Report: AI Driving 28.7% CAGR in Liquid Cooling Market
A new market analysis forecasts that the data center liquid cooling market will grow at a compound annual rate of 28.7%. The growth is primarily driven by escalating thermal loads from high-density GPU deployments for AI workloads. The trend reflects a structural transformation in data center design toward liquid-first cooling solutions to meet performance and sustainability targets.
- The power consumption of individual GPUs is a primary driver for this shift, with NVIDIA's H100 GPU drawing 700W and the newer B200 GPU requiring 1000W to 1200W. This rising power demand makes traditional air cooling increasingly impractical for dense AI computing clusters. - There are two main liquid cooling methods: direct-to-chip (DTC) and immersion cooling. DTC uses cold plates to cool specific components like CPUs and GPUs, while immersion involves submerging entire servers in a non-conductive dielectric fluid. DTC is currently the more dominant solution for AI workloads. - Adopting liquid cooling can significantly improve a data center's Power Usage Effectiveness (PUE), a key efficiency metric. While a typical air-cooled data center may have a PUE between 1.4 and 1.6, liquid cooling can lower that figure to below 1.2, with some direct liquid cooling solutions achieving a partial PUE of 1.02 to 1.03. - While the initial capital expenditure for liquid cooling can be double that of air cooling for a new 1MW facility, it can lead to a 39% reduction in the total cost of ownership (TCO) over a 10-year period due to significant energy and maintenance savings. For retrofits, liquid cooling can also be more practical than overhauling existing air-based infrastructure to handle high-density racks. - Hyperscalers are leading the adoption trend, with Microsoft Azure, Google Cloud, and Meta all using liquid cooling for their large-scale AI infrastructure. Google uses custom liquid cooling for its TPUs, and Microsoft's Azure regions that support OpenAI's models are liquid-cooled. - The supply chain for this market includes not only cooling hardware specialists like Vertiv, CoolIT Systems, and Schneider Electric but also chemical and energy giants like Shell, Dow, and Exxon Mobil, who provide the specialized dielectric fluids required for these systems. - This shift has a significant sustainability impact, particularly on water usage. A large data center can use up to five million gallons of water per day for cooling. Liquid cooling technologies can reduce water consumption by 30% to 50% compared to traditional evaporative cooling towers.
