On September 19, Zhu Zhaosheng, the Head of Huawei's Computing Strategy and Business Development Department, announced that Ascend has successfully navigated past the ecological inflection point, positioning itself with both the capability and the golden opportunity to construct a cutting-edge AI ecosystem in the Agentic era. In terms of community engagement, the Ascend CANN open-source community boasts an average of over 5,200 monthly active users. Since June of this year, it has emerged as the most vibrant open-source community in China, with the number of non-Huawei developers surpassing that of Huawei's own developers, and an impressive average of over 30,000 lines of new code being integrated daily. At the model development front, more than 40 large-scale language and multimodal models, pre-trained on the Ascend platform, have been successfully developed, establishing Ascend as the sole AI computing power platform in China that supports pre-training. On the ecological front, in addition to open-sourcing CANN itself, Ascend has seamlessly integrated with mainstream open-source frameworks like PyTorch, further enhancing its ecosystem's robustness and versatility.