Artificial intelligence infrastructure is undergoing a thermal revolution that's reshaping how entire data center campuses are designed and built. As AI workloads demand denser GPU clusters and higher power concentrations, liquid cooling is shifting from a nice-to-have facility feature into a core strategic decision that affects compute performance, energy efficiency, water usage, and deployment speed. The data center liquid cooling market is estimated at USD 3.39 billion in 2025 and is projected to reach USD 23.24 billion by 2035, growing at a compound annual rate of 21.21% from 2026 to 2035.

Why Is Cooling Becoming the Bottleneck for AI Infrastructure?

The thermal challenge is straightforward but urgent: modern AI racks are generating heat at levels that traditional air cooling simply cannot handle. Rack power densities have climbed from manageable levels to 30 kilowatts to 100 kilowatts or more, making conventional air-based systems impractical for contemporary AI workloads. NVIDIA's reference architecture for Rubin systems shows rack power approaching 225 kilowatts, while power suppliers are already designing systems for 1 megawatt racks and larger configurations. At campus scale, thousands of racks operating simultaneously create a concentrated heat rejection requirement that must be coordinated with electrical substations, backup power systems, and water infrastructure.

North America currently holds about 35.6% of the global liquid cooling market, with direct-to-chip cooling representing about 42.9% of technology demand. This concentration reflects the region's dominance in hyperscale AI infrastructure development, but the thermal challenge is becoming universal as AI deployment accelerates worldwide.

What Cooling Technologies Are Competing for Market Share?

The competitive landscape spans multiple cooling approaches, each suited to different deployment scenarios and facility constraints. The market is not consolidating around a single winner; instead, operators are selecting technologies based on their specific infrastructure stage and operational requirements.

• Direct-to-Chip Cooling: Removes heat directly from the processor while preserving familiar rack layouts and service practices. Research published in 2026 showed that optimized cold plate channel designs reduced maximum chip temperature by more than 35 degrees Celsius compared with baseline designs, demonstrating how material engineering and fluid path optimization can improve thermal performance.
• Rear Door Heat Exchangers: Can extend the useful life of existing air-cooled facilities without requiring complete infrastructure redesign. These systems remain important for operators managing mixed environments with legacy servers and new GPU racks.
• Single-Phase Immersion Cooling: Surrounds electronic components with dielectric fluid to reduce fan energy and enable compact server configurations. Adoption depends on server warranties, maintenance workflows, fluid compatibility, and industry standardization efforts.
• Spray and Jet Cooling: Emerging approaches for localized hot spot management and future chip packages as heat flux at the chip level continues to increase.

Solutions account for about 74.5% of market revenue, reflecting strong demand for integrated systems rather than individual components. Hyperscale facilities represent about 38.9% of revenue, while AI and machine learning workloads account for about 34.8% of demand. Less than 30% of global data centers currently use liquid cooling, which leaves substantial conversion potential as new server platforms make liquid readiness a standard procurement requirement.

How Can Operators Optimize Cooling Efficiency at Scale?

Beyond hardware selection, software and control systems are becoming critical to cooling economics. A 2026 study based on the Frontier exascale supercomputer found that coordinated optimization of flow rate and supply temperature could deliver energy savings of about 27.8% under practical operating constraints. This finding suggests that the future cooling market will include intelligent software platforms that optimize thermal systems across changing AI workloads, rather than relying on static operating settings.

• Digital Twin Technology: Creates virtual models of cooling systems to predict performance and identify efficiency opportunities before physical deployment.
• Adaptive Control Systems: Monitor temperature, pressure, flow, and coolant quality in real time across thousands of nodes, adjusting operations to match current workload demands.
• Heat Recovery and Reuse: Captures waste heat from cooling systems for district heating, industrial processes, or heat pump applications, improving site economics and sustainability metrics.
• Continuous Monitoring Platforms: Track coolant quality, detect leaks early, and maintain consistent reliability across gigawatt-scale campuses with thousands of interconnected cooling nodes.

How Does Regional Geography Shape Cooling Investment Decisions?

Cooling technology selection is increasingly shaped by local constraints and regulatory environments. In the United States, data centers are expected to account for about half of electricity demand growth through 2030, according to the International Energy Agency. Large campuses are being planned in regions with grid access, land, and gas infrastructure, yet water availability and interconnection delays are influencing facility design decisions. Closed-loop liquid systems, dry coolers, and warm water operation can reduce water dependence and support faster permitting in constrained markets.

Europe is placing greater emphasis on energy efficiency, heat recovery, and urban integration. Liquid cooling can produce higher quality waste heat than conventional air systems, improving the economics of district heating and industrial heat reuse where local networks exist. The opportunity extends to heat pumps, thermal storage, and digital control platforms that match cooling output with external heat demand. Operators that can monetize recovered heat may improve both sustainability performance and site economics.

Global data center electricity consumption is projected by the International Energy Agency to reach about 945 terawatt-hours by 2030, more than double the 2024 level. Cooling and other infrastructure account for a meaningful share of this increase. As AI-focused capacity expands, thermal efficiency will directly influence how much revenue a campus can generate from a fixed power connection, fundamentally changing cooling economics.

What Political and Community Headwinds Are Slowing Data Center Expansion?

While cooling technology advances are accelerating, community opposition and regulatory scrutiny are emerging as material constraints on data center development, even in markets that have actively courted the industry. In Texas, Governor Greg Abbott issued an order on August 3 pausing all new data center connections to the Electric Reliability Council of Texas (ERCOT) grid until a comprehensive audit is complete. This pause could delay 49.8 gigawatts of new data center electricity demand, nearly 20% of the U.S. development pipeline. BloombergNEF estimates that delays could cost projects between USD 8 billion and USD 15 billion cumulatively, depending on the proportion of AI-related capacity affected.

The Texas audit targets roughly 474 gigawatts of pending large-load requests in the interconnection queue, about 90% of them data centers. Abbott cited grid reliability concerns, noting that the queue is "more than five times Texas' record peak electricity demand," and pointed to compliance failures where some data centers did not fully respond to the Public Utility Commission's survey measuring water and power usage. However, BloombergNEF suggests the timing is politically motivated, noting that Abbott is facing reelection in November 2026 and the pause is "likely intended to take the controversial data center issue off the table until after the voters have their say".

Community opposition is becoming increasingly visible and confrontational. In Salem, Oregon, a city council meeting on July 27 to discuss a proposed USD 5.1 billion Oakline at Mill Creek data center by Verrus drew protesters who brought a guillotine to the venue, signaling strong public discontent. The consultation lasted more than six hours, and Verrus representatives were confronted by opposition members and had to leave midway through the proceedings for safety reasons. One week after the meeting, Salem announced a data center moratorium that would halt all data center developments in the city for one year as it considers zoning laws to accommodate such projects.

Salem joins a growing list of jurisdictions that have passed similar moratoriums, including Seattle, Washington, and the State of New York. Governor Tina Kotek announced that Oregon will terminate a deal to sell a 32-acre plot of land that formed part of the three plots on which the Verrus project was planned, though Verrus still retains approximately 43 acres. Despite the moratorium, Verrus's application remains under processing because it was filed three days before the moratorium announcement, though the project still must pass an extensive permitting process including technical studies, permit conditions, and enforceable agreements.

The regulatory landscape is shifting rapidly across multiple states. New York Governor Kathy Hochul signed Executive Order 62 on July 14, establishing the nation's first statewide moratorium on new hyperscale data centers of 50 megawatts or more. Illinois Governor JB Pritzker directed the state's Department of Commerce and Economic Opportunity to pause processing of Data Center Investment Program agreements starting July 1. Florida Governor Ron DeSantis signed Senate Bill 484, which became effective on July 1, defining large-scale data centers as facilities with anticipated monthly peak load of 50 megawatts or more and prohibiting utilities from shifting service costs to residential and small-business ratepayers.

The cooling technology market is expanding rapidly, but the political and community environment for data center deployment is tightening. Operators investing in advanced liquid cooling systems must now navigate not only technical and infrastructure challenges but also increasingly restrictive regulatory frameworks and organized community opposition. The USD 23 billion cooling market opportunity exists, but realizing it will require addressing the social license to operate alongside the thermal engineering challenges.

" }