UL Solutions has launched a new certification program for liquid cooling subassemblies designed to cool high-performance chips directly, targeting one of the fastest-changing parts of AI data center infrastructure. The program focuses on components used in direct-to-chip liquid cooling systems, where coolant travels through a cold plate attached to the processor and pulls heat away from the chip. As AI workloads push computing equipment toward higher power densities, operators increasingly need cooling architectures that can handle heat loads beyond the practical limits of conventional air cooling. UL Solutions is positioning its certification as a way for equipment manufacturers and data center operators to bring more confidence to that increasingly complex hardware layer.
AI systems generate substantial heat because their processors run at high power while handling demanding workloads. When cooling cannot keep pace, equipment can throttle performance, shut down, or suffer damage, creating an infrastructure problem that reaches beyond simple thermal management. Direct-to-chip liquid cooling addresses the challenge by placing the cooling mechanism much closer to the heat source, with coolant absorbing thermal energy through a cold plate and carrying it away through the cooling loop. That architecture has become increasingly relevant as AI accelerators and other high-performance computing platforms demand greater rack-level power and more sophisticated thermal control. “As AI data centers handle more powerful computing, they depend on even more advanced cooling systems to operate safely and reliably,” said Jennifer Scanlon, president and CEO of UL Solutions. “Our certification helps IT equipment manufacturers demonstrate product performance and safety while giving data center operators greater confidence during integration.”
Liquid Cooling Creates A New Integration Problem
The shift toward liquid cooling changes the risk profile inside an AI data center because operators must manage fluid, pressure, materials and mechanical connections alongside compute hardware. A cooling loop can introduce risks that traditional air-based systems largely avoid, including leaks, corrosion, contamination and compatibility failures between components. Those issues can affect equipment reliability while also undermining the thermal performance that operators expect from a liquid-cooled deployment. For AI infrastructure buyers, the cooling system therefore becomes an engineering dependency that deserves the same scrutiny as servers, networking equipment and power systems.
The challenge becomes more pronounced when data center operators assemble cooling infrastructure from multiple suppliers. Components can physically connect while still requiring different coolant specifications, pressure ranges or operating conditions, creating potential integration problems that may not become obvious during procurement. Industry research has also pointed to limited standards and integration challenges as obstacles to broader liquid cooling adoption. UL Solutions is addressing that gap by evaluating eligible components and subassemblies before they become part of a larger cooling architecture.
UL 4501 Brings Testing To Cooling Components
The new UL Solutions program covers eligible liquid-filled components and subassemblies that move coolant through information and communication technology equipment. The scope includes several critical parts of a direct-to-chip cooling architecture, including cold plates that capture heat from processors, manifolds that distribute coolant across servers and racks, and quick disconnects that allow equipment servicing without draining an entire cooling loop. These components may appear relatively small compared with GPUs or servers, but their performance can determine whether a liquid cooling system operates reliably at scale. Certification therefore puts greater attention on the interfaces that connect compute equipment to its thermal infrastructure.
UL Solutions evaluates eligible liquid cooling subassemblies under UL 4501, its Outline of Investigation for Safety, Performance and Interoperability of Liquid-Filled Components and Assemblies Used in Liquid Cooling Systems. The evaluation examines coolant compatibility, pressure and leak integrity, durability, liquid flow, interoperability, production controls and installation requirements. That combination matters because liquid cooling performance depends on more than whether a component can move fluid from one point to another. A component also needs to remain compatible with the operating environment over time while maintaining the mechanical and fluid characteristics required by the broader system.
Certification Could Shift How Operators Buy Cooling
Testing individual components before they enter a complete cooling system can give manufacturers and operators an earlier view of potential failures. Instead of discovering incompatibilities only after multiple components come together inside a finished cooling architecture, organizations can use independent evaluation to establish whether eligible products meet applicable requirements earlier in the process. This approach can reduce repeated testing and help manufacturers substantiate product claims when presenting cooling equipment to prospective customers. For data center operators, the benefit comes through more informed decisions around procurement, integration and system-level deployment.
The move also reflects a broader change in AI infrastructure procurement, where thermal performance increasingly intersects with reliability and operational risk. Cooling equipment no longer sits quietly behind the server as a supporting utility; it directly influences how much compute infrastructure can operate within a rack and how reliably that infrastructure can sustain its workload. As operators combine equipment from multiple vendors, standardized evaluation can help create a clearer baseline for comparing components before they enter increasingly dense AI environments. The strategic value of certification, therefore, extends beyond a compliance exercise into the practical problem of scaling liquid-cooled compute without multiplying integration risk.
UL Solutions Pushes Independent Validation
“AI infrastructure must scale without sacrificing safety or reliability,” said Milan Dotlich, vice president and general manager of Data Center Solutions at UL Solutions. “Independent evaluation of liquid cooling components can help organizations deploy advanced computing systems with greater confidence.” UL Solutions’ program arrives as AI infrastructure moves toward higher-density computing and greater dependence on advanced thermal architectures. The company’s focus on component-level evaluation highlights an important reality for the next generation of data centers: the reliability of an AI rack depends on a chain of interconnected systems, and cooling is becoming one of the most technically sensitive links. Direct-to-chip liquid cooling can provide a path toward managing higher heat loads, but its success depends on components working together across fluid, mechanical and thermal boundaries.



