Wikipedia: MetaX

SECONDARY aggregator; the underlying citations are DIGITIMES (Li, 2025-01-22 and Liang, 2025-04-10) and IT Home. Carries the only public N100/C500 specs: C500 is 7 nm, 15 TFLOPS FP32 (~75% of A100), MACA 2.0, CUDA-compatible, multi-GPU interconnect; N100 is the inference part, HBM2E, 160 TOPS INT8 / 80 TFLOPS FP16, 128-ch encode and 96-ch decode of 8K HEVC/H.264/AV1/AVS2, mass production 2023. Product-line taxonomy is useful: MXN = inference, MXC = training/HPC, MXG = graphics. C500 ran out of stock during 2024 and a downgraded C280 was submitted to TSMC; BAAI pre-trained a MoE model on C-Series clusters in June 2024; >10,000 GPUs across 9 clusters in commercial operation by end-2024. SSE IPO 17 December 2025 (ticker 688802).