Investment Rating - The industry investment rating is "Positive" and maintained [8] Core Insights - Recent updates of overseas large models have led to significant performance improvements, driving the AI application landing phase [4][5] - The demand for computing infrastructure is surging due to the increased usage of large models, indicating a high growth cycle for the computing infrastructure industry [5][28] - Domestic AI applications are growing but at a slower pace compared to overseas counterparts, with potential acceleration as domestic models close the performance gap [6][34] Summary by Sections Overseas Large Models and AI Application Growth - Continuous updates from major overseas model developers like xAI, Google, and Anthropic have enhanced model performance, leading to a substantial increase in AI application usage [4][16] - For instance, Google's AI Token monthly usage surged from 9.7 trillion in April 2024 to 480 trillion in April 2025, marking a 50-fold increase [25] Domestic AI Model Development - Domestic AI applications are experiencing growth, with notable MAU figures for various applications as of June 2025, such as Quark (15,618 million) and Doubao (12,594 million) [31][34] - The performance gap between domestic and overseas models is narrowing, particularly with advancements from companies like DeepSeek and Qwen [34] Investment Recommendations - Focus on infrastructure tailored to Chinese characteristics, emphasizing data, computing optimization, and adaptation, particularly in IDC and domestic computing supply chains [7][38] - Highlight significant application scenarios for AI, such as education, justice, and healthcare, where AI applications are expected to thrive [7][38]
AI应用拐点已至,聚焦Infra与大场景