市场策略丨深度报告 [Table_Title] AI 革命和泡沫分析框架（一）：AI 的 1998—— 科网泡沫再审视 %% %% %% %% research.95579.com 1 丨证券研究报告丨 报告要点 [Table_Summary] 本文对科网泡沫复盘，希望以史为鉴，对当下 AI 产业演进过程提出一些思考。衡量泡沫是否存 在主要有八大维度，可以分为三个层面。宏观：叙事催化剂和杠杆水平；产业：故事性、应用 场景、未来竞争格局、商业模式；市场：经验不足的投资者、纯粹可投资标的。从宏观环境整 体来讲，美国当前的经济环境不及科网泡沫时期，流动性相比当时更好。产业端来讲，商业模 式能否闭环，企业能否产生最终盈利是市场最为关注的问题。而市场端来看，当前估值水平和 市场构成相比科网泡沫时期更优。 分析师及联系人 [Table_Author] 戴清 SAC：S0490524010002 SFC：BTR264 请阅读最后评级说明和重要声明 2 / 26 %% %% %% %% research.95579.com 2 [Table_Title AI 革命和泡沫分析框架（一 2] ）：AI 的 1998—— 科网泡沫 ...

Core Viewpoint - The article discusses the escalating competition among global tech giants for electricity supply, driven by the increasing energy demands of AI development, highlighting China's significant advantage in power generation capacity compared to the U.S. [5][6][12] Group 1: Electricity Demand and Supply Challenges - Tech giants are entering a "power grab" as AI's energy needs surge, with Meta, Microsoft, and Google forming partnerships with nuclear power companies to secure long-term electricity contracts [8][9][11]. - The U.S. energy consumption is projected to reach a historical high of 29,000 terawatt-hours by 2026, while AI data centers are expected to consume 20%-25% of U.S. electricity by 2030 [5][12]. - The average annual electricity growth rate is only 2%-4%, creating a significant supply-demand gap as AI's energy requirements grow exponentially [12][14]. Group 2: U.S. Energy Infrastructure Issues - The lengthy approval process for new energy projects in the U.S. can take 7-10 years, with over 50% of projects being rejected, leading to delays in meeting energy demands [19][20]. - Aging infrastructure, with over 70% of transmission lines built before the 1970s, exacerbates the energy supply issues, particularly in tech-heavy regions like California and Texas [20][21]. - The conflict between tech companies' energy consumption and local residents' electricity costs has led to significant price increases, with some areas seeing a 200% rise in electricity bills [22][24]. Group 3: China's Energy Advantage - China is projected to add approximately 470 gigawatts of new power generation capacity by 2025, seven times that of the U.S., which is only expected to add 64 gigawatts [26][28]. - The establishment of a nationwide power grid in China allows for efficient energy distribution, overcoming geographical limitations and ensuring a stable supply for AI development [29][30]. - The cost of electricity in China's western regions is significantly lower than in the U.S., providing a competitive edge for Chinese AI companies in terms of operational costs [31][32]. Group 4: Strategic Importance of Energy in AI Competition - The article emphasizes that energy supply is becoming a strategic resource in the AI competition, with countries needing robust energy systems to support technological advancements [37][38]. - The geopolitical implications of energy resources are highlighted, with the U.S. engaging in actions to secure energy supplies globally, reflecting the critical role of energy in future technological competition [40].

Core Insights - Apple's AI division is experiencing significant turbulence, with a notable exodus of talent, particularly from its AI ranks, as the company struggles to keep pace with competitors in the artificial intelligence sector [2][12] - The decision to outsource some technology to Google has caused dissatisfaction among staff, contributing to the ongoing talent drain [2][9] Departures and Talent Loss - Recent exits from Apple include Yinfei Yang, Haoxuan You, Bailin Wang, and Zirui Wang, with Yang starting a new company and You and Wang joining Meta [1][12] - Zirui Wang and Apple executive Stuart Bowers have joined Google DeepMind, which is assisting Apple in developing core AI models for new features, including an upgraded Siri [3][12] Organizational Changes - Apple's AI efforts underwent a major reorganization last year, with CEO Tim Cook relieving AI chief John Giannandrea and assigning responsibilities to software head Craig Federighi [6][12] - The Foundation Models (AFM) team, responsible for the underlying technology of Apple's AI platform, has faced scrutiny due to delays in new Siri features and a lackluster reception of current AI offerings [7][12] Product Development - Apple is preparing two new versions of Siri: a near-term update utilizing personal data for queries and a more ambitious chatbot-style overhaul set for later this year, both powered by models developed by Google's team [9][12] - The company continues to rely on its own models for on-device Apple Intelligence features, although it is unlikely to depend on external partners indefinitely due to competitive pressures in the AI landscape [11][12] Financial Performance - Despite reporting over $85 billion in iPhone sales, Apple's AI challenges have contributed to a stock slump this year, complicating efforts for a turnaround [5][12]

法庭文件显示，历经五年多激烈且艰难的诉讼，才终于形成一份和解方案。2020年，原告用户诉称，在谷歌应用程序被关闭、位置共享功能被禁用、屏幕被 锁定等情况下，谷歌仍能通过对安卓用户手机进行特殊设置，收集其移动蜂窝数据。 值得关注的是，与类似案件通常主张侵犯用户隐私不同，原告律师格伦·萨默斯（Glen Summers）认为，谷歌非法收集用户数据并用于产品研发和个性化广 告推送的行为构成了"非法侵占财产"，即一方未经允许占有他人财产。萨默斯表示，一旦和解协议达成，此案将成为史上和解金额最高的"侵占财产案"。 和解协议提到，谷歌将支付1.35亿美元（折合人民币约9.4亿元），赔偿对象包括自2017年11月12日起，所有使用搭载安卓系统移动设备的用户，每位成员的 赔偿金上限为100美元。据第三方公司的估测，美国安卓系统用户约占整个移动市场的40%。 和解协议明确了针对谷歌的一系列整改措施。包括停止被用户指控的非法数据收集和传输行为；修改谷歌应用商店的服务条款，披露数据收集的相关情况， 并为用户提供数据传输"一键关闭"按钮；在修改安卓系统手机初始设置时，必须获得用户的明确同意等。 1月29日，谷歌同意支付1.35亿美元解 ...

Group 1: Google Project Genie - Google DeepMind has launched Project Genie, an AI tool that allows users to create interactive virtual worlds using text prompts or images [1] - The tool is currently available to users with a Google Ultra account in the U.S. who are 18 years or older [1] - Project Genie is based on the previously showcased Genie 3 model and integrates Google's Nano Banana Pro image generation model and Gemini multimodal model [1] Group 2: Apple AirPods Pro 3 Demand - Apple CEO Tim Cook stated that the demand for AirPods Pro 3 has exceeded the company's expectations since its release in September 2025 [3] - The "Wearables, Home, and Accessories" category saw a revenue decline of approximately 2% year-over-year in Q1 of fiscal year 2026 [3] - Cook clarified that the revenue drop does not reflect true market demand, attributing it to production constraints affecting AirPods Pro 3 [3] Group 3: Apple Siri Development - Apple initially considered using Anthropic's Claude model for the new Siri but ultimately chose Google's Gemini platform due to cost negotiations [5][7] - Anthropic's proposal included high fees, demanding tens of billions annually from Apple, which led to the decision to switch to Gemini [7] - Despite not being selected for Siri, Anthropic continues to play a significant role in Apple's internal systems [7] Group 4: Xiaomi's UK Expansion - Xiaomi plans to open 150 stores in the UK over the next four years, having generated approximately 1 billion RMB in revenue last year [10] - The company has diversified its product lines, including smartphones, wearables, home appliances, and vehicles [10] Group 5: Tencent's AI Research - Tencent has appointed Pang Tianyu as the Chief Research Scientist for its AI division, focusing on multimodal reinforcement learning [12] - Pang has a strong academic background in machine learning and has published multiple papers in top conferences [12] Group 6: Apple iPhone 18 Pro Features - Apple is negotiating with SpaceX to integrate Starlink satellite communication into the upcoming iPhone 18 Pro, allowing users to connect without additional hardware [14] - Currently, Apple offers emergency SOS satellite services through Globalstar, but this is limited to urgent situations [14] Group 7: Alibaba's DeepPlanning - Alibaba Qianwen has launched a new benchmark test called DeepPlanning, which focuses on complex real-world planning tasks for AI [13] - Current top AI models still show limitations in global optimization and long-term consistency, indicating a gap in achieving full autonomous decision-making capabilities [13] Group 8: Robotics and AI Insights - Yu Shu Technology's CEO Wang Xingxing emphasized the importance of developing large models for robots, suggesting that success in this area could lead to significant recognition [17] - The company reported a production output of over 5,500 humanoid robots in 2025, with plans for further advancements [18]

Core Viewpoint - The announcement of Google's "Project Genie," an AI prototype capable of generating interactive virtual worlds, has led to significant declines in the stock prices of major gaming companies, raising concerns about potential disruptions in the gaming development industry [1][3]. Group 1: Market Reaction - Major gaming companies experienced substantial stock price drops, with Unity Software falling over 24%, Roblox down approximately 13%, and Take-Two Interactive decreasing nearly 8% [3]. - International gaming companies, including Nexon, CD Projekt, Ubisoft, Nintendo, and Konami, also saw their stock prices affected by the news [3]. Group 2: Project Genie Overview - "Project Genie" allows users to create virtual worlds by describing scenes, characters, and perspectives, and it simulates physics and interactions in real-time [3]. - The prototype is currently available to AI Ultra subscription users in the U.S. and is seen as a potential game-changer for the gaming development model [3]. Group 3: Industry Implications - Concerns exist that if AI can significantly simplify world-building, it may impact the demand for professional game engines like Unreal Engine and Unity Engine [4]. - Platforms like Roblox, which rely on user-generated content, may lose their appeal if users can easily create worlds using AI [4]. - Project Genie could shorten development cycles and reduce costs, as some AAA games currently take five to seven years and cost hundreds of millions to produce [4]. Group 4: Analyst Perspectives - Analysts from Wells Fargo maintain that the introduction of AI tools will accelerate game development, which is a long-term positive for the industry [5]. - Unity's CEO believes that advancements in "world models" will serve as a powerful accelerator for video game creation rather than a threat [5]. - The application of AI in game development is becoming a trend, with nearly 90% of game developers reportedly using AI tools in their work [5].

Core Viewpoint - Google has officially launched the experimental research prototype Project Genie, which allows users to create and interact with 3D worlds using AI technology [1][17]. Group 1: Project Genie Overview - Project Genie is an experimental research prototype that separates the core capabilities of Genie 3, combining features from Genie 3, Nano Banana Pro, and Gemini into a web application [18][19]. - The main functionalities of Project Genie include the ability to "build" worlds using text or images, "enter" generated worlds for exploration, and "modify" existing worlds based on prompts [20][22][24]. Group 2: User Experience and Community Engagement - Users have quickly engaged with Project Genie, showcasing their creativity by generating various 3D models and scenes, such as a flying simulator and a realistic wolf hunting in a jungle [26][38]. - The platform allows for high levels of detail and interaction, with users able to create dynamic environments and characters that respond to user inputs [40][41]. Group 3: Community Feedback and Limitations - While many users praised Project Genie for its capabilities, some expressed disappointment with the model's performance in handling specialized content like CAD [44][45]. - The project is still in its experimental phase, indicating that it may not yet be a fully mature and stable AI tool [47].

Core Viewpoint - The announcement of Google's "Project Genie," an experimental AI prototype capable of generating interactive virtual worlds, has led to significant declines in the stock prices of major gaming companies, raising concerns about potential disruptions in the gaming development industry [1][2]. Group 1: Market Reaction - Major gaming companies experienced substantial stock price drops, with Unity Software falling over 24%, Roblox down approximately 13%, and Take-Two Interactive decreasing nearly 8% [1]. - International gaming companies, including Nexon, CD Projekt, Ubisoft, Nintendo, and Konami, also saw their stock prices affected by the news [1]. Group 2: Project Genie Overview - "Project Genie" allows users to create virtual worlds through descriptions of scenes, characters, and perspectives, and it can simulate physics and interactions in real-time [2]. - The technology is currently available to AI Ultra subscription users in the U.S. and is expected to change the game development landscape, compelling developers to adapt to rapidly evolving technologies [2]. Group 3: Industry Implications - Concerns exist that if AI can significantly simplify world-building, it may reduce the demand for professional game engines like Unreal Engine and Unity Engine [2]. - Platforms like Roblox, which rely on user-generated content (UGC), may see diminished attractiveness if users can easily create worlds using AI [2]. - The potential for AI to shorten development cycles and reduce costs is notable, as some AAA games currently take five to seven years and cost hundreds of millions to produce [2]. Group 4: Analyst Perspectives - Wells Fargo analysts maintain that the introduction of AI tools will accelerate development in the gaming industry, viewing it as a long-term positive [3]. - Unity's CEO expressed that advancements in "world models" will serve as a powerful accelerator for video game creation rather than a threat [3]. - The application of AI in game development is seen as an inevitable trend, with nearly 90% of game developers reportedly using AI tools in their work [3].

Core Insights - The market reacted positively to Meta Platforms' quarterly results, while Microsoft and Tesla's December-quarter numbers disappointed investors [1] Group 1: Company Performance - Microsoft reported a +28.1% earnings growth and +16.7% revenue growth, exceeding estimates, but faced concerns over Azure's +38% revenue growth and underwhelming guidance [2][3] - Meta's Q4 earnings and revenues increased by +9.3% and +23.8%, respectively, but highlighted margin pressures; however, its effective use of AI in advertising improved click rates by +3.5% and conversion rates by +1% [4] - Meta plans to increase its capital expenditure to $135 billion for the year, up from $72 billion in 2025 and $39 billion in 2024, citing capacity constraints as a reason for the increase [5] Group 2: Market Trends - The Mag 7 group, which includes major tech companies, is projected to see Q4 earnings up +21.9% year-over-year with +18.1% higher revenues, although individual contributions vary significantly [7] - As of now, 167 S&P 500 members have reported Q4 results, showing a +13.1% increase in earnings year-over-year on +7.6% higher revenues, with 77.8% beating EPS estimates [9][19] - The Mag 7 group is expected to account for 25.2% of all S&P 500 earnings in 2025, up from 23.2% in 2024 and 18.3% in 2023, indicating a strengthening earnings outlook [14]

Group 1 - Eli Lilly's stock performance is influenced more by new reports on GOP-1 clinical trials than by its earnings [1] - Alphabet is highlighted as a strong performer, with significant contributions from its products like Gemini, YouTube, and self-driving cars [2] - The healthcare sector is facing scrutiny from the president, particularly targeting companies in the healthcare chain [3] Group 2 - The drug distribution sector is criticized for high profits with minimal contributions to healthcare, with McKesson being implied as a key player [4] - Amazon's stock has become volatile, with earnings reports leading to significant fluctuations in its stock price [5] - The travel sector is mentioned as a focus for investment, despite recent underperformance [6]