基本面观察2月第2期：AI叙事的转变

Group 1: AI Narrative Shifts - The global AI narrative is experiencing significant marginal changes, with at least three layers of transformation observed[4] - The first narrative shift indicates a divergence regarding the Scaling Law, highlighting physical constraints, data bottlenecks, and diminishing marginal returns on investment in AI models[5] - The second narrative shift reflects a transition from "rewarding" CAPEX to anxiety over slow monetization, with projected AI-related capital expenditures in the U.S. exceeding $700 billion by 2026, representing over 2% of GDP[6] Group 2: Market Concerns and Impacts - The third narrative shift involves deeper concerns about AI's disruptive potential across various industries, evolving from changing search methods to transforming software applications and business processes[7] - The anticipated capital expenditures by major U.S. tech firms will consume approximately 90% of their operating cash flow in 2026, up from 65% in 2025, raising concerns about negative free cash flow[6] - The market is currently pricing in a relatively worst-case scenario due to panic-driven sentiment, despite resilient fundamentals in many affected companies[10] Group 3: Investment Strategies - Investors are advised to shift from a broad "buy a basket of AI" approach to a more refined selection of targets, focusing on which changes are likely to occur and which are not[11] - Key investment perspectives include identifying hardware segments with strong supply constraints, competitive model layers with proprietary data, and application layers that can quickly demonstrate AI's value[12] - The differences in AI development paths between China and the U.S. suggest that investment logic in China may focus more on "industrial empowerment" rather than mere labor replacement[14]