One-phase immersion-cooling rack for high-density GPU servers
A rack-level one-phase immersion-cooling system that circulates dielectric fluid around server equipment and transfers absorbed heat to chilled water through an external heat exchanger.
Problem solved
Addresses the thermal-management constraint of high-density AI and GPU server deployments by removing heat from the complete server assembly rather than relying only on air cooling or chip-level cold plates.
Application contexts
- High-density AI and GPU server environments
- HPC systems
- Edge computing installations
- High-load computing systems
Technical principle
Servers operate in a fluid tank containing dielectric cooling fluid. A pump circulates warmed fluid to a heat exchanger, where it exchanges heat with chilled water from external cooling equipment; cooled fluid returns to the rack in a repeated cooling cycle.
Typical use cases
- Rack-level cooling for AI and GPU servers
- Thermal management for high-density compute racks
Partnership route
How global engineering teams typically engage this asset.
Helpful to include: server model and rack configuration · GPU and CPU thermal design power · target rack power density · available chilled-water temperatures and flow · site floor loading and installation constraints · required maintenance and monitoring scope
Have specific parameters to run against this asset?
Submit a Challenge BriefProvider
- Source: Immersion cooling equipment ICEraQ series
- Original language: JA
- Summary AI-assisted; translation status: Unreviewed