来源：市场资讯 （来源：太阳投研札记） 洪灏在《展望 2026：持而盈之》上下篇中，从全球经济格局、中美市场表现、政策走向及资产配置等 多个维度，分析了 2026 年市场的风险与机遇，核心观点可总结为以下两大方面： 特朗普发起的全球贸易战未能改善美国贸易赤字，反而因中国制造业优势和稀土行业整合，让美国谈判 筹码不足；关税政策推高美国通胀预期，进一步复杂化美联储决策。 美国经济与市场：风险积聚，泡沫隐现 1. 货币政策失去独立性，陷入 "三难" 困境 美联储货币政策逐渐从属于美国财政政策，因美国政府债台高筑，美联储需通过购债为财政赤字融资， 其决策受经济增长、通胀高企和金融稳定三重矛盾制约。 美国回购市场流动性紧张，隔夜逆回购工具近乎归零，美联储大概率将停止量化紧缩并重启扩表，但扩 表方式（买短端 / 长端债券）将影响资产价格走势，短期利好市场，长期恐因久期问题难改美股冲高回 落趋势。 2.经济周期步入晚期，半导体与经济周期背离 量化模型显示美国经济未来数月将放缓，而半导体周期仍在冲高，美股涨幅集中于少数科技巨头和 AI 相关企业，市场宽度持续收窄，科技股估值已接近甚至超过 2000 年互联网泡沫水平。 美国私 ...

点击小程序查看报告原文 近期，AI泡沫卷土重来、且似有愈演愈烈之势，这也成为引发美股、A股和港股三地市场AI相关成长风格、甚至整体风险资产大跌的罪魁祸首之一。过去 一周，AI泡沫的担忧导致全球科技股下挫，美元计价下，恒生科技（-7%）、创业板（-6%）和"美股七姐妹"（-6%）领跌。 图表：过去一周全球大类资产表现中，比特币、恒生科技、创业板以及"美股七姐妹"领跌 资料来源：Bloomberg，FactSet，中金公司研究部 虽然与此同时，美国政府关门、回购市场流动性紧张、美联储降息预期降温，国内数据转弱也都来"凑热闹"，压制了风险偏好， 但AI泡沫担忧才是最核 心的问题，毕竟这是当下中美三地市场最主要的景气方向之一。 例如，从2022年底ChatGPT发布以来，美股"七姐妹"最高上涨283%，大幅跑赢同期标普 500指数扣除M7后的69%；中国市场也是如此，2025年初DeepSeek发布以来，中国科技"十巨头"最高上涨81%，大幅跑赢恒指扣除"十巨头"后的19%，年 化表现甚至一度超过M7。可见，不论是市场表现、还是信用周期，AI趋势都牵一发而动全身，一荣俱荣、一损俱损（《 2026年展望：跟随信用扩张的 ...

Core Viewpoint - The article discusses the recent volatility in the global AI technology sector and the strong rebound of the Hong Kong stock market, particularly the Hang Seng Internet ETF, which has shown resilience amidst concerns about an "AI bubble" [1][2]. Market Performance - The Hang Seng Internet ETF (513330) tracking the Hang Seng Internet Technology Index surged by 3.88%, making it the best-performing tech index in Hong Kong [2]. - From November 17 to the present, the Hang Seng Internet Technology Index has experienced a cumulative decline of 2.25%, marking it as the least affected tech index in Hong Kong [2]. - The article highlights the contrasting views in the market regarding AI, with some fearing a bubble similar to the 2000 internet crash, while others see it as a final opportunity to invest in AI [5][6]. AI Sector Analysis - The article compares the current AI market to the 2000 internet bubble, noting that while there are concerns about overvaluation, many leading companies in the "M7" group have substantial revenues and profits, with annual revenues exceeding $100 billion and profits in the hundreds of billions [7][9]. - The capital expenditure to operating cash flow ratio for the "M7" group is 49%, lower than the 56% seen during the internet bubble peak, indicating a healthier financial position [8]. - The overall valuation of the "M7" companies is around 33 times earnings, which is lower than the 60 times seen at the peak of the internet bubble [9]. Domestic AI Landscape - In China, the AI sector is also experiencing a correction, but there are significant breakthroughs, such as Alibaba's AI assistant achieving over 10 million downloads in just one week [12]. - The capital expenditure of Chinese tech giants is projected to turn positive, with a growth rate of 18% by March 2025, reflecting a similar trajectory to overseas AI giants [12]. - The Hang Seng Internet Technology Index, which includes major companies like Alibaba, Tencent, and Baidu, has a price-to-earnings ratio of 20.66, significantly lower than the 33 times seen in the US market [18]. Investment Outlook - The article emphasizes that the core investment value in the AI sector is driven by genuine demand and supported by financially healthy companies, marking the beginning of a technological revolution [24]. - The potential for growth in the Chinese AI market is substantial, particularly in consumer applications, which are still in their infancy compared to the US market [23]. - Investors are encouraged to focus on the long-term value of AI applications rather than short-term market fluctuations, with a particular emphasis on companies with strong fundamentals and financial health [25].

Core Viewpoint - The article discusses the current volatility in the AI sector, highlighting the contrasting views on whether there is a bubble similar to the 2000 internet crash, while emphasizing the resilience of the Hang Seng Internet Technology Index amidst market fluctuations [1][3][4]. Market Performance - The Hang Seng Internet ETF (513330) tracking the Hang Seng Internet Technology Index surged by 3.88%, making it the best-performing tech index in Hong Kong [1]. - The Hang Seng Internet Technology Index has experienced a cumulative decline of 2.25% since November 17, marking it as the least affected tech index in Hong Kong [1]. AI Sector Sentiment - There are two opposing perspectives in the market: one side believes AI is nearing a bubble similar to the 2000 internet crash, while the other sees it as a final opportunity for latecomers to invest [3][4]. - The debate centers around the quality of companies and their valuations, with comparisons drawn to the unsustainable valuations of internet companies in 2000 [4]. Financial Metrics - The "M7" companies, which significantly contribute to the Nasdaq's rise, have substantial revenues exceeding $100 billion and profits in the hundreds of billions, indicating strong financial backing [5][6]. - The capital expenditure to operating cash flow ratio for "M7" stands at 49%, lower than the 56% during the internet bubble peak, suggesting a healthier financial environment [6][7]. Valuation Comparisons - The overall valuation of "M7" companies is around 33 times earnings, which is lower than the peak of 60 times during the internet bubble, indicating a more rational market [8]. - The Hang Seng Internet Technology Index has a PE (TTM) of 20.66, significantly lower than the 33 times of the US market, reflecting a valuation gap and potential investment opportunities [14]. Domestic AI Developments - China's AI sector is also experiencing a downturn, but recent breakthroughs, such as Alibaba's AI assistant achieving over 10 million downloads, indicate strong growth potential [9][10]. - The capital expenditure of Chinese tech giants is projected to turn positive, with a significant increase from -9.5% in 2024 to 18% by March 2025 [9]. Growth Potential - The AI revenue growth rate for major Chinese internet cloud companies has increased from 11.7% at the end of 2024 to 23.2% by the second quarter of 2025, marking a shift towards commercialization [19]. - The current focus of AI applications in China is primarily in the B2B sector, with significant potential for growth in consumer applications in the future [19]. Investment Outlook - The core investment value in the AI sector is driven by genuine demand, supported by financially healthy companies that are beginning to generate revenue [20]. - The ongoing technological revolution in AI is expected to yield compounding returns over the next 10-20 years, emphasizing the importance of focusing on long-term value rather than short-term market fluctuations [21].

Group 1 - Huadian Co., Ltd. plans to acquire a 15% stake in Shengwei Technology along with patents and technical assets [1] - Huarong Chemical has not yet introduced customers in the photolithography or photolithography machine sectors [1] - Wenfeng Co., Ltd. faces an investigation with its deputy general manager Qin Guofen being placed under detention [1] Group 2 - The Ministry of Housing and Urban-Rural Development is working on improving urban health assessment systems and integrating urban health checks with urban renewal [2] - The Ministry of Commerce reported that foreign investment in China from January to October reached 621.93 billion yuan, a year-on-year decrease of 10.3% [2] - Beijing is set to revise the sales contracts for commercial housing for the first time in eight years [2] - The company "Dark Side of the Moon" is expected to initiate a new round of financing with a valuation of 4 billion USD and plans for an IPO next year [2] - Innovent Biologics has been included in the Hang Seng Index, while Leap Motor has been added to the Hang Seng Technology Index [2]

周五，美国三大股指全线收涨，家得宝涨超3%，默克涨近3%，领涨道指。 芯片股多数上涨，费城半导体指数涨0.86%，德州仪器涨逾3%，恩智浦半导体涨超3%，微芯科技涨逾 3%，美光科技涨超2%。跌幅方面，博通跌逾1%，阿斯麦跌超1%，超威半导体跌逾1%。 中概股多数上涨，纳斯达克中国金龙指数涨1.23%，万得中概科技龙头指数涨1.08%。个股方面，涂鸦 智能涨超6%，阿特斯太阳能涨逾5%，亚朵涨近5%，老虎证券涨超4%，蔚来涨超3%；爱奇艺跌超2%， 小马智行跌逾1%，大全新能源跌超1%，再鼎医药跌近1%。 此外，国际贵金属期货收盘涨跌不一，COMEX黄金期货涨0.07%报4062.8美元/盎司，本周累计下跌 0.77%；COMEX白银期货跌1.27%报49.66美元/盎司，本周累计下跌2.02%。 美联储官员近期释放出对货币政策的分歧信号，副主席杰斐逊强调当前金融体系稳健且具韧性，认为 AI热潮与互联网泡沫时期不同，企业拥有实际收益且杠杆使用有限。与此同时，美联储理事米兰称数 据影响"明显偏向鸽派"，劳动力市场数据不如预期强劲，暗示可能支持降息。 截至收盘，道指涨1.08%报46245.41点，标普500指数 ...

这些企业迄今尚未严重依赖债务融资，可能会降低当AI市场情绪转变时，通过信贷市场向 经济体传导更广泛风险的程度。 不过他也警示，AI行业的杠杆率可能会上升，一旦AI情绪转变，损失也会增加。他将密切关注这一发 展趋势。 美联储副主席Philip Jefferson表示，当前人工智能相关股票的涨势不太可能重演上世纪90年代末互联网 泡沫破裂的情形，主要原因在于AI企业已建立起实际盈利能力，且金融体系保持稳健。 周五，Jefferson在克利夫兰联储会议上发表讲话称，与当年投机性质浓厚的互联网公司不同，当前AI相 关企业是成熟企业，拥有相对成熟且在增长的盈利来源。他指出： Jefferson强调，AI行业迄今对债务融资的依赖有限。 他认为，这种有限的杠杆使用"可能会降低当AI市场情绪转变时，通过信贷市场向更广泛经济体传导风 险的程度"。 然而，未来的趋势值得警惕。Jefferson援引一些分析师的预测称，未来对AI基础设施的投资将需要更 多债务融资。 AI企业有盈利，估值更低 Jefferson详细阐述了当前AI热潮与互联网泡沫时代的三个关键区别。 首先，也是最核心的区别在于企业的基本面。他指出，与90年代末许多仅凭 ...

Core Insights - The current AI hype is being compared to the internet bubble of the late 1990s, with Goldman Sachs highlighting five warning signals that resemble those seen before the internet bubble burst [5][10]. Group 1: Warning Signals Identified by Goldman Sachs - The first signal is peak investment spending, with capital expenditures for the five major tech giants projected to reach $533 billion by 2026, up from 3% of GDP in 1995 to 4.5% in 2000 [5]. - The second signal indicates a decline in corporate profits, with macro-level profit growth showing signs of fatigue despite stable profit margins [5]. - The third signal is the rapid rise in corporate debt, exemplified by Meta raising $30 billion through bond issuance to support its AI spending plans [6]. - The fourth signal is the Federal Reserve's interest rate cuts, with a 25 basis point cut in October and expectations for another cut in December [7]. - The fifth signal is the widening credit spread, which has increased from 2.76% to 3.15% for U.S. high-yield bonds, indicating a trend despite still being at low levels [8]. Group 2: Comparison with the Internet Bubble - Unlike the internet bubble, the current AI hype does not exhibit key macroeconomic imbalances, as corporate profitability remains solid and financial health is relatively stable [10]. - The net profit margin for the AI industry stands at 27.7%, contrasting sharply with the 14% profitability of ".com" companies during the internet bubble [10]. - Valuation metrics differ significantly, with the forward P/E ratio for the Nasdaq 100 at 26.7 compared to 60 during the internet bubble [10]. Group 3: Historical Lessons from the Internet Bubble - The internet bubble saw 68% of Nasdaq tech stocks unprofitable, while 72% of private AI companies are currently unprofitable [11]. - The average valuation for private AI companies is 35 times sales, compared to 28 times during the internet bubble [11]. - Companies that survived the internet bubble, like Amazon, adapted through innovation and diversification, highlighting the importance of sustainable business models [13]. Group 4: Current AI Investment Landscape - Despite significant investments in generative AI, 95% of organizations report negligible returns, indicating a disconnect between investment and actual benefits [15]. - Many companies treat AI as a plug-and-play tool, failing to recognize its need for continuous learning and adaptation [15]. - Investors are advised to use options for risk management while being cautious of the dual risks in the interest rate market [17]. Group 5: Conclusion on AI Investment - AI is recognized as a major technological shift, but not all participants will benefit, as evidenced by the 95% of AI investments yielding no returns [19]. - Companies that thrive in the AI landscape will likely share traits such as continuous technological advancement, sustainable profit models, and robust governance [19][20]. - The current AI wave presents opportunities, but investors should remain vigilant and selective in their investments [20][21].

11月21日，美联储副主席菲利普‧杰斐逊表示，人工智能的崛起与1990年代末的互联网泡沫有着明显的 不同。杰斐逊表示："当今市场与AI技术联系最紧密的公司，普遍拥有相对成熟且在增长的盈利来 源。"他指出，相比之下，1990年代末和2000年代的许多互联网公司"几乎没有已实现盈利，收入前景也 只是投机性的；即便如此，它们仍在投资者热情高涨的氛围下相对容易地获得外部融资并在公开市场上 市"。他补充称，以市盈率衡量，AI相关公司的估值迄今仍"远低于"当年互联网公司的峰值水平。 ...

美联储 副主席 杰斐逊称他认为人工智能的崛起与20世纪90年代末的互联网繁荣之间有明显的区别。杰 斐逊在为周五在克利夫兰联储举行的活动准备的讲话中表示：相比之下许多上世纪90年代末期和2000年 代的互联网公司"几乎没有实现利润只有投机性的收入前景即使它们获得了外部资金，并且在投资者热 情高涨的情况下相对轻松地在交易所上市"杰斐逊补充说以市盈率衡量的人工智能相关公司的估值迄今 为止仍远低于互联网公司的峰值水平。他说在互联网泡沫繁荣的顶峰时期有超过1,000家公司以互联网 公司的身份上市而如今有大约50家专注于人工智能的上市公司。杰斐逊还表示现在判断人工智能的崛起 对美联储货币政策决策的影响可能还为时过早。杰斐逊表示金融体系仍然稳健且富有弹性。 来源：滚动播报 ...