AI Accelerators & Compute Servers
Full catalogue of Chinese domestic AI compute hardware: accelerator cards, compute servers, supernodes and modules from Huawei Atlas, MetaX, Hygon, Biren, Moore Threads and Alibaba. Select a model to view specifications and request a quotation.
Huawei Atlas
Atlas 200I DK A2 Developer Kit
Easy-to-use AI development kit for fast innovation View details →
Huawei Atlas
Atlas 200I A2 Acceleration Module
Edge AI accelerator module embedded in smart devices, robots and drones View details →
Huawei Atlas
Atlas 300I A2 Inference Card
High-performance AI inference card integrated into servers View details →
Huawei Atlas
Atlas 300I Pro Inference Card
Inference accelerator card integrated into servers View details →
Huawei Atlas
Atlas 300I Duo Inference Card
Dual-chip high-density inference accelerator card View details →
Huawei Atlas
Atlas 300V Video Analysis Card
Video analytics inference card for HD video decoding and AI parsing View details →
Huawei Atlas
Atlas 300V Pro Video Analysis Card
High-channel video analytics inference card View details →
Huawei Atlas
Atlas 500 A2 Intelligent Edge Station
Fanless edge AI appliance for harsh outdoor edge deployment View details →
Huawei Atlas
Atlas 500 Pro Intelligent Edge Server (Model 3000)
High-compute edge inference server View details →
Huawei Atlas
Atlas 800 Inference Server (Model 3000)
Data-center AI inference server View details →
Huawei Atlas
Atlas 800T A2 Training Server
High-density AI training server View details →
Huawei Atlas
Atlas 800I A2 Inference Server
Modular large-model inference server with NPU interconnect View details →
Huawei Atlas
Atlas 900 AI Cluster
Large-scale AI training cluster for AI computing centers View details →
Huawei Atlas
Atlas 900 A2 PoD Base Unit
Liquid-cooled training cluster building block (PoD) View details →
Huawei Atlas
Atlas 900 A3 SuperPoD
384-NPU liquid-cooled AI supernode View details →
Huawei Atlas
Atlas 800T A3 Supernode Server
10U air-cooled training supernode server, composable to 384-NPU supernode View details →
Huawei Atlas
Atlas 800I A3 Supernode Server
10U air-cooled inference supernode server View details →
Huawei Atlas
Atlas 150
Training/inference standard accelerator card integrated into servers for strong AI compute View details →
Huawei Atlas
Atlas 350
Training/inference standard accelerator card integrated into servers for strong AI compute View details →
Huawei Atlas
Atlas 650E
AI server with strong compute power View details →
Huawei Atlas
Atlas 850E
Best fit for large-scale data centers, trillion-parameter model training and centralized... View details →
MetaX
N100
Intelligent Inference GPU View details →
MetaX
N260
Large-Model Inference GPGPU View details →
MetaX
N300
Large-Model Inference GPGPU View details →
MetaX
C500
General-Purpose GPU (GPGPU) View details →
MetaX
C550
General-Purpose GPU OAM Module View details →
MetaX
C500X
Optical-Interconnected GPGPU View details →
MetaX
C588
Flagship GPGPU OAM Module View details →
MetaX
C600
Next-Gen GPGPU OAM Module, Air/Liquid Cooling View details →
MetaX
N260 Server
Rackmount GPU Inference Server View details →
MetaX
N300 Server
High-Density AI Inference Server View details →
MetaX
C500 Server
Rackmount GPU Server View details →
MetaX
C550 OAM Server
Dual-Socket Flagship OAM Server View details →
MetaX
C588 OAM Server
Dual-Socket Flagship OAM Server, Air/Liquid Converged View details →
MetaX
C600 OAM Server
Dual-Socket Flagship OAM Server View details →
MetaX
N260 Workstation
Actively-Cooled GPU Workstation View details →
MetaX
C550 Shanghai Cube Liquid-Cooled Cabinet
High-Density Liquid-Cooled Full Cabinet (128 GPUs/rack) View details →
MetaX
C500X Optical-Interconnect Supernode
Optical-Interconnect Supernode (up to 8 servers / 64 cards) View details →
MetaX
C550 3D Mesh Supernode
Electrical 3D Mesh Supernode (up to 8 servers / 64 cards) View details →
Hygon
Hygon Deep Computing Processor DCU (specific model not disclosed)
High-Performance Deep Computing Processor (DCU) on a mainstream general-purpose parallel... View details →
Hygon
Heterogeneous Computing Server (specific model not disclosed)
Dedicated Heterogeneous Computing Server form factor based on Hygon processors View details →
Biren
BIREN 106B
FHFL Dual-Wide PCIe AI Training & Inference Accelerator Card View details →
Biren
BIREN 106M
Air-Cooled OAM Module for AI Training View details →
Biren
BIREN 166C
Datacenter High-Performance Inference PCIe AI Accelerator View details →
Biren
BIREN 166L
Train-Inference Converged Liquid-Cooled (Cold-Plate) OAM Module View details →
Biren
BIREN 166M
Train-Inference Converged Air-Cooled OAM Module (OAM V1.1) View details →
Moore Threads
MTT S5000
Universal GPU for AI Training and Inference View details →
Moore Threads
MTT S4000
Large-model AI inference and training GPU with Tensor Cores View details →
Moore Threads
MTT S3000
Full-featured server-class GPU View details →
Moore Threads
MTT S2000
Low-power full-featured server GPU for cloud-native workloads (translation draft) View details →
Moore Threads
MTT SGX5000
Integrated AI training/inference server with 8 S5000 OAM modules (translation draft) View details →
Moore Threads
MCCX D800 X1
8-GPU AI training/inference integrated server (translation draft) View details →
Moore Threads
MTT KUAE
Full-Stack Solution for AI Data Centers View details →
Moore Threads
MTT E300
Edge AI module built on the ChangJiang SoC (translation draft) View details →
Alibaba
Hanguang 800 NPU (AI Inference Chip)
Alibaba's first AI inference chip (NPU) for cloud search, recommendation, advertising and... View details →
Alibaba
Zhenwu M890 All-Scenario Training & Inference AI Chip
Fully self-developed train-and-inference AI chip supporting FP8/FP4 low precision and... View details →
Alibaba
Zhenwu V900 Training & Inference AI Chip
High-performance train-and-inference AI chip with 216 GB memory, 1200 GB/s interconnect... View details →
Alibaba