中昊芯英全自研 TPU 架构 AI 芯片与计算集群重磅亮相2025WAIC

Group 1 - The WAIC 2025, held from July 26-29 in Shanghai, is the highest-level and most influential event in China's artificial intelligence sector, co-hosted by multiple national departments and the Shanghai municipal government [1] - Zhonghao Xinying showcased its self-developed high-performance TPU AI chip "Shan Na" and AI server "Tai Ze" at the event, highlighting its core competitiveness in domestic innovation in computing infrastructure [1][2] - The event featured a dual exhibition model that presented three core scenarios: computing hardware, solutions, and cloud platform experiences, attracting numerous professional audiences [2] Group 2 - Zhonghao Xinying's CMO shared insights on AI trends and the company's technological practices, emphasizing the declining value of CUDA and the advantages of AI-specific chips like TPU for future infrastructure [3] - The company’s chief solution architect discussed the importance of system-level aggregation and efficient scheduling in TPU architecture, showcasing its full-stack AI infrastructure platform [3] - Zhonghao Xinying has collaborated with various sectors, including finance, healthcare, and education, to promote the application of AI technology [3] Group 3 - At the K&H Data Computing Infrastructure Forum, Zhonghao Xinying participated in strategic partnerships with multiple companies across the AI and IT service sectors, aiming to build a more comprehensive AI industry ecosystem [4] - The "North China Computing New Engine" event highlighted the challenges of computing power for large parameter models and how the GPTPU architecture can reshape AI computing efficiency [5] - The Tianjin Mobile TPU Intelligent Computing Center was inaugurated during WAIC, marking a significant collaboration among Tianjin Mobile, Zhonghao Xinying, and other partners to establish a new benchmark for domestic computing power [6][7] Group 4 - The collaboration among Tianjin Mobile, Taiji Co., Zhonghao Xinying, and Bozhong Intelligence aims to enhance AI computing power and contribute to the high-quality development of national computing infrastructure [7] - Zhonghao Xinying's commitment to promoting domestic TPU architecture and its determination to drive the industrialization of AI were emphasized during the event [7]