Hua Er Jie Jian Wen·2025-12-12 12:13Core Viewpoint - The current AI boom is not a single bubble but consists of three distinct types: valuation bubble, investment bubble, and technology bubble [1][2]. Valuation Bubble: Identifying Real Risks - The Shiller Cyclically Adjusted Price/Earnings ratio has exceeded 40, nearing the 44 times level seen at the peak of the 2000 internet bubble, indicating potential market overheating [2]. - Despite high valuations, they are primarily driven by profit growth rather than speculation, with the S&P 500 index showing a 22.7% annual growth trend since October 2022 [4]. - Large tech stocks have a valuation premium of about 60%, supported by over 20% profit growth differences [6]. - Private companies exhibit significantly higher valuations, with OpenAI's revenue forecast leading to a price-to-sales ratio of 38 times, while public tech giants like Nvidia and Microsoft have more reasonable valuations of 22 times and 12 times, respectively [9]. Investment Bubble: Cash Flow Support vs. Debt Risks - Capital expenditures for major cloud service providers are projected to reach $500 billion by 2026, potentially totaling $4 trillion by 2030, which is unprecedented but still within reasonable limits [11]. - The annual growth rate of global tech capital expenditures has been 12.3% since 2013, indicating that current growth remains within this trend [13]. - Investment returns for large tech companies have been rising, driven by demand for cloud services and AI tools [14]. Technology Bubble: Usability and Scalability Concerns vs. Technological Progress and Demand Growth - Generative AI still faces challenges such as errors and hallucinations, making large-scale application difficult, and physical bottlenecks may soon hinder rapid expansion [16]. - Unlike the debt-driven internet bubble, current AI investments are primarily supported by free cash flow, with Google reporting an operating cash flow of $48 billion in Q3 [17]. - Demand for AI is growing, with Google processing 1.3 quadrillion tokens monthly, and less than 10% of U.S. businesses currently using AI, indicating significant growth potential [20]. - The launch of Gemini 3 by Google demonstrates ongoing advancements in AI capabilities, surpassing previous models in various tests [21]. - Cost reductions in AI models have been significant, with the cheapest large language models dropping in cost by 1000 times, driving consumption growth [24]. Potential Triggers for Bubble Burst - Recent complex agreements, such as OpenAI's $1.4 trillion computing purchase commitment, may introduce systemic risks due to valuation opacity [27]. - Even cash-rich cloud service providers are increasing debt issuance, with investment-grade bond issuance exceeding $35 billion in 2025, raising concerns about rising debt costs [30]. - The report highlights diminishing returns on scale, with training costs for AI models skyrocketing from $10 million to over $1 billion [32]. - Growing skepticism towards AI, particularly in developed markets, may lead to customer resistance and regulatory challenges [34]. - Energy supply may become a significant barrier to AI adoption, with electricity demand projected to quadruple by 2030 [37].