GPU & AI servers from China
We build GPU nodes on Dell R760xa, HPE DL380a, Lenovo SR675 V3 and xFusion chassis with NVIDIA H100 NVL and H200, L40S for inference, RTX PRO 6000, and A100 for budget builds. Air- or liquid-cooled, 8-way capable, with the NVLink, InfiniBand and PSU headroom the accelerator actually needs.
Flagship with integrated 8-GPU modules for training neural networks.
4U with up to 10 DW GPUs for AI inference.
2U GPU server with DDR5 and PCIe 5. 0.
4U GPU server for AI inference.
Flagship for supercomputer clusters.
Supercomputing platform with DLC.
4U AI platform V4 with the latest GPUs.
2U AI-optimized version of the SR650 V4.
HPC node AMD + NVIDIA GPU with Neptune.
HPC node on AMD EPYC with Neptune.
Node with NVIDIA GPU and Neptune DWC.
Node with Xeon Max Series (HBM) and Neptune.
Neptune direct liquid cooled unit.
4U AI server with DLC for LLM training.
4U AI platform with AMD EPYC and 8× NVIDIA H100/H200.
4U AI platform with 8× SXM5 GPUs.
5U liquid cooling, 8× NVIDIA H200 / B200 (Blackwell).
Composable HPC module for AI clusters.
2U on NVIDIA GH200 NVL2 (Grace CPU + Hopper GPU).
4U AI version with NVIDIA GH200 NVL2 support.
HPC node for Apollo 2000 Gen10.
A higher-density GPU variant of the Server 9712.
Platform with DLC for NVIDIA GB200 rackmount solutions.
Flagship platform for supercomputer clusters.
GPU platforms for AI training and inference
We build GPU nodes on Dell R760xa, HPE DL380a, Lenovo SR675 V3 and xFusion chassis with NVIDIA H100 NVL and H200, L40S for inference, RTX PRO 6000, and A100 for budget builds. Air- or liquid-cooled, 8-way capable, with the NVLink, InfiniBand and PSU headroom the accelerator actually needs.
Every order ships EXW Hangzhou, so you see a clean equipment price and keep control of the freight. If you would rather not import, we arrange a forwarder for DAP or DDP to your city. Units are burned in for 24 to 72 hours and carry a 1-year warranty with advance replacement.