Investment Rating - The report rates the industry as "Positive" [13] Core Insights - The Doubao large model has undergone significant updates, which are expected to drive the prosperity of the AI application market. The updates include enhancements in visual understanding and 3D production capabilities, with a cost reduction of 85% compared to industry averages, promoting commercial growth in AI applications [13][58]. - The AI industry has crossed the threshold of large-scale commercial applications of AI large models, leading to increased investments in AI Agent and other next-generation AI applications by domestic manufacturers. The demand for computing power driven by the performance improvements of AI large models is expected to accelerate the construction of computing power infrastructure [13][58]. - According to IDC, global AI capital expenditure is expected to reach 632 billion by 2028, with a compound annual growth rate (CAGR) of 29%. The capital expenditure for generative AI is projected to grow at a CAGR of 59% from 2024 to 2028, significantly higher than the 22% growth for other AI technologies [13][58]. Summary by Sections Section 1: Doubao Large Model Product Strength Enhancement - The Doubao large model's product matrix has been enriched, significantly enhancing its capabilities and reducing usage costs, which is expected to drive the commercial prosperity of AI applications [13][58]. Section 2: Accelerated Growth of the AI Industry - The AI industry is experiencing rapid growth, with applications and computing power as two major pillars. The demand for AI computing power is expected to grow significantly due to the development of large models [13][58]. Section 3: Computing Power Industry Chain - The demand for AI computing servers is expected to grow rapidly, with the construction scale of servers increasing significantly. For instance, training a large model with 175 billion parameters requires approximately 2,917 Nvidia A100 servers [10][11]. - The global shipment of ordinary AI servers and high-end AI servers is projected to reach 725,000 and 543,000 units in 2024, respectively, representing year-on-year growth of 54.2% and 172.0% [19]. Section 4: Chip Industry - The report highlights the significant gap in performance between domestic AI chips and international leaders like Nvidia. Domestic companies are accelerating the development of next-generation AI chips to bridge this gap [41][43]. Section 5: Optical Module Industry - The demand for optical modules is expected to increase significantly due to the high data transfer requirements of AI large models. The global optical module market is projected to exceed $20 billion by 2027, with a CAGR of 11% from 2025 to 2027 [50][55]. Section 6: Investment Recommendations - The report suggests focusing on investment opportunities in servers, liquid cooling equipment, chips, and optical modules, with specific companies highlighted for each category [58].
人工智能行业专题研究：豆包大模型迎来重磅更新，AI应用繁荣推动算力基础设施建设
源达信息·2024-12-31 08:05