Transform any air-cooled Multi-GPU server into a high-performance liquid-cooled system. Unlock sustained GPU performance without thermal throttling.
Our standalone Drycooler CDU enables rapid deployment of liquid cooling in any environment. Ready-to-install kits for standard servers — start cooling in days, not months.
+30% PerformanceSustained GPU boost clocks
-40% PowerFacility power consumption
-50% Cooling CostData center energy savings
+40°C AmbientHarsh environment ready
Zero Throttling24/7 continuous operation
NVIDIA: H200, H100, A100, L40S
AMD: MI300X, W7900
Precision-engineered waterblocks for each GPU model
Supermicro: SYS-421GE-TNRT, SYS-821GE-TNHR
Gigabyte: G593-SD0, G593-ZD2
Dell: PowerEdge XE9680
HPE: ProLiant DL380a Gen11
And other 8x GPU platforms
Also available for CPU-intensive servers — convert air-cooled platforms to liquid cooling for overclocking and noise reduction. Supports Dual EPYC and Dual XEON waterblocks.
Engineered for reliability, not just assembled. Every component is designed for 24/7 operation in demanding data center environments.
Custom-designed cold plates for each GPU model ensure optimal thermal contact and heat transfer efficiency
Tool-free, drip-free connections for easy maintenance and hot-swap capability without draining the loop
Connects to Comino InRack Drycooler for standalone deployment or to facility CDU for data center integration
Automotive-grade fittings and pressure-tested manifolds ensure reliable 24/7 operation
Choose the deployment model that fits your infrastructure.
Connect to Comino InRack Drycooler — a self-contained cooling solution that fits in a standard 42U rack alongside your servers. No external infrastructure required. Perfect for edge deployments, research labs, and office environments.
Connect to facility CDU (Cooling Distribution Unit) or building chilled water loop via standard quick-disconnect manifolds. Compatible with existing liquid cooling infrastructure. Ideal for high-density deployments.
The Integration Kit is compatible with multi-GPU server platforms from Supermicro, Gigabyte, Dell, HPE and other manufacturers. We support NVIDIA H200, H100, A100, L40S GPUs, as well as AMD MI300X and W7900. Contact our engineering team for specific model compatibility.
Installation typically takes 4-6 hours for a qualified technician. All components are designed for tool-free installation where possible, using TheQ Quick Disconnects for easy maintenance.
No external infrastructure is required if you use the Comino InRack Drycooler, which fits in a standard 42U rack alongside your servers. Alternatively, you can connect to your facility's existing CDU (Cooling Distribution Unit) or building chilled water loop via standard quick-disconnect manifolds.
You can expect up to 30% higher sustained performance as GPUs maintain boost clocks without thermal throttling during extended AI training and inference workloads. The exact improvement depends on your workload characteristics and ambient conditions.
Yes, the installation is designed to be reversible. All components can be removed without permanent modification to your server, allowing you to restore the original air-cooled configuration if needed.
The liquid cooling system requires minimal maintenance. TheQ Quick Disconnects allow for easy component replacement without draining the entire loop. We recommend annual inspection of fittings and periodic coolant replacement (typically every 2-3 years) depending on your operating environment.
The Integration Kit components come with a standard warranty. For specific warranty terms and coverage, please contact our sales team.
