Domestic Large Model Landscape - China's domestic large model sector is experiencing a surge in development, characterized by rapid iteration and competitive releases [1] - Kimi K3 has been launched with a parameter scale of 2.8 trillion (2.8T) and a context window capacity of 1 million (1M) tokens [1] - Qwen 3.8 is rumored to feature a parameter scale of 2.4 trillion (2.4T) [1] - Domestic AI firms are engaging in aggressive price wars and actively pursuing open-source strategies for international expansion [1] Operational Challenges and Strategic Shifts - High demand for domestic models has led to temporary suspensions of new subscriptions and highlighted performance bottlenecks in processing speed [1] - The industry's competitive focus is shifting from raw model parameters to user product experience [1] - Global AI leaders, including Claude and OpenAI, are implementing retention strategies to counter market competition [1] Industry Outlook and Investment Logic - The AI sector is entering a phase of direct competition between Chinese and American models over the next two to three years [1] - Market attention remains concentrated on the AI infrastructure supply chain, including GPU, HBM, storage chips, and high-speed optical modules (800G/1.6T) [1] - Investment focus encompasses the entire AI ecosystem, ranging from cloud computing and data centers to power infrastructure and advanced packaging technologies like CoWoS [1]