MetaX C500 64GB

MetaX C-500 is MetaX's first AI training accelerator on GPGPU/XCORE architecture using a 7nm process. Equipped with 64 GB HBM2e memory, it delivers approximately 75% of NVIDIA A100 performance.

The MXMACA 2.0 platform provides CUDA ecosystem compatibility. MetaX server configurations support up to 16 C-500 cards per unit - exceeding the standard 8-card configuration.

MetaXLink supports scaling at 384 GB/s.

C-500 is widely applied in AIGC (AI Generated Content) workloads - text, image, and video generation.

Application scenarios

Neural network and LLM training in enterprise data centers.
AIGC: text, image, and video generation workloads.
HPC computing and big data analytics.
Scalable MetaXLink clusters (up to 16 cards per server).
CUDA-compatible workloads on domestic hardware.