“读万卷书”不如“行万里路”!芯原股份掌舵人戴伟民详解AI芯片下一站:端侧推理与场景落地

Core Insights - The Shanghai Stock Exchange International Investor Conference highlighted the significant growth in demand for AI customized chips (AI ASIC) as articulated by Dai Weimin, Chairman and CEO of Chipone [1][3] - The relationship between GPU and AI ASIC was clarified, emphasizing that they complement each other rather than operate independently, with AI ASIC offering cost-effectiveness and GPU providing flexible deployment [3][4] - The evolution of AI requires a "world model" that goes beyond text and image processing to include spatial, physical, and contextual information, thus demanding diverse computational power [4][5] Industry Trends - The industry is witnessing a clear division in computational power needs between "cloud" and "edge" computing, with edge computing representing a significant value opportunity [5][7] - The rise of edge inference is seen as a critical area for AI applications, particularly in devices like smartphones, cars, and IoT devices, which will drive AI commercialization [5][6] - Chipone is focusing on core IP and AI ASIC solutions to support the shift towards edge computing, positioning itself strategically in this emerging market [8] Market Opportunities - The potential for AI applications in smart glasses and AI toys was highlighted, showcasing how customized chips can enhance user experiences and address market gaps [7][8] - The company believes that empowering end devices in various industries will lead to the next trillion-dollar market opportunity, emphasizing the importance of offline capabilities for privacy and security [7][8]