Immersion liquid cooling enables high-density, energy-efficient server rack operation by submerging servers in dielectric fluids for superior heat transfer compared to air cooling.

Overview

Immersion liquid cooling involves submerging servers or components in a thermally conductive, electrically non-conductive fluid, allowing heat to be removed directly from CPUs, GPUs, and memory modules. This method is increasingly adopted in AI, HPC, and high-density data centers where traditional air cooling is insufficient for modern thermal loads .

Key Benefits

• High Power Density: Systems can support up to 4 kW per 1RU or more, with some enclosures handling 140 kW per rack, enabling more compute power in the same footprint .
• Energy Efficiency: Immersion cooling reduces energy usage significantly, often enabling complete heat recovery and lowering PUE (Power Usage Effectiveness) compared to air-cooled systems .
• Reduced Noise and Maintenance: Fewer fans are required, lowering noise levels, and servers remain accessible for maintenance without messy tanks or complex disassembly .
• Scalability and Flexibility: Modular designs allow single or dual-level configurations, supporting standard 19” racks and various server architectures, including 21” and ORV3 .

Deployment Options

1. Single-Phase Immersion: Servers are submerged in a dielectric fluid that absorbs heat and is circulated to a heat exchanger.
2. Two-Phase Immersion: The fluid boils at a low temperature, and the vapor is condensed externally, providing highly efficient heat removal.
3. Direct-to-Chip (DLC): Combines immersion with targeted cooling of high-heat components, enhancing thermal performance while minimizing fluid volume .

Integration Considerations

• Compatibility: Modern immersion enclosures support standard server form factors, allowing integration into existing racks without major infrastructure changes .
• Redundancy and Reliability: Systems often include 2N redundant pumps and heat exchangers to ensure uninterrupted operation .
• Coolant Selection: Engineered dielectric fluids like ThermaSafe R provide low viscosity, stable thermal performance, and efficient convective heat transfer .
• Space Optimization: Immersion enclosures can achieve 2-3x more compute per rack compared to traditional air-cooled setups, maximizing data center footprint efficiency .

Leading Providers

• LiquidCool Solutions: Offers plug-and-play immersion-cooled servers with direct-to-chip technology, supporting high-density computing up to 4 kW per server .
• Supermicro: Provides rack-scale liquid cooling solutions for AI and HPC, including in-rack or in-row CDUs, supporting up to 80 kW per rack .
• DCX: Delivers modular immersion enclosures like the H1000, optimized for AI, HPC, and edge deployments, with cooling capacities up to 140 kW per rack .

Conclusion

Immersion liquid cooling is a highly effective solution for modern high-density server environments, offering superior thermal management, energy efficiency, and scalability. It is particularly suited for AI, HPC, and edge computing applications, enabling data centers to maximize compute density while maintaining reliability and operational flexibility .