未知机构：国产算力6算力紧缺主流大模型适配国产芯片印证算力需求旺盛东北计算机-20260224

Summary of Conference Call Records Industry Overview - The conference call discusses the domestic computing power industry, particularly focusing on the demand for computing resources driven by AI applications during the Spring Festival in 2026, highlighting a significant supply gap in computing resources [1][2]. Key Points and Arguments 1. Beanbag App Performance: The Beanbag app, owned by ByteDance, experienced a surge in user traffic during the Spring Festival, leading to a temporary suspension of its video calling feature on February 16 due to computing power constraints. This incident illustrates the immediate impact of resource shortages on AI applications [2]. 2. AI Interaction Metrics: On New Year's Eve, the Beanbag app handled 1.9 billion AI interactions, with its large model achieving a peak inference throughput of 63.3 billion tokens per minute. This reflects the critical demand for computing infrastructure in deploying AI large models [2]. 3. Competitive Landscape: Major tech companies like Tencent and Alibaba are intensifying their investments in AI applications, leading to fierce competition for consumer traffic. This competition is expected to further escalate the demand for computing power in AI model training and inference [2]. 4. Trends in Chip Development: There is a growing trend for large companies to develop their own chips, while third-party chip manufacturers are accelerating the iteration of GPU and CPU technologies. This shift is likely to solidify competitive positions among leading firms [2]. 5. Integration of Domestic Models and Chips: The release of the new flagship model GLM-5 by Zhipu on February 12 marks a significant milestone in the integration of domestic large models with domestic computing hardware, achieving deep optimization and stable operation on major domestic chip platforms [3]. 6. Industry Chain Development: The increasing demand for computing power due to the large-scale deployment of AI applications is expected to accelerate the development of the entire industry chain, including domestic computing hardware, service operations, and infrastructure construction [2][3]. Additional Important Content - Related Companies: The call mentions several companies involved in computing power and storage, including Haiguang Information, He Sheng New Materials, and Cambrian Technology, among others, indicating a broad ecosystem supporting the computing power industry [3]. - Risk Factors: Potential risks highlighted include slower-than-expected policy progress, lower-than-expected downstream demand, and risks associated with artificial intelligence [4].