推理侧打开长期空间，中欧基金代云锋、杜厚良如何看AI投资？

Core Insights - The 2026 market is at a critical juncture of cyclical iteration and structural transformation, with a focus on technology, cyclical industries, consumption, and new energy sectors [1][6] - The investment strategy meeting highlighted how the industrialized research system of the fund can adapt to complex market changes while maintaining a long-term perspective [1][6] Technology Sector - The North American market has seen a commercial closure on AI inference, shifting the investment focus from training to inference demand, which supports long-term industry prosperity [1][6] - The fund manager believes that breakthroughs in AI applications will emerge, particularly in AI infrastructure and smart hardware, with significant investment opportunities in humanoid robots and autonomous driving [1][6] AI Application Directions - Three key areas for AI applications include: 1. Consumer entertainment, with new scenarios like AI advertising and AI education emerging [2][7] 2. Optimization of enterprise operations through AI, enhancing efficiency and reducing costs [2][7] 3. New paradigms in humanoid robots and intelligent driving [2][7] Autonomous Driving Focus - In the autonomous driving sector, three focal points are identified: 1. Flexibility in vehicle manufacturers, which may lead to benchmark companies [2][7] 2. Demand for new software and hardware in the components sector, emphasizing domestic supply chain advantages [2][7] 3. The potential for Robotaxi proliferation to create new business models and revenue-sharing opportunities [2][7] Supply Chain and Demand - The fund manager anticipates a significant increase in demand for computing power in both training and inference segments in 2026, with current supply bottlenecks due to reliance on overseas foundries for domestic chip production [2][7] - A mismatch between supply and demand is expected in 2026, necessitating rapid technological advancements and capacity building in domestic foundries [2][7] Market Dynamics - The fund manager warns that AI applications and new software may become a "gray rhino," indicating potential risks as resources concentrate on core AI areas, leading to supply chain pressures and increased upstream costs [2][7] - Innovations in end-user applications are likely to experience a "first dip then rise" pattern, with acceleration in enabling technologies like robots and autonomous driving expected after foundational models mature [2][7]