Data Center Liquid Cooling Market Size & Share Analysis - Growth Trends and Forecast (2027 - 2032)

Market Trends and Insights

Surging rack densities (>30 kW) in AI and HPC facilities

High-density computing concentrates the cooling opportunity in the racks placing the greatest demands on facility infrastructure. Uptime Institute’s 2026 global survey found more operators reporting peak rack densities of 30 kW or higher for advanced computing workloads.^[1] It linked rising rack power demands to AI and other enterprise applications, while average modal densities continued their slow upward movement. This combination makes a facility-wide density average a poor guide to the cooling requirements of its most demanding computing installations. Peak rack heat loads can justify targeted liquid-cooling investment even when the facility’s average density does not call for conversion.

Hyperscale operators' net-zero roadmaps accelerating liquid adoption

Microsoft’s 2030 carbon-negative and water-positive commitments place cooling design within a broader hyperscale environmental roadmap. Microsoft reported that about 90% of its owned data-center fleet operated with low- to zero-water cooling systems in 2025. It described its liquid-cooled AI data centers as using closed-loop, direct-to-chip cooling for chip-level temperature control without water evaporation. That operating characteristic serves the water objective through the cooling loop itself, connecting an environmental commitment with a specific thermal architecture. For Microsoft’s liquid-cooled AI facilities, adding chip-cooling capacity can support computing demand without adding evaporative losses from that cooling operation.

OEM warranties now covering direct-to-chip loops

Warranty-backed direct-to-chip cooling can make responsibility for the cooled server a more explicit part of the deployment decision. In March 2025, Accelsius reported offering server-level warranties globally for leading OEM systems alongside its two-phase direct-to-chip cooling support program. It also said SKU-level warranties were being introduced after testing and validation by OEM providers. Linking coverage to server-level and SKU-level validation gives buyers a more specific basis for judging the protected configuration than a cooling-system claim alone. Validated server coverage can reduce the buyer’s exposure to a gap between thermal-system support and protection of the computing hardware.

Nvidia and AMD liquid-ready reference designs driving ecosystem

Cooling suppliers gain a more explicit integration target when processor platforms specify rack infrastructure alongside compute. In March 2026, NVIDIA announced its Vera Rubin DSX AI Factory reference design, including partner guidance for power, cooling and control systems across the broader infrastructure stack. AMD showcased its Helios rack-scale AI reference design with quick-disconnect liquid cooling in October 2025 for adoption and customization by OEMs, ODMs and hyperscalers. In July 2026, AMD reported that Helios solutions were in production for deployment by leading AI companies at gigawatt scale. Embedding liquid connections in an adoptable rack design makes cooling part of platform replication, so each customized build carries a defined thermal integration task.

Limited field expertise among facility engineers

Liquid-cooling expertise creates a deployment constraint within the broader data-center workforce. A 2025 peer-reviewed perspective by Intel Corporation and Intel Labs China researchers identified a lack of skilled workers as a challenge hindering liquid-cooling technologies. It said installation, operation, maintenance and troubleshooting generally require specialized knowledge and skills from operators. Across data-center positions generally, Uptime Institute’s 2026 survey found that more than half of respondents had difficulty finding qualified candidates. An operator relying on a general facilities workforce may carry that skills gap into commissioning and maintenance, limiting how much new cooling equipment it can put into dependable service.

High upfront retrofit costs for brown-field sites

For brownfield operators, the capital burden of liquid cooling can extend from the equipment package to the existing facility. The 2026 IEA 4E report describes complete overhauls at existing data centers as often impractical and very expensive, prompting less disruptive upgrades. One potential retrofit replaces a row of air-cooled racks with a single high-density AI rack served by a liquid-to-air coolant distribution unit. The AI rack and distribution unit may be substantially heavier than conventional equipment, so load spreaders may be required to avoid excessive floor loading. Even this simplified route may add structural work such as load spreaders, so the retrofit cost can extend beyond the cooling hardware itself.

*Our forecasts treat driver/restraint impacts as directional, not additive. The impact forecasts reflect baseline growth, mix effects, and variable interactions.

Data Center Liquid Cooling Market Segment Analysis

Direct-to-chip liquid cooling accounted for a 42.79% share of the data center liquid cooling market size in 2026, making it the largest segment. In February 2026, CoolIT Systems Inc., acquired by Ecolab in July 2026, reported that its coldplates cooled more than 5 million GPUs, CPUs and AI accelerators worldwide. It also reported use of its direct liquid-cooling technology in 7 of the top 10 supercomputers, including El Capitan. Those footprints connect chip-level cooling with both a large processor population and demanding computing installations. A coldplate design’s commercial reach can multiply across processor positions within each system, linking component supply to computing deployment scale.

Immersion cooling is forecast to be the fastest-growing segment, at a CAGR of 26.12% from 2027 to 2032. Shell reported in 2026 that its own immersion-cooling fluids were in use at its data centers in Houston and Amsterdam. Rear-door heat exchangers offered NorthC another route at Münchenstein (Basel) 1 in Switzerland, where Legrand’s USystems equipment was installed without interrupting operations by November 2025. Shell develops its immersion fluids in its own laboratories, combining the fluid-developer and data-center-operator roles in this example. Combining those roles can connect fluid development directly with operating requirements, giving immersion adoption a path within the same organization.

Single-phase hydrocarbon fluids represented a 45.08% data center liquid cooling market share in 2026, making them the largest segment. Intel certified Shell Lubricants’ immersion fluids in 2025 for compatibility with 4th and 5th Gen Intel Xeon processors, the first fluid line to earn its Data Center Certification for Immersion Cooling. Shell’s single-phase S3 X offering uses a GTL-based hydrocarbon fluid. Such processor-specific validation gives fluid selection a defined relationship with the computing platform. Its commercial value lies in qualifying the fluid for an identified hardware environment, rather than relying on the broad hydrocarbon category as a sufficient specification.

Two-phase fluorocarbon fluids are forecast to be the fastest-growing segment, at a CAGR of 25.14% from 2027 to 2032. Among the remaining coolants, water/glycol solutions have an operating use in some Microsoft data centers, where a propylene-glycol solution serves closed-loop cooling systems. In May 2025, Chemours agreed with Navin Fluorine International Limited to manufacture Opteon two-phase immersion fluid, with initial manufacturing capacity planned for 2026. In August 2026, Navin Fluorine’s investor presentation targeted completion of the Chemours project by the end of September 2026. Navin Fluorine's project schedule therefore sets when contract-manufactured Opteon capacity can begin.

Hyperscale facilities held the largest segment share, at 39.17% of market revenue in 2026. Google reported in April 2025 that it had deployed liquid cooling at gigawatt scale across more than 2,000 TPU Pods over the preceding 7 years. This footprint gives cooling suppliers a computing environment extending across many deployment units, creating a broader implementation task than a single isolated equipment installation. For suppliers, one operator relationship can connect equipment sales to many pod installations over a multiyear deployment program.

Edge and micro data centers are forecast to be the fastest-growing segment, at a CAGR of 24.48% from 2027 to 2032. Cogent reported 84 edge data centers in its company-wide on-net building inventory at the end of Q2 2026. That inventory illustrates a multi-site operating footprint, making the engineering effort repeated at each location a potential cost multiplier. For an edge operator adopting liquid cooling, reusable site-level installation designs could spread that engineering effort across multiple locations.

Artificial intelligence and machine learning were the largest segment with a 35.79% share in 2026 and are forecast to be the fastest-growing segment, at a CAGR of 25.48% from 2027 to 2032. Oracle made liquid-cooled NVIDIA GB200 NVL72 systems generally available on OCI Supercluster in June 2025, scalable to a maximum of 131,072 NVIDIA GPUs. The offering connects liquid cooling with commercial AI training and inference capacity, placing the thermal arrangement within a cloud service. This gives the cooling architecture a role in the service’s capacity proposition, with its value tied to the size of the computing configuration the operator can offer.

Cloud and virtualization make the provider’s deployment stack an important context for cooling implementation. NVIDIA reported in April 2025 that Oracle had stood up and optimized its first wave of liquid-cooled GB200 NVL72 racks in its data centers. The deployment included high-speed Quantum-2 InfiniBand and Spectrum-X Ethernet networking for scalable, low-latency performance, alongside NVIDIA and OCI software and database integrations. Because the racks serve customers through NVIDIA DGX Cloud and OCI, the cooling reaches the end user as part of a rented computing service.

Geography Analysis

North America Data Center Liquid Cooling Market

North America accounted for 36.97% of the data center liquid cooling market size in 2026, the largest regional share. In its September 2025 announcement, Microsoft said closed-loop liquid cooling served more than 90% of capacity at its first Fairwater facility in Mount Pleasant, Wisconsin, United States.^[2] It confirmed the first Mount Pleasant facility was fully operational in June 2026, while its second facility was under construction. For cooling vendors, the high liquid-cooling coverage at this AI facility connects regional opportunity to the computing installation as a whole, extending its scope beyond individual chip-level components.

APAC Data Center Liquid Cooling Market

APAC is forecast to be the fastest-growing region, at a CAGR of 27.17% from 2027 to 2032. In May 2026, Equinix announced KL2 in Cyberjaya, Malaysia, with more than 2,200 cabinets at full build-out and a substantial portion of capacity supporting liquid cooling for AI and HPC. AirTrunk’s planned JHB2 campus in Johor Bahru, Malaysia, was designed for more than 270 MW of capacity and a design PUE of 1.37 using liquid cooling. AirTrunk announced development financing in July 2026, and IFC recorded financing investment in August 2026. Both campuses specify liquid cooling in their designs, and KL2 had reached structural construction by July 2026, so the question for suppliers is when each site reaches equipment installation.

Outside the leading regions, cooling projects connect high-density readiness with different stages of physical implementation. In South America, ODATA, an Aligned Data Centers company, announced Aligned’s DeltaFlow, developed with Munters, at SP04 in Brazil in September 2026, with installation and first-phase testing under way. Europe’s SINES DC in Portugal brought its first building into operation in 2025, coupling closed-loop freshwater process cooling with seawater heat rejection. Within Middle East and Africa, Teraco, majority-owned by Digital Realty, completed its JB4 expansion in Ekurhuleni, South Africa, in 2025, enabling clients to deploy direct-to-chip cooling for denser AI workloads. Halls ready for clients to deploy cooling can create a later opportunity for equipment sales than an installation already under way.

Regulatory Framework and Outlook

This comparison of selected jurisdictions assesses regulatory conditions for data-center operators deploying liquid cooling. Attractiveness reflects incentive access and the burden of the reviewed performance, reporting and power-connection frameworks. Germany requires covered data centers with at least 300 kW of non-redundant connection capacity to publish and transmit prior-year Annex 3 information by March 31, excluding predominantly non-processing network nodes. Brazil’s Redata regime, in force from September 15, 2026, links eligibility to an annual cooling WUE commitment at or below 0.05 L/kWh.

Regional tariff review frames the US Medium rating, while interconnection rulemaking at a proposed stage with an undetermined November 2026 next action supports near-term No change. Brazil’s High rating reflects conditional relief for qualifying operators, while the December 31, 2026, endpoint for specified tax benefits supports Hardening. Germany's reporting burden supports Medium, while the European Commission's minimum-performance proposal, planned for the second quarter of 2027, points to Hardening through potential additional operating requirements. China’s Medium rating reflects the current standard’s air-cooling scope, while the proposed extension to eligible liquid-cooled facilities, with implementation proposed 12 months after publication, points to Hardening.

Value Chain Stages

The value chain compares system engineering and equipment supply with deployment and maintenance, following how suppliers can earn across an installation’s life. CoolIT Systems Inc., acquired by Ecolab in July 2026, engineers and mass-produces direct liquid-cooling systems for AI and HPC data centers. Rittal GmbH and Co. KG supplies hyperscale cooling equipment, including rear-door heat exchangers and coolant distribution units. CoolIT also provides professional services to design, deploy and maintain liquid-cooled systems, linking its manufacturing activity to implementation and after-sales support.

Integrated suppliers can address the handoffs between equipment, deployment and maintenance within a single service relationship. Their additional earning opportunity depends on converting hardware knowledge into separately valuable implementation and maintenance work. For equipment specialists, clearly defined interfaces preserve a role alongside integrated providers without requiring participation in every service stage.

Key Opportunities for Market Participants

The opportunities compare cooling-equipment packages, existing-site upgrades and integration work for liquid-cooled computing projects. Construction of the Pohang AI Data Center began in June 2026, with operations targeted for the second half of 2027. AI Factory Pohang PFV commissioned the project, with Hyundai E&C as construction contractor and liquid cooling planned for its AI servers. In its August 2026 update, IREN planned liquid-cooled deployments during 2027 at Mackenzie, Canal Flats and Prince George in British Columbia to use existing power headroom.^[3]

The IEA 4E report identified standardization, high initial costs and long-term reliability concerns as sector-wide adoption barriers in 2026. IREN’s plan to use existing power headroom makes its opportunity commercially different from an entirely new AI facility, because cooling additions can be tied to power already available at named sites. That creates a focused entry point for suppliers able to fit incremental cooling equipment to the operator’s deployment plan.