UII UPDATE 431 | NOVEMBER 2025
For the past few years, applications such as high-performance computing (HPC) and AI training have put pressure on data center cooling systems to accommodate escalating power demands of processors and increasing rack density. Many organizations have responded by deploying some direct liquid cooling (DLC). DLC removes heat from IT more efficiently, potentially unlocking higher IT performance (e.g., overclocking of CPUs) or chillerless and waterless heat rejection (lower capital expenditure and power/water use). These two objectives tend to oppose each other in a trade-off, necessitating the optimization of cooling systems to favor one over the other.
The Uptime Institute Cooling Systems Survey 2025 recorded a slight, but notable, drop in the share of respondents naming environmental sustainability as a driver of DLC adoption compared with past results. Rising rack densities and individual server heat loads retained the top rankings. This suggests that maximizing IT performance is not only a strong incentive to convert to liquid cooling, but also that operators will likely continue to prioritize application performance over sustainability when the two are at odds.
Uptime Intelligence’s research draws on both survey data and operator briefings to identify the most influential drivers of DLC in current deployments and short-term plans.
The 2025 survey respondents represent a diverse cross-section of the industry and are closely examining DLC. Nearly one in five (18%, n=388) of the 2025 respondents are already using liquid cooling, and a small majority (62%) are considering DLC but not yet deploying it.
In recent years, a steady minority of respondents has named environmental sustainability as a primary driver of DLC adoption. In 2025, this figure slipped down to just above one in five (22%), the lowest since Uptime Intelligence began gathering data in 2021 (Figure 1).
Figure 1 Dense IT still drives direct liquid cooling
Currently, which of the following options are the primary drivers for direct liquid cooling adoption? Choose no more than two.

Respondents described the specific factors they say define a sustainable cooling system. The top responses were energy efficiency (87%, n=466), environmental impact from coolants or refrigerants (35%) and water consumption (35%).
DLC affords organizations a demonstrated opportunity to optimize for some efficiency goals (such as higher facility water temperature, lean heat rejection with zero or minimal use of chillers or evaporative cooling), but it also complicates data center operations and resiliency design. As a result, its adoption remains concentrated in applications where return on investment is most evident: usually where IT thermals (rack density, individual server heat loads) render air cooling impractical or altogether insufficient. To date, HPC and AI training applications have tended to produce the strongest business cases to justify the challenges of aggressive IT densification (see AI embraces liquid cooling, but enterprise IT is slow to follow).
In 2024, Uptime Intelligence predicted that much of DLC’s sustainability potential would go unrealized (see DLC will not come to the rescue of data center sustainability). The reason for this is neither a deficiency in cooling performance, nor operator opposition to sustainability initiatives themselves. Instead, the primary driver for organizations to explore DLC tends to be the need to densify for IT performance — and this outweighs sustainability in cooling design and optimization stages. Improved data center water and energy efficiency will require comprehensive improvements to cooling systems from IT to heat rejection, as well as increased attention to efficiency in IT hardware.
This year’s decrease in respondents naming sustainability as a DLC driver does not necessarily mean the industry is disinterested in energy-efficient or water-efficient operations. Each year, more data center organizations are advancing their liquid cooling plans: some are making early inquiries with equipment vendors, while others are proceeding with proof-of-concept installations or their first deployments in production environments. Still, some organizations will abandon or postpone DLC plans where the business case is unclear or does not justify the challenges specific to liquid coolants.
Regardless, all these organizations now have a clearer understanding of the factors that are moving their DLC plans either forward or backward. Operator perceptions of DLC drivers indicate that sustainability, although valuable, remains secondary to IT performance in driving the deployment of liquid cooling. If these priorities reverse in the future, it will likely be due to regulatory changes rather than operating expenses for electricity or water.
Other related reports published by Uptime Institute include:
AI embraces liquid cooling, but enterprise IT is slow to follow
DLC will not come to the rescue of data center sustainability