintenance considerations all call for retraining and skill sets.

The high initial expenditures, specific infrastructure requirements, and a process that hasn’t yet unified on a single model are the main obstacles to immersion today. Therefore, is it too disruptive to consider seriously?

The Case for Immersion

History has shown us that often the “niche” of today becomes the necessity of tomorrow.

Immersion cooling solves problems that DTC can only mitigate.

• Unmatched thermal performance: Immersion can absorb heat from all components, not just CPUs and GPUs, by completely submerging servers in dielectric fluids.
• Extreme density potential: Racks can be packed much more densely without needing to depend on airflow. This enables higher compute density per square foot.
• Energy efficiency: Immersion cooling can significantly lower power usage associated with cooling. Power usage effectiveness (PUE) levels of 1.02 to 1.05 have been observed by some operators.
• Sustainability: Immersion does not use water as the primary cooling method, which is a developing benefit in areas with water scarcity compared to evaporative cooling technologies. However, water may still be used indirectly to carry heat away from the dielectric fluid via heat exchangers and via chillers or cooling towers to reject heat from the building.
• Hardware longevity: Immersion can increase the useable life of servers by removing hot spots and thermal cycling.

Additionally, immersion cooling eliminates the need of airflow from data center design. Operators can completely redesign facility layouts without the requirement for air-handling infrastructure, raised floors, or large HVAC systems. However, experts in both air and immersion cooling still predict a future where both air and immersion will coexist in the industry depending on specific cooling needs.

What Will Force the Shift?

Operators cannot afford ineffective thermal management as power costs grow, and the trajectory of compute power continues to rise. Rack power densities are already increasing beyond that which air and DTC cooling can sustainably handle as processors get near 1000W+ TDP levels. Energy prices remain volatile and demands for efficiency and sustainability are being driven by environmental challenges.

Can immersion’s higher upfront investment be justified when weighed against long-term energy savings, environmental benefits, and the ability to extend hardware lifecycles?

Tipping Point

In the near term, DTC will remain the workhorse, as it’s good enough to buy operators time without demanding a full architectural reset. However, immersion’s tipping point won’t arrive because the technology suddenly becomes more appealing. It will arrive when nothing else works. DTC won’t scale forever.

The long-term goal will likely be immersion. Workload trends, power densities, and sustainability requirements all point to a future in which immersion is not only viable but the only workable option.

At that point, immersion will go from trailing to prevailing, but the changes will go beyond thermal efficiency. Data center architecture will evolve, altering not just cooling strategies but the data center’s architecture. A new approach will be unlocked by tank-based designs, fluid-centric maintenance, and different hardware form factors, where cooling will no longer be a limitation but rather a facilitator of sustainability, performance and efficiency.

This is about enabling the next generation of computational infrastructure, not simply about cooling.

Conclusion

The issue is not if immersion will catch up, but rather when high-density workloads will force the shift.

When it does, will operators be prepared? Those who still view immersion as a fringe experiment run the danger of having to rush to catch up when DTC reaches its limit, whereas those early adopters will be able to take the lead since they will already be training, developing, and cultivating knowledge and expertise.

For data centers, the future of cooling won’t be about being cautious. It will belong to those who are brave enough to wager on immersion before the tipping point happens.

Africa’s data center industry is expanding at record speed, but a critical shortage of skilled professionals threatens its ability to sustain long-term growth.

By Subzero Engineering

Rapid growth is redefining infrastructure across Africa, exposing a critical gap in skilled human capital.

Africa’s data center capacity has more than doubled over the last five years, and analysts predict that the industry will have more than 1,000 MW of installed IT load by 2030.

With new facilities rushing to go online at an unprecedented rate, the meaning of “infrastructure” for African economies has changed because of the fintech revolution, e-commerce, and the rapid increase of cloud use.

However, beneath the optimism, Africa’s data center sector faces bottlenecks in financing, power reliability, and regulation. One of the most urgent issues to resolve is an acute shortage of skilled workers. The human infrastructure necessary for the operation and maintenance of this digital ecosystem is trailing significantly behind investments in physical infrastructure.

The faster the industry grows, the more exposed it becomes to its own human-capital deficit, and without a pipeline of skilled professionals, the continent’s data center ambitions risk outpacing its ability to sustain them.

The Infrastructure Challenge

Land, power, and regulation are the tangibles that data center investors typically concentrate on, but how these elements interact shows the difficulties of doing business in African marketplaces.

Grid instability is continuing to be a problem in more developed African economies like South Africa, but power problems are far more severe in Nigeria, where grid reliability can be less than 50% in many parts. As a result, energy costs can be significantly higher than comparable facilities in the Middle East or Europe.

Complicated licensing processes, shifting tax regimes, and inconsistent specifications and certifications all add to regulatory uncertainty, which hinders planning and delays projects. In addition, the devaluation of African currencies makes it difficult to fund major infrastructure projects, discourages long-term commitments, and increases the cost of imported equipment.

Nevertheless, the continent would still find it difficult to satisfy the expectations of its digital infrastructure sector even if these challenges were resolved tomorrow. The most critical issue is the lack of a robust workforce.

The Talent Gap

Operators across the continent report difficulty in finding competent engineers and technicians who understand the complex interactions that are required to run a modern facility.

Building and operating Tier III or Tier IV data centers requires specialized skills that are acquired over years. Professionals with specific expertise in international certification requirements, redundancy design, or high-density facilities are scarce at this level.

It doesn’t help that Africa is exporting the very skills it most urgently needs. Data center professionals are drawn to opportunities in Europe, the Middle East, and North America, where compensation is higher and professional development pathways have more clarity.

As a result, the continent trains but fails to retain its most prized human resources. Facilities lose experienced engineers, project teams lose continuity, and the next generation loses mentors. Organizations are forced into reactive hiring wars, poaching staff from competitors instead of expanding the market’s talent pool.

Invisible Industry

The shortage of skilled professionals in Africa’s data center sector is the predictable consequence of the way technical education, industry structure, and investment priorities have evolved.

For many young African engineers, they’ve been educated that ‘tech’ does not refer to the actual infrastructure but rather to software, coding, finance, or cloud apps.

Ironically, one of the most technically complex career pathways is the data center business, which is situated at the convergence of digital, electrical, and mechanical systems. However, without visibility, neither students nor early-career professionals will know that.

Fragmented Certification

Programs in critical infrastructure engineering and data center operations are not commonly offered at African educational institutions. Despite producing skilled graduates in IT, mechanical, and electrical engineering, it is rare to find those that have worked with mission-critical infrastructure.

Few regional institutions are certified to deliver Tier or BICSI training. Cost, travel, and the lack of recognized local training facilities are the main barriers to accessing these programs in Africa. The result is that only a small elite can afford to obtain global credentials, with most operators relying on informal or vendor-led training.

Underinvestment in Workforce Development

With limited funding and short project timelines, many operators choose to forego investment in structured development programs that foster long-term capability in favor of expecting recruits to arrive ‘job ready’.

When margins are tight, training budgets are frequently the first to be slashed. Businesses complain about the lack of trained labor, but few are building the foundation to generate it. Even when training is offered, it is often inward facing, instructing employees on specific site operations without developing those broader skill sets that are transferable across the industry.

Turning the Gap into an Opportunity

Classroom schooling alone is unlikely to be effective. Apprenticeship programs that blend academic instruction with practical training and rotating mentorship programs across power, cooling, and network teams expedite and establish operational readiness.

Organizations collaborating with local colleges to establish training centers can facilitate access to resources, educational programs, and certification systems that meet globally recognized standards. Successful cases in Asia and the Middle East show that such partnerships can revolutionize national skill sets in just a few years.

Practical governmental policy levers include co-funding vocational programs, accelerating the certification process, and offering tax incentives to companies that invest in training. Pilot projects are yielding promising results in South Africa and Keya, but an ongoing commitment is needed to scale these programs across the continent.

Sustainable Growth

Africa will continue to attract investment due to its thirst for data, artificial intelligence, and cloud computing, but this investment won’t result in sustained growth unless the industry develops its people infrastructure with the same vigor as its physical infrastructure.

Personal growth, not just compensation, determines retention. Operators that provide possibilities for international certification, exposure to innovative technologies, and clear career development pathways will see that employees are far less likely to leave when they see opportunities for advancement at home.

The fastest return on investment in the data center industry is the investment in people. It is important to remember that every engineer trained today can become a teacher for the next generation. Organizations that invest in workforce development, embed training, and retain experienced professionals will pull away from those that don’t. The race is not against time or technology; it’s against the widening talent gap, and Africa’s young population represents its greatest opportunity to build a sustainable digital backbone.