All Products
Explore our cutting-edge hardware solutions to meet general server, AI server, data storage, and switch needs.
General Servers
-
Based on Intel Xeon 4th and 5th generation scalable processors with up to 350W TDP
32 DIMM slots support with up to 8TB of memory; RDIMMs up to DDR5-5600
CXL memory expansion on demand
-
Up to 12x 3.5-inch HDDs
Up to 28x 2.5-inch HDDs
Up to 8x standard PCIe slot and 1x OCP 3.0 slot
Up to 2x DW GPU slots
2U2S General Server
GPU Servers
-
Easy Operation & Maintenance
Optical modules in cold air intake zone for enhanced network reliability
-
CPU-Switch with X32/X16 bandwidth interconnection
Compatible with NVIDIA HGX B200 liquid cooling module
Optional full/half-height network cards
Designed for liquid-cooled data centers
-
N+N redundant PSUs
Efficient liquid cooling design for stable high-temperature operation
HGX B200 GPU Liquid-Cooled Server
-
Easy Operation & Maintenance
Optical modules in cold air intake zone for enhanced network reliability
-
CPU-Switch with X32/X16 bandwidth interconnection
Compatible with NVIDIA HGX B200 air cooling module
Optional full/half-height network cards
-
N+N redundant PSUs
Efficient thermal design for stable operation in high ambient temperatures
HGX B200 Air-Cooled Server
4U 8 PCIe-GPU Server
-
Based on Intel® Xeon® 4th & 5th SPR&EMR Scalable
Processors, with a TDP of up to 350W
Support the latest DDR5 and PCIe Gen5 technologies
Front and rear IO maintenance – friendly design compatible with both new and legacy server rooms
-
CPU-GPU supports passthrough and switch
solutions for up to 10 DW 350W GPUs
CPU-Switch allows for x16 or x32 lane interconnections achieve high-bandwidth communication
Supports various CPU-GPU interconnection topologies suitable for diverse scenarios
-
Strong computing performance with up to 350W CPU and 32 DIMM DDR5 memories
Compatible with the latest NVIDIA® HGXTM H100/ H200 module and Intel Gaudi®2 OAM module
-
Flexible x16 or x32 PCIe link bandwidth between CPU and Switch
8x PCIe 5.0 x16 FHHL (from switch) + 2x PCIe 5.0 x16 FHHL slots (from CPU)
SmartNIC supported
-
Front IO access
Air cooling and cold plate cooling optional
Reservation support for next-gen NVIDIA® HGXTM and OAM 2.0 module
8U HGX H100/H200 8-GPU Server
Network Switches
-
128 × QSFP112 400GbE ports supported
Up to 102.4 Tbps full-duplex switching capacity
-
Redundant design for power supply and fan
Software real-time monitoring of system status
-
Pre-loaded with Open Network Install Environment (ONIE)
Supports open-source SONIC and multiple commercial NOS
128x400G(QSFP-112G) 4RU TH5 112G Network Switch
-
24 × QSFP56 100GbE + 8 × QSFP-DD 400GbE ports
Up to 16 Tbps full-duplex switching capacity
-
Redundant design for power supply and fan
Software real-time monitoring of system status
-
Pre-loaded with Open Network Install Environment (ONIE)
Supports open-source SONIC and multiple commercial NOS
24x200G + 8x400G 1RU TD4 Network Switch
-
2RU design with 64×800G ports delivers 102.4T non-blocking switching capacity (full duplex)
-
PHYless architecture ensures superior signal integrity
Air-cooling support
-
Pre-loaded with Open Network Install Environment (ONIE)
Supports open-source SONIC and multiple commercial NOS
64x800G 4RU TH5 Network Switch
Rack
-
High density design: 64 GPUs interconnected in one High- Bandwidth Domain
Scalability: Expandable to 128/256/512 GPUs
High-Speed Cable Tray Design: 112G/224G
capacity, totaling 6144 copper cables
Hybrid Cooling Design: Combines liquid and air cooling to enhance data center PUE
Multi-Level High-precision Position Alignment: Supports blind insertion
Rack Power: up to120KW+
Support Rack-Level Transportation
and Deployment
Support Node and Rack-level Liquid Leakage Detection