CDUCoolant Distribution Unit
Coolant Distribution Units for high-density data centers.Direct-to-chip liquid cooling for AI, NVIDIA GPU, and HPC.
Air Is No Longer Enough
With rack densities reaching 200 kW and AI chips with TDPs over 1000W, air cooling has reached its limits. The CDU is the solution.
High-Power GPUs
NVIDIA H100, B200, and GB200 generate heat that air systems cannot dissipate efficiently.
1000x Efficiency
Water has 1000x the thermal capacity of air, allowing for more efficient and compact cooling.
40% Savings
Significant reduction in energy consumption compared to traditional air conditioning systems.
Higher Density
Allows for 100+ kW racks in the same space, maximizing data center capacity.
Helioterm CDU: 50 kW to 1 MW
Solutions for all data center scales, from edge computing to hyperscale.
Liquid-to-Liquid
In-rack solution for direct-to-chip cooling of high-density servers. Integration with existing chillers.
Capacity
100 kW
Type
Liquid-to-Liquid
Technical Specifications
Recommended Application
In-Rack, NVIDIA GPU Servers
- Redundant pumps (N+1) with automatic failover
- 316L stainless steel plate heat exchanger
- 25 μm filtration system with saturation indicator
- Integrated leak detection with alarm
- PLC controller with 7" touchscreen HMI
- Protocols: Modbus TCP/IP, BACnet, SNMP
- Dual power inputs (optional)
- RAL 7035 steel cabinet with anti-corrosion treatment
Configurations for Every Need
Choose the ideal architecture for your data center: in-rack, in-row, or gallery.
CDU mounted inside the rack for localized thermal control.
- Up to 100 kW per rack
- Ideal for edge and GPU clusters
- Quick installation
CDU positioned between rows of racks for centralized cooling.
- 100-500 kW per unit
- Serves multiple racks
- Modular and scalable design
Installation-scale system for large data centers.
- 500 kW to 1+ MW
- Cools multiple rows
- Coordinated control
Where the Helioterm CDU Excels
Our CDUs are designed for the most demanding high-performance computing environments.
Cooling for NVIDIA H100, B200, and GB200 GPU clusters, ensuring maximum performance in training and inference.
Precise thermal control for complex simulations, scientific research, and large-scale data modeling.
Scalable solutions that grow with demand, optimizing PUE and reducing operational costs (OPEX).
Get Your Questions Answered
Answers to the most common questions about our CDUs.
Need a Custom Solution?
Our engineering team develops custom CDUs to meet your project's specific requirements.
Talk to a Specialist