凭借在精密陶瓷领域的技术壁垒， TOTO 已卡位 AI 算力背后的关键一环 ——NAND 存储芯片制造，且该业务已贡献了公司 40% 的营业利润。 硬·AI 作者 | Kozmon 编辑 | 硬 AI 在东京股市的传统板块中，TOTO（东陶）长期被视为卫浴行业的防御性资产。但在伦敦激进对冲基金 Palliser Capital眼中，这家拥有百年历史的"马桶大王"，是一只被严重误读且低估的AI存储芯片概念股。 01 从烧制马桶到生产晶圆 这家公司最不为人知的杀手锏是"静电卡盘"。在半导体制造环节，这是一种利用静电力吸附并固定硅晶圆 的关键组件。随着AI浪潮推动NAND存储芯片向更高层数、更复杂结构演进，制造工艺中对"低温蚀刻"的 需求激增。 这正是TOTO的护城河所在。利用自1980年代以来在卫浴制造中积累的精密陶瓷技术，TOTO生产的静电 卡盘能够在极低温度下保持极高的稳定性。Palliser指出， TOTO 拥有长达五年的竞争优势，短期内竞争 对手难以望其项背。 尽管TOTO早已涉足此领域，但直到近年AI需求爆发，这项业务才从边缘走向舞台中央。Palliser强调， TOTO已悄然从一家传统的卫浴冠军，进化 ...

作者 | Kozmon 硬·AI 编辑 | 硬 AI Fintool 创始人 Nicolas Bustamante 最近在 X 平台上发了一篇"杀人诛心"的深度长文，直接点破了最近软 件股万亿市值蒸发背后的残酷真相。 作为一位曾经打造过欧洲最大法律科技平台（Doctrine）、现在又投身 AI 金融（Fintool）的"双栖"创业 者， 他站在新旧时代的交界点上，详细拆解了垂直 SaaS 行业赖以生存的十大护城河是如何被大模型一 一瓦解的。 Nicolas认为，LLM（大语言模型）正在系统性地拆除垂直软件过去赖以生存的护城河，以前靠"软件难 用"和"流程复杂"赚取高昂溢价的日子结束了，市场正在经历一场残酷的价值重估。 我们给大家简单划了下这篇文章的重点： 1."难用"不再是护城河 LLM 正在系统性地拆除垂直软件过去赖以生存的护城河，以前靠 " 软件难用 " 和 " 流程复杂 " 赚取高昂溢价的日子结束了，市 场正在经历一场残酷的价值重估。 以前像彭博终端这种软件，最牛的护城河其实是「难用」，用户花了很长时间学会了那些复杂的快捷键和 代码，学会了就不想换。但现在，LLM把所有复杂的界面都坍缩成了一个聊天框，用 ...

Core Viewpoint - Tencent's recent initiatives, including the distribution of 1 billion yuan in red envelopes and the NBA All-Star Game live streaming, are not merely promotional tactics but strategic moves to dominate the "context" landscape in AI [5][24]. Group 1: Product Overview - The "Yuanbao Pai" product is more than just a chat room; it serves as a "super container" for context, integrating various media and social interactions into a single AI-driven platform [8][11]. - Unlike traditional applications that operate in silos, Yuanbao Pai combines resources from Tencent's video, music, and sports platforms, allowing AI to understand user interactions across different contexts [11][21]. Group 2: AI Context Learning - Current AI models struggle with complex contexts, achieving only a 23.7% accuracy rate when faced with new and intricate scenarios, as highlighted in the "CL-bench" paper by Tencent's AI team [15][16]. - The challenge lies in AI's reliance on past knowledge, which often leads to misunderstandings in real-time interactions, necessitating a shift towards context learning rather than mere memorization [18][20]. Group 3: Strategic Implications - The integration of AI into high-frequency usage scenarios, such as social interactions and content consumption, is seen as a sustainable competitive advantage for Tencent [19][21]. - By leveraging user interactions within Yuanbao Pai, Tencent aims to gather valuable reinforcement learning data to improve AI's contextual understanding, potentially increasing accuracy from 23.7% to much higher levels [25][26].

若短缺持续至2027年，价格仍有上涨空间。 作者 | 龙 玥 编辑 | 硬 AI 随着存储巨头们四季度财报的披露，一个清晰的信号正在释放：存储行业不仅走出了低谷，更在AI浪潮的推动下，开启了一轮强度罕见的"超级周期"。 本周，美银美林团队总结了存储巨头财报电话会议的精华，并结合韩国半导体展（Semicon Korea）的一线见闻，指出当前市场正处于库存极低、价格 飙升且资本开支大幅扩张的强劲上升期。 美银美林指出，SK海力士库存周转天数降至127天的低位，成品库存仅剩2-3周。三星DRAM售价环比大涨40%，现货价格创25年新高。行业高管认 为，本轮周期强度已超越2017-18年的云繁荣期，若短缺持续至2027年，价格仍有上涨空间。 他进一步解释道，与高度定制化的晶圆代工不同，存储器仍被视为大宗商品（遵循JEDEC标准），这意味着如果短缺持续到2027年，价格还有进一步 上涨的空间。 美银的"存储指标"（Memory Indicator）也佐证了这一点。该指标在12月已回升至124的"上行周期"水平，而2025年上半年的平均值仅为103。 01 财报季"五大关键信号" 美银美林分析师Simon Woo团队在研 ...

Core Viewpoint - Alibaba's Qwen 3.5 model represents a significant leap in AI architecture, emphasizing efficiency and performance over sheer parameter size, positioning itself as a leading open-source model in the industry [3][19][32]. Group 1: Model Performance and Architecture - Qwen 3.5 features a total of 397 billion parameters, activating only 17 billion during inference, resulting in a 60% reduction in deployment memory usage and a 19-fold increase in inference throughput compared to its predecessor [4][20]. - The model's API pricing is set at 0.8 yuan per million tokens, making it significantly cheaper than competitors like Gemini 3 Pro, which is 18 times more expensive for similar performance [7][20]. - The model's architecture incorporates a mixed expert framework, allowing for dynamic attention allocation and efficient processing of long texts, enhancing both efficiency and accuracy [21][22]. Group 2: Multi-Modal Capabilities - Qwen 3.5 evolves from a language model to a native multi-modal model, capable of understanding and integrating text, visuals, and audio seamlessly, unlike many existing multi-modal solutions that rely on separate modules [11][12]. - The model's training involves joint learning from mixed data types from the outset, enabling it to understand deep semantics from images and construct corresponding visuals from text [12][13]. - This native integration allows for advanced capabilities such as pixel-level visual localization and understanding complex video content over extended durations [15][18]. Group 3: Market Position and Ecosystem - Alibaba's strategy includes a dual approach of releasing state-of-the-art models while maintaining an open-source ecosystem, allowing developers worldwide to access and utilize these models freely [24][30]. - The company has established a significant presence in the AI cloud market, with a projected market share increase from 33% to 36% by 2025, driven by the demand for AI-related products [26][27]. - Recent financial reports indicate a 34% year-over-year growth in Alibaba Cloud's public cloud revenue, with AI-related product revenues maintaining triple-digit growth for nine consecutive quarters [28]. Group 4: Industry Impact - The launch of Qwen 3.5 signifies a paradigm shift in the AI industry, moving from high-cost, high-complexity models to more accessible and efficient solutions that democratize AI technology [31][32]. - The model's success is expected to redefine industry standards, making AI a productivity tool available to a broader audience, thus reshaping the global AI landscape [32].

Core Viewpoint - The rapid evolution of AI technology is set to disrupt white-collar jobs significantly within the next one to five years, necessitating immediate adaptation by professionals to integrate AI into their workflows and develop irreplaceable skills such as critical thinking [2][5][16]. Group 1: AI's Evolution and Impact - AI has transitioned from being an "assistive tool" to an "independent executor," capable of completing complex tasks autonomously in a fraction of the time previously required [3][4]. - The ability of AI to self-iterate and improve its own systems marks a critical turning point, breaking the limitations imposed by human researchers and leading to exponential growth in capabilities [4][9]. - The speed of AI advancements is accelerating, with significant improvements occurring within months rather than years, creating a narrowing window for professionals to adapt [6][8][42]. Group 2: Employment and Professional Adaptation - The nature of work across various sectors, including law, finance, and creative industries, is being fundamentally reshaped, with repetitive and standardized tasks being automated [10][11][43]. - Professionals are encouraged to engage in scenario-based learning, embedding AI into daily workflows to gain practical experience and understanding of its boundaries [12][13]. - Maintaining relevance in the job market requires a focus on developing core competencies that AI cannot easily replicate, such as strategic judgment and interpersonal skills [5][12]. Group 3: Societal and Economic Implications - The AI revolution is redefining wealth distribution, educational foundations, and occupational structures, leading to a "winner-takes-all" effect in various industries [5][14]. - The philosophical implications of AI's capabilities challenge traditional notions of work and human value, as machines increasingly perform tasks previously thought to require human judgment [14][44]. - The potential for AI to solve complex global issues, such as medical research, is immense, but it also poses significant risks if not managed responsibly [54][55].

Core Viewpoint - The CEO of Anthropic, Dario Amodei, predicts that by 2026-2027, AI will evolve into a "Country of Geniuses in a Datacenter," with intelligence comparable to thousands of Nobel laureates working together [2][8][9] - Anthropic is experiencing a staggering annual revenue growth of 10 times, expecting to reach $10 billion by 2025, driven by advancements in AI capabilities [2][11] Group 1: AI Growth and Predictions - Amodei asserts that AI is nearing the end of its exponential growth phase, with significant qualitative changes expected in the next 2-3 years [5][6] - The transition from "smart high school student" to "professional-level" AI models has been rapid, with improvements in programming and mathematical capabilities [6][8] - Amodei expresses high confidence in achieving the vision of a genius AI nation within the next decade, citing a 90% certainty for a 10-year timeline and a 50/50 chance for the next 1-2 years [9][42] Group 2: Revenue Growth and Financial Strategy - Anthropic's revenue trajectory is described as "bizarre 10x per year growth," with projections of $1 million in 2023, $10 million in 2024, and $9-10 billion in 2025 [11][12] - Amodei explains the cautious approach to capital investment in computing power, emphasizing the need for revenue growth to align with capacity expansion to avoid bankruptcy risks [13][14] Group 3: AI in Software Engineering - Amodei outlines three stages of AI evolution in software engineering, with the first stage already achieved where models write 90% of code lines [16][50] - The second stage will see models handling 90% of end-to-end tasks, while the third stage will involve models taking over complex engineering tasks [18][53] - The expectation is that AI will significantly enhance productivity in software engineering without leading to mass unemployment among engineers [20][54] Group 4: Challenges and Future Developments - Amodei acknowledges potential geopolitical risks and societal upheavals as variables that could impact the timeline for achieving advanced AI capabilities [9][13] - The company is actively researching continuous learning capabilities for AI, which may be realized in the next couple of years [108][109] - There is an ongoing discussion about the efficiency of AI in learning and adapting compared to human learning processes, with a focus on the need for models to achieve a level of contextual understanding [100][101]

01 高盛合伙人Rich Privorotsky认为，市场陷入"先卖出、后提问"恐慌，本质是对企业护城河的全面检验，建议关注拥有真 正护城河的公司、实物资产和工业股，看好航空航天板块，但需警惕银行股风险。高盛预计CTA将抛售15-20亿美元美 股，标普若跌破中期阈值6723点，将加速抛售幅度。 硬·AI 作者 | 鲍亦龙 编辑 | 硬 AI 高盛合伙人Rich Privorotsky警告称，一场由人工智能驱动的"创造性破坏"正实时席卷全球各行业，本质上 这是一次对企业护城河的全面检验。 从上周软件行业遭遇冲击，到本周初先是保险和财富管理类股，下半周则轮到房地产服务类和物流板块。 AI最初被视为对股市的利好因素，但现在正在激进地检验哪些企业真正具有可防御的竞争优势。 "先卖出、后提问"的情绪在市场扩散，抛售速度加快，但除了AI担忧外并无明确催化剂。高盛合伙人Rich Privorotsky认为这是一次护城河检查： 企业的业务是否能抵御技术冲击？如果有一支机器人大军，能否颠覆现有企业？企业是否必须竞相 投入或收购，否则就会被取代？ Privorotsky进一步强调，需警惕美国各大股指中的CTA（商品交易顾问）触发 ...

分析认为，在现实世界复杂任务中， 由于大规模推理与长链路生成将消耗大量token，豆包2.0的成本优 势将成为关键竞争力 。这标志着字节跳动在大模型商业化应用上迈出重要一步。 01 多模态能力达到世界顶尖水平 豆包2.0全面升级了多模态能力，在视觉推理、感知能力、空间推理与长上下文理解等任务上表现突出。 字节发布豆包2.0，旗舰版Pro全面对标GPT-5.2与Gemini 3 Pro。新模型在多模态、数学及编程等领域达到业界顶尖， 同时将推理成本降低约一个数量级，显著提升Agent应用性价比。目前已接入豆包App、TRAE及火山引擎API。 硬·AI 作者 | 董 静 编辑 | 硬 AI 字节跳动旗下豆包大模型正式进入2.0阶段，推出面向Agent时代的系统性升级版本。 新版本在保持与 GPT-5.2和Gemini 3 Pro相当性能的同时，将推理成本降低约一个数量级 ，为大规模生产环境下的复杂任 务执行提供更具竞争力的解决方案。 2月14日，字节跳动宣布，豆包2.0系列包含Pro、Lite、Mini三款通用Agent模型和专门的Code模型。 其 中旗舰版豆包2.0 Pro全面对标GPT-5.2与Gemin ...

Core Viewpoint - Microsoft AI's chief executive warns that a majority of white-collar jobs may be automated within the next 12 to 18 months, a timeline that is significantly earlier than the expectations of the business community and policymakers [1][4]. Group 1: AI Impact on Employment - The report from Challenger indicates that in January 2023, 7,624 job losses were attributed to AI, accounting for 7% of total layoffs that month. By 2025, the total number of layoffs linked to AI is projected to reach 54,836 [1][4]. - Since the beginning of 2023, a total of 79,449 planned layoffs have been attributed to AI [1]. Group 2: Training AI with Human Labor - The startup Mercor has employed thousands of white-collar contractors, including professionals from fields such as medicine, law, finance, and engineering, to train AI systems that may eventually replace them. These contractors earn between $45 to $250 per hour [5]. - This model highlights the short-term demand for "data labeling and feedback labor" in the AI industry, while also raising concerns about long-term job stability and salary structures [5]. Group 3: Diverging Opinions on AI's Timeline - Not all analysts agree with the rapid timeline for job replacement. Morgan Stanley suggests that the impact of AI may take longer to manifest in economic data, with significant disruptions potentially occurring in the late 2020s or beyond [7]. Group 4: AI Risks Identified by Industry Leaders - Anthropic's CEO, Dario Amodei, outlines six major risks associated with AI, including large-scale unemployment, the potential for AI to possess state-level power, and the rise of terrorism threats due to advancements in biology [9]. - He expresses concern that AI could empower authoritarian regimes and highlights the risks posed by AI companies themselves, which control significant data and influence over users [9][10].