AI Infrastructure at Risk: Critical Flaws in Data Center Cooling Systems Threaten Industry Growth, New Research Warns

Technical deficiencies - poor direct-to-chip liquid cooling systems, misuse of glycol, and flawed industry guidance – may put billions of AI investments at risk

HOUSTON, TX, UNITED STATES, February 9, 2026 /EINPresswire.com/ -- As artificial intelligence workloads push data center rack densities to unprecedented levels, a comprehensive peer-reviewed technical paper reveals alarming deficiencies in current liquid cooling system design and implementation that could jeopardize the reliability of critical AI infrastructure, as well as the lifecycle of the chips and hardware that fuel AI’s performance.

The research, authored by cooling systems expert Loraine Huchler, P.E., CMC^® of MarTech Systems, Inc., exposes fundamental flaws in published industry guidance for direct-to-chip (DTC) cooling systems – the technology increasingly relied upon to manage heat from high-performance Graphics Processing Units (GPUs) powering AI applications.

The Stakes Are Higher Than Ever
Data Center Frontier conducted a 2025 annual benchmarking survey (N=172) across a wide variety of organizations that showed that 48% of the respondents were currently using for direct-to-chip cooling systems. These findings carry immediate implications for billions of dollars in AI infrastructure investments. The paper reveals that current industry standards contain “inaccurate information” about crucial system components including chemical treatment, materials selection, fluid specifications, filtration, and – perhaps most importantly – the choice between water and glycol cooling fluids.

The Glycol Misconception Putting Systems at Risk
Among the most concerning findings: a widespread misunderstanding about glycol solutions in cooling systems is leading operators to make choices that compromise both data center performance and energy sustainability.
“Some operators assert that inhibited glycol solutions have higher stability and greater reliability compared to water,” Huchler notes, “but the data tells a different story.” The research reveals multiple problems with glycol use in DTC cooling:

• Massive Performance Penalties: Comparing water to glycol systems, glycol requires 15-20% higher flowrate requirements and 25-30% higher pumping power due to glycol’s heat transfer properties at typical operating conditions. Glycol has a measurable increase in energy consumption precisely when the industry is focused on efficiency.
• Chemical Degradation Risk: Unlike water, glycol is vulnerable to chemical degradation from continuous operating temperatures, creating organic acids that can damage systems over time.
• False Security on Bacteria: “An engineer from a chip manufacturer adamantly—and incorrectly—insisted that 25% propylene glycol would kill bacteria and therefore there would be no risk of fouling,” Huchler reports. In reality, glycol concentrations above 23% only provide biostatic protection that slow bacterial growth but don’t eliminate the threat.
• Dye Interference: Many glycol products contain dye for leak detection, but this prevents real-time fluid quality testing using colorimetric and spectrophotometric methods. “Operators need to balance the risk of leaks versus the ability to have real-time fluid quality data to ensure efficient heat transfer,” the paper warns.
• Unnecessary for Operating Racks: Except for shipping, there’s no need for freeze protection in operating racks, eliminating one of the considerations for glycol use.
• Sustainability Concerns: In addition to increased energy consumption, water provides higher sustainability compared to glycol’s manufacturing process and additional disposal considerations during leaks and operational changeouts.

Crucial Vulnerabilities Beyond Glycol
The research identifies several additional high-risk areas where current practices fall short:

• Microchannel Fouling Risk: Cold plates microchannels are approximately 100 microns wide—yet current filtration guidance varies wildly and often contradicts fundamental engineering principles, creating significant fouling risks that could cripple cooling performance.
• Incompatible Testing Standards: Industry documents reference ASTM qualification tests that have “no relevance” to actual operating conditions, using wrong materials, wrong temperatures, and wrong fluid conditions—essentially validating fluids under conditions that don’t match real-world deployment.
• Material Compatibility Issues: Widely used oxygen-permeable plastic tubing can introduce corrosion risks that degrade heat transfer efficiency, yet published guidance fails to adequately address material selection criteria for recirculating cooling systems that have a closed design.
• Fluid Dynamics Confusion: Published sources provide “confounding explanations” of flow requirements, with some claiming transition or turbulent flow is beneficial for the wrong reasons—potentially selecting flowrates that create “hot spots” damaging expensive processors.

Misinformation Compounding the Problem
The paper documents troubling examples of technical misinformation circulating in the industry. Beyond the glycol-bacteria misconception, Huchler reports an engineer insisting that algae was causing fouling in laboratory cold plates—despite algae being a plant requiring light to survive, making it virtually impossible in closed cooling systems.
“Unfortunately, inaccurate technical information and unverified claims continue to create risks for data center owners and operators,” the paper states.

“Cooling Is King” … But Current Approaches May Be Failing
“The performance of chip cooling systems has not matched the increased power consumption and heat rejection needed to ensure reliable performance,” warns Huchler. “We’re seeing increased adoption of liquid cooling, but without rigorous technical standards, the industry is building on a shaky foundation.”

The paper notes that inadequate fluid monitoring and treatment programs create particular risk. Even minor particulate contamination or calcium scale formation on cold plate surfaces can incrementally reduce heat transfer efficiency—yet many facilities lack the inline sensors and testing protocols necessary to detect – and remediate - problems before they cause failures.

The Business Continuity Threat
For data center operators supporting AI workloads, cooling system failures pose direct threats to business continuity. As Huchler notes, “cooling has become a central issue for business continuity and capital planning” as computing speeds and heat rejection continue to increase.

The research emphasizes that water treatment suppliers and operations staff must adopt more rigorous procedures, stating: “The best water treatment design, poorly managed, is less effective than a good water treatment design, properly managed.”

Call for Industry Action
The paper calls for fundamental reforms by the AI and data center industry “Suppliers, water treatment professionals, and consultants must serve the data center community by creating validated technical information to optimize heat transfer efficiency and system reliability,” Huchler concludes.

The full peer-reviewed technical paper published by The Cooling Technology Institute (www.cti.org), “TP26-06 Optimizing Lost Heat Transfer in Single-Phase Cold-Plate Liquid Cooling Systems,” provides detailed analysis of cooling system design, fluid selection, materials compatibility, filtration strategies, and monitoring requirements for direct-to-chip, single-phase liquid cooling in data centers.

Mark Corallo
Corallo Communications
+1 703-402-2249
mark@corallocommunications.com

Legal Disclaimer:

EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.