Date: October 6, 2026
Author: Catie Owen, Regional Content & Insights Lead
Publication: Capacity Global
Executive Overview
The explosive global expansion of artificial intelligence (AI) is routinely measured in billions of dollars, gigawatts of electrical capacity, and millions of square feet of real estate. Yet, beneath the digital sheen of AI’s ascent lies a fundamental, unyielding physical reality: the massive generation of heat. As the industry accelerates past traditional operational paradigms, modern server rack densities are aggressively scaling past 100 kilowatts (kW), with many cutting-edge deployments pushing into the 130 kW threshold.
This unprecedented thermal footprint has turned the data centre industry upside down. The central engineering question of the digital age is no longer simply how to feed more compute clusters with reliable power, but rather how to cool those systems effectively without exhausting local water tables, straining regional electrical grids, or colliding with community sustainability mandates.
In this shifting landscape, water and liquid cooling have transformed from alternative engineering strategies into core elements of a facility’s "license to operate." To unpack the complexities of modern thermal management, Capacity sat down with Matthew McLaughlin, Corporate Account Manager for Ecolab’s dedicated data centre business.
According to McLaughlin, the industry can no longer afford to view power, water, and heat rejection as siloed operational challenges. Instead, navigating the AI boom requires a unified approach—one that marries the superior physics of liquid cooling with real-time chemical monitoring and a holistic accounting of resources both inside and outside the facility fence.
The Physics of Thermal Management: Why Air Cooling Has Reached Its Limit
For decades, the data centre industry relied on conventional air-cooling systems. Massive arrays of chillers, computer room air handlers (CRAHs), and raised floors pushed colossal volumes of air across server components to dissipate heat. While sufficient for legacy workloads, this linear, air-based model is fundamentally incompatible with the extreme thermal densities of modern AI accelerators and high-performance computing (HPC) clusters.
McLaughlin urges stakeholders to return to fundamental physics to understand why change is mandatory.
"Liquid—specifically water—moves heat far more effectively than air," McLaughlin explains. "Water can carry roughly 3,500 times more heat than the same volume of air and transfer heat about 23 times faster."
To conceptualize this, McLaughlin draws a parallel to mechanical engineering:
"Think of a hot engine: a fan can cool its surface, but circulating coolant through the engine removes heat directly from where it is generated. The same principle applies to high-density compute. At today’s densities, that difference is decisive. Air cooling depends on fans and chillers pushing enormous volumes of air across hot components, and that movement takes significant power."
By bringing liquid significantly closer to the heat source—whether via a facility loop or a specialized dielectric fluid circulating through a direct-to-chip technical loop—operators can capture heat directly at the point of origin. This allows heat to be rejected at higher temperatures with substantially less auxiliary energy spent moving air through massive server halls.
The Water-Energy Nexus: Unveiling the Hidden Footprint
A critical oversight in legacy data centre planning has been the artificial separation of water metrics from energy metrics. For years, operators tracked Power Usage Effectiveness (PUE) as the primary indicator of efficiency, treating Water Usage Effectiveness (WUE) as a secondary, localized concern. McLaughlin warns that this isolated optimization creates a dangerous blind spot.
"It is tempting to frame cooling as a choice between competing technologies," McLaughlin notes. "Where operators can go wrong is optimising one metric in isolation. An approach that uses less water at the point of cooling may need more energy, and generating that energy can consume more water upstream at the power plant."
This realization brings light to the complex "water-energy nexus." Much of the world’s baseload electricity relies on water somewhere along its supply chain—from cooling thermal power generation plants (coal, nuclear, natural gas) to feeding hydroelectric turbines. Consequently, every kilowatt-hour saved by an efficient on-site cooling architecture also eliminates the embedded water consumed upstream to generate that power.
"This part of the equation often goes uncounted because it happens outside the data centre’s fence," McLaughlin states. "But it’s real, and it means efficient cooling improves the water math in two places at once. On site, water is used purposefully and kept in circulation. Off site, lower power demand reduces the water embedded in the grid that feeds the facility."
To scale responsibly, data centre operators must move past linear models—where fresh resources flow in and waste heat and water flow out—toward resilient circular systems centered on recycling, reuse, and closed-loop liquid architectures.
Designing the Modern Cooling Chain: A Hybrid Approach
Despite the undeniable advantages of liquid cooling, McLaughlin emphasizes that the industry is not facing a binary choice between air and liquid. High-density hyperscale and colocation sites operate via a sophisticated, multi-tiered cooling chain.
Heat naturally flows through a sequential pathway:
- From the chip to the immediate cold plate or immersion fluid.
- From the rack via local distribution manifolds and heat exchangers.
- From the facility to the external atmosphere.
Because each stage of this journey presents unique engineering requirements, state-of-the-art facilities leverage a hybrid mix of cooling technologies rather than relying on a single silver bullet. Direct-to-chip liquid cooling handles the blistering thermal loads of heavy AI accelerators, while traditional air cooling or localized evaporative systems manage peripheral rack and room loads.
At the facility level, heat rejection is typically handled through:
- Evaporative cooling towers
- Dry coolers
- Adiabatic systems
- Hybrid configurations
The selection of these systems is dictated entirely by three governing site variables: local climate, local power grid carbon/capacity profiles, and local water availability (such as operating within water-stressed watersheds).
"The operators leading the way are designing for local conditions rather than copying a template," McLaughlin observes. "Evaporative cooling remains one of the most energy-efficient ways to reject heat, but reusing water or tapping into non-potable, recycled municipal sources is what makes it truly sustainable."
Precision Infrastructure: Why Chemical Health Dictates Uptime
Deploying advanced water-based or liquid-cooled architectures introduces a new operational imperative: the fluid itself must be treated as critical, mission-critical infrastructure.
In closed-loop liquid systems, efficiency and hardware safety depend entirely on maintaining fluid integrity. Without active, rigorous management, variables such as corrosion, microbiological growth, suspended particulates, water chemistry fluctuations, and coolant drift can severely degrade heat-transfer efficiency over time, leading to catastrophic hardware failures.
"Even in direct-to-chip liquid cooling applications, water and coolant fluids are not ‘set-and-forget’ utilities," McLaughlin stresses. "Real-time monitoring of coolant chemistry, temperature, and flow helps operators identify out-of-spec conditions and manage risks proactively, so they can protect servers and mitigate risks."
Real-World Risk Mitigation
To illustrate the vital importance of real-time fluid telemetry, McLaughlin points to recent field deployments:
- The Hyperscale Case Study: At a major hyperscale facility, continuous real-time monitoring detected microscopic corrosive particles circulating within a secondary loop—contaminants that standard, periodic manual maintenance checks had completely missed. Early detection allowed engineering teams to remediate the issue before pitting or seal degradation could occur.
- The Colocation Commissioning Case Study: During the commissioning phase of a high-density colocation build, automated sensor arrays flagged diluted coolant concentrations. Immediate intervention prevented thermal inefficiencies and kept the facility’s go-live timeline strictly on schedule.
In both instances, proactive chemical and fluid management eliminated the precursors to unscheduled downtime, preserving both hardware integrity and enterprise peace of mind.
Strategic Playbook: What This Means for Data Centre Operators
As the industry looks ahead toward accelerating AI deployments, data centre planners, architects, and facility managers must adapt their methodologies. McLaughlin outlines four core principles to guide successful high-density builds and retrofits:
- Start with the Site: Let local environmental realities—climate, grid profile, watershed stress, and workload distribution—shape your cooling design. Never force a standardized template onto a mismatched regional environment.
- Unify WUE and PUE Models: Model water and energy efficiency simultaneously. Always factor in the indirect, upstream water embedded in the electrical power consumed by your facility.
- Match the Method to the Heat: Deploy direct-to-chip or immersion liquid cooling where thermal density explicitly demands it, and couple it with facility-level heat rejection systems optimized for local climate dynamics.
- Treat Fluid as Mission-Critical Infrastructure: Implement real-time monitoring platforms from the initial commissioning phase onward. Maintaining strict control over fluid chemistry ensures that efficiency gains hold stable over the entire lifecycle of the equipment.
Future Outlook: Sustainable Compute at Scale
The relentless march of artificial intelligence will inevitably drive continuous demand for higher compute density, greater electrical power, and advanced cooling capabilities. However, industry leaders are increasingly aware that digital growth cannot come at the expense of local communities and natural ecosystems.
While electrical capacity can theoretically be expanded through new generation assets and grid investments, water availability remains strictly bound by local hydrological constraints. Unregulated resource consumption carries severe human and environmental consequences that modern enterprises can no longer ignore.
"Operators that use water and liquid cooling effectively—and manage water and energy as one connected system—will be better positioned to deliver more compute with fewer resource demands," McLaughlin concludes.
Ultimately, the future of the data centre industry belongs to those who recognize that sustainable digital infrastructure is not built in isolation, but in harmony with the physical resources that support it.
