For data centers expanding to meet the demands of AI workloads, cooling is no longer a background requirement. It has become the defining constraint. Energy consumption, water usage, and heat rejection now influence where facilities can be built, how large they can grow, and whether they meet sustainability targets. According to Anders Almhem, Sales Manager at Bürkert Fluid Control Systems, solving that challenge requires more than adding larger chillers or more pumps. It requires inventing new approaches to fluid control.
“Cooling efficiency is no longer just a line item on an operating budget,” Almhem explains. “It is directly tied to sustainability goals, water stewardship, and long-term scalability. If we do not engineer the cooling systems intelligently, we limit the future of data center expansion.”
Cooling Has Become the Bottleneck
Modern AI-driven facilities generate extraordinary heat densities. Liquid cooling has emerged as the preferred solution for managing those loads, particularly as traditional air-cooling approaches reach their limits. However, simply switching to liquid does not automatically guarantee efficiency.
Most liquid-cooled data centers operate with two distinct fluid loops. The first, often called the primary or facility loop, circulates water sourced from municipal systems or groundwater to remove heat from the building. The second loop carries treated coolant through heat exchangers and into server racks. Each loop requires pumps, monitoring, and energy. Each loop represents both opportunity and inefficiency.
“When you look at the architecture,” Almhem says, “you see that every loop is drawing power, and in many cases systems are running at full capacity even when the workload does not demand it. That is where optimization becomes critical.”
AI workloads fluctuate constantly. Some racks operate at peak capacity while others idle. Without precise control, cooling systems tend to overcompensate, pushing unnecessary flow and consuming excess energy. The result is higher power usage and, in evaporative systems, increased water consumption.
Precision as a Sustainability Strategy
For Bürkert, the solution lies in intelligent fluid control. The company’s expertise centers on integrating sensors, valves, and control systems so that cooling responds dynamically to real-time conditions rather than operating at fixed output levels.
“The closer you bring proportional control to the actual heat source, the more efficient the system becomes,” Almhem explains. “If a section of servers is not active, you should not be running full flow through that loop. When demand increases, the system should respond instantly and precisely, not broadly and inefficiently.”
By combining temperature, flow, pressure, and conductivity measurements with rapid-response proportional valves and variable pumping strategies, cooling systems can be tuned continuously. This reduces pumping energy, limits unnecessary water throughput, and minimizes waste heat. In an era where data center approvals increasingly depend on environmental performance, that precision can determine whether projects move forward.
A History of Solving Extreme Cooling Challenges
Although liquid cooling at scale is relatively new within the data center industry, Bürkert’s experience with complex thermal management spans decades. The company’s technologies have been deployed in highly demanding environments, including cooling applications associated with the Large Hadron Collider. The LHC accelerates protons and ions to nearly the speed of light, creating conditions that demand extraordinary thermal stability and control.
“In projects like the LHC, there was no off-the-shelf solution,” Almhem says. “It required collaboration with manufacturers and engineers to create systems that had never existed before. That same mindset is what data centers need now.”
The relevance is not about scale alone, but about philosophy. When thermal loads intensify and tolerances narrow, the answer is not simply more cooling capacity. It is smarter, more controlled fluid architecture.
Partnering with Manufacturers to Invent What’s Next
Today, Bürkert is applying that collaborative approach within the data center ecosystem, particularly with coolant distribution unit manufacturers. CDUs serve as the bridge between facility water and server-level coolant loops, and their performance depends heavily on accurate flow measurement, pressure regulation, and fluid quality monitoring.
“If a CDU is rated to remove a specific heat load, it must be supported by reliable data,” Almhem notes. “Flow meters, conductivity sensors, and proportional control valves are not optional add-ons; they are fundamental to delivering that performance.”
As chip developers continue to increase power densities, some manufacturers are exploring even more granular cooling strategies. Concepts such as rack-level or chip-level proportional control are moving from theory toward implementation. Bürkert’s modular, sensor-integrated systems allow manufacturers to experiment, adapt, and refine their architectures as the technology evolves.
Enabling Sustainable Growth
The broader implication is clear. Data centers are often described as the new power plants of the digital economy, yet their expansion is increasingly constrained by cooling efficiency and water availability. If computing power continues to grow exponentially, cooling systems must evolve just as quickly.
“Liquid cooling is not a new or frightening concept,” Almhem emphasizes. “Precise fluid control has been around for decades. What is new is applying that experience to the scale and speed of AI infrastructure.”
For engineers and commercial manufacturers entering this space, the message is reassuring. They do not need to reinvent fluid control from scratch. They need partners with a proven history of developing custom solutions for complex thermal challenges.
Ultimately, the sustainability of tomorrow’s digital infrastructure will depend not only on the performance of its processors but also on the intelligence of its cooling systems. Bürkert’s role, as Almhem describes it, is to help invent those systems, ensuring that data centers can expand efficiently, responsibly, and without compromise.
“The cloud feels invisible,” he says. “But the fluid systems behind it are very real. And if we engineer them correctly, we can support both innovation and sustainability at the same time.”