Z.AI’s Landmark AI Model Trained on Huawei Chips
- Z.AI released an open-source image generation model trained entirely on Huawei processors, marking a first for major AI models.
- The model, GLM-Image, uses a hybrid autoregressive-diffusion architecture with a total of 16 billion parameters.
- Z.AI’s IPO in Hong Kong raised $558 million, and its stock has risen by approximately 80% since listing.
- Chinese customs have blocked Nvidia H200 chips from entering the country, signaling a push for domestic technology reliance.
Z.AI’s release of the GLM-Image model demonstrates China’s growing capability to develop advanced AI systems using domestic hardware like Huawei’s Ascend processors. This move challenges Nvidia’s dominance in the AI chip market and aligns with China’s strategy to reduce dependence on U.S.-made technology.
This development underscores China’s commitment to achieving self-reliance in AI technology, as evidenced by Z.AI’s successful IPO and subsequent stock performance. Source
