瑞银：当前AI热潮处潜在泡沫早期 关键见顶信号未现

Core Viewpoint - UBS's global equity research team indicates that the current market is in the early stages of a potential bubble, but key signals that typically indicate a bubble peak—extreme valuations, long-term overheating catalysts, and short-term topping events—are currently absent [1] Group 1: Market Conditions - The U.S. stock market has met all seven prerequisites for bubble formation, including a 14 percentage point annualized return over bonds in the past decade, significant new technologies, a 25-year gap since the last bubble, overall profit pressure, market concentration, retail investor buying, and loose monetary conditions [3] - UBS emphasizes that simply comparing the current AI boom to historical bubbles is overly simplistic, as the logic behind its formation is more rational in two aspects [3] Group 2: AI and Productivity - Generative AI shows unique disruptive potential and adoption speed, with OpenAI attracting 800 million users in just three years, compared to Google's 13 years for the same user base [3] - If generative AI can temporarily boost productivity growth by 2% as expected, it could support a 20-25% upside in the stock market [3] Group 3: Macro Risk Structure - The macro risk structure has fundamentally changed; unlike the budget surplus during the 2000 internet bubble, the current U.S. government debt-to-GDP ratio is double that of the past, with high fiscal deficits [3] - In contrast, corporate balance sheets are relatively robust, with the tech giants' price-to-earnings ratio, excluding Tesla, at 35 times, significantly lower than the 45-73 times during the bubble period [3] Group 4: Semiconductor Market Potential - UBS estimates that if semiconductor industry spending reaches 1.3% of global GDP by 2030, the current valuation would be justified, with investment logic still based on earnings and cash flow [4] - Long-term structural factors that typically lead to bubble bursts are not currently evident, as over-investment signals have not appeared, and U.S. telecom technology investment as a percentage of GDP remains below the 2000 peak [4] Group 5: Debt Financing and Market Stability - Debt financing risks are low, as leading data center companies' capital expenditures to sales ratio is close to the 2000 telecom level, but tech giants primarily rely on cash flow rather than debt for investments [4] - UBS calculates that these companies would need a 40% increase in capital expenditures to start utilizing debt financing, contrasting sharply with the 3.5 times net debt/EBITDA ratio of telecom companies during the internet bubble [4] Group 6: Market Sentiment - The current market breadth is not as extreme as in 1999, and overall U.S. corporate profits remain stable [5] - Despite this, UBS finds that the market perceives a 20% probability of a bubble forming and advises investors to identify key signals that indicate a potential bubble burst as a core aspect of future investment decisions [5]