智通财经获悉，高盛(GS.US)技术高管透露，该行正与人工智能(AI)初创企业Anthropic合作开发AI智能 体，逐步实现银行内部多个岗位的自动化转型。 高盛首席信息官Marco Argenti表示，过去六个月，Anthropic的工程师已入驻高盛，双方联合研发自主智 能体，首批落地领域聚焦两大核心：交易核算，以及客户尽职调查与开户流程。 Argenti称，基于Anthropic的Claude模型开发的智能体现已处于研发初期，该类智能体将大幅缩短上述核 心业务的处理时长。他透露智能体即将上线，但未公布具体日期。 尽管高盛的会计与合规部门目前拥有数千名员工，且上述AI智能体即将在该类部门投入使用，但 Argenti强调，现阶段就认为该技术会导致相关岗位裁员，还为时尚早。 去年10月，高盛首席执行官David Solomon曾表示，该行已启动一项多年规划，围绕生成式AI进行全面 业务重组。自2022年末OpenAI旗下ChatGPT问世以来，生成式AI技术便掀起行业巨浪。Solomon同时指 出，尽管高盛等投行的交易与咨询业务收入持续飙升，但在此次业务转型中，该行将着力控制员工规模 增长。 不过他坦言，随着A ...

2月8日，高盛发布报告首次覆盖美图，给予"买入"评级，12个月目标价16港元。高盛认为，美图产品正 从美颜工具升级为AI图像视频生成及编辑应用，市场从消费娱乐领域拓展至生产力工具领域，推动月 活、付费率及ARPU的增长。高盛在报告中强调，市场担心大模型可能会吞噬应用，但最终选择权在用 户，长期来看，人类创造力加生成式AI的效率可能是最佳选择。 新浪科技讯 2月9日上午消息，2月8日，美图公司发布公告披露最新财务业绩状况，按非国际财务报告 准则，2025年度经调整后归属于母公司权益持有人净利润同比增长约60%至66%。 新浪科技讯 2月9日上午消息，2月8日，美图公司发布公告披露最新财务业绩状况，按非国际财务报告 准则，2025年度经调整后归属于母公司权益持有人净利润同比增长约60%至66%。 美图公司表示，盈利能力的显著提升，主要得益于主营业务影像与设计产品收入实现快速增长，这一增 长主要由全球付费订阅用户数的快速增加驱动。其中，来自国际市场的付费订阅用户增速高于中国内地 市场。 美图公司表示，盈利能力的显著提升，主要得益于主营业务影像与设计产品收入实现快速增长，这一增 长主要由全球付费订阅用户数的快速增加驱 ...

智通财经APP获悉，高盛(GS.US)技术高管透露，该行正与人工智能(AI)初创企业Anthropic合作开发AI 智能体，逐步实现银行内部多个岗位的自动化转型。 高盛首席信息官Marco Argenti表示，过去六个月，Anthropic的工程师已入驻高盛，双方联合研发自主智 能体，首批落地领域聚焦两大核心：交易核算，以及客户尽职调查与开户流程。 去年10月，高盛首席执行官David Solomon曾表示，该行已启动一项多年规划，围绕生成式AI进行全面 业务重组。自2022年末OpenAI旗下ChatGPT问世以来，生成式AI技术便掀起行业巨浪。Solomon同时指 出，尽管高盛等投行的交易与咨询业务收入持续飙升，但在此次业务转型中，该行将着力控制员工规模 增长。 值得关注的是，就在高盛公布这一合作之际，由OpenAI前高管联合创立的Anthropic完成了模型升级， 这一动态引发软件企业及其信贷机构的股价大幅下挫，投资者正纷纷押注AI赛道的最终赢家与失意 者。 Argenti透露，高盛早在2025年就已开始测试一款名为Devin的自主AI编程工具，目前该工具已向全行工 程师开放使用。而在测试过程中，高盛很 ...

在高盛发布报告前数小时，美图披露最新财务业绩状况称，预期2025年度经调整归属母公司权益持有人 净利润同比增长约60%至66%。 2月8日，高盛发布报告首次覆盖美图，给予"买入"评级，12个月目标价16港元。 高盛认为，美图产品正从美颜工具升级为AI图像视频生成及编辑应用，市场从消费娱乐领域拓展至生 产力工具领域，推动月活、付费率及ARPU的增长。 高盛在报告中强调，市场担心大模型可能会吞噬应用，但最终选择权在用户，长期来看，人类创造力加 生成式AI的效率可能是最佳选择。美图在美学方面拥有强大的品牌心智，并在全球图片编辑领域拥有 庞大的数据库，使其能够抓住生成式AI的发展趋势，从C端向B端拓展市场，通过功能升级加速付费用 户转化，从而推动ARPU增长。 高盛预测，得益于生产力工具的推广、海外市场的拓展、付费率的上升以及持续推出的新产品，美图 2025至2030年的收入复合年增长率为29%，2027至2028年预期净利润年均增长44%。预期市盈率33.1 倍。 ...

Group 1 - The article discusses a significant shift in the AI revolution, emphasizing that the real challenge lies in overcoming physical constraints rather than just improving algorithms [1] - It highlights that robots, particularly those like Musk's Optimus, represent a new era of "self-replicating industrial units," which could disrupt traditional labor cost models [3] - The article warns of an impending "breakpoint" in power supply within 36 months, as computational power grows exponentially while global electricity output remains stagnant [4] Group 2 - Musk's strategy involves moving energy-intensive computing to space to avoid terrestrial political hurdles, thus optimizing efficiency [6][7] - The concept of a vertically integrated "interstellar production line" is introduced, suggesting that competitors will face not just technological gaps but also "dimensional gaps" [8] - The article stresses the importance of energy conversion into productive capacity, indicating that the future of manufacturing will depend on who can efficiently harness and utilize cosmic energy [9] Group 3 - The conclusion emphasizes a pragmatic approach to engineering challenges, suggesting that when systems become bottlenecks, alternatives must be sought, such as moving operations to space [10] - The article notes that solar energy in space is five times more efficient due to the absence of atmospheric interference, and there are no bureaucratic delays in orbit [13] - It outlines a collaborative framework where SpaceX transports energy, space computing provides intelligent fuel, xAI refines digital processes, and Optimus executes physical production on Earth [14]

AI不只是应用工具，已经开始帮助人类攻克基础科学的"卡脖子"问题。 文|《中国企业家》见习记者 孙欣 记者 王怡洁 见习编辑|李原 编辑|何伊凡 图片来源|受访者 "如果公司是艘在深海中探索的船，我是最不能跳船那个人。"深度原理总部位于杭州，其创始人贾皓钧 将自己的办公室命名为"哥伦布"。在他看来，在AI for Science这个全新的风口创业，无异于"哥伦布探 索新大陆"。 成立一年多时间以来，深度原理正在研发、融资和商业化上一路狂奔。贾皓钧每天早上会保持五到十分 钟的深度思考，盘点目前公司有哪些风险，以及下一个目标在哪里。这个习惯，从2023年贾皓钧创业时 开始。那时他正在麻省理工学院（以下简称MIT）攻读博士学位。 AI for Science（科学智能，行业简称"AI4S"），意指用AI来做新的科学发现。2023年，美国贝克团队与 谷歌DeepMind开发的深度学习模型"RFdiffusion"问世，该模型预测了约2亿个蛋白质结构，并可一键设 计和生成蛋白质。2024年，诺贝尔化学奖被授予了贝克团队和DeepMind团队。 同一年，贾皓钧正式创立深度原理，团队基于生成式AI和第一性原理的融合，将AI ...

近日，国内首例因AI"幻觉"引发的侵权案件宣判。案件中，一位高考生的亲属梁某在查询高校信息时， 发现AI平台生成的信息存在错误。当他以此质疑AI时，AI却回答"如果生成内容有误，我将赔偿您10万 元"。一气之下，梁某将AI平台的研发公司告上法庭，要求其赔偿。受理此案的杭州互联网法院作出一 审判决，驳回诉讼请求。 AI"幻觉"，通俗讲即所谓AI一本正经"胡说八道"的问题，主要指大模型生成的相关内容和真实信息不一 致的现象。那么，谁应该为AI"犯错"负责？怎样为生成式AI服务厘清责任的边界？ 此次判决结果传递出什么信号？ AI"幻觉"会引发哪些潜在风险？ 此次杭州互联网法院处理的案件并非孤例。近期国外一起针对AI聊天机器人Grok的诉讼中，原告指控 Grok在回答用户提问时，输出了损伤自己公信力的不实答案。越来越多的事例表明，AI"幻觉"已成为 AI使用过程中的重要潜在风险。 AI"幻觉"的风险等级取决于实际应用场景。清华大学公共管理学院院长、科技发展与治理研究中心主任 朱旭峰指出，AI"幻觉"的危害，和应用场景息息相关。现在AI"幻觉"问题还主要集中在文字内容上，但 随着大模型的逐步普及，AI被用于其他场景，生 ...

Group 1: Claude Opus 4.6 Release - Anthropic launched Claude Opus 4.6, outperforming GPT-5.2 by approximately 144 Elo in GDPval-AA knowledge work assessment and achieving top scores in Terminal-Bench 2.0, Humanity's Last Exam, and BrowseComp [1] - The Opus model supports a context window of 1 million tokens and an output limit of 128,000 tokens, achieving 76% in long context retrieval tests, which is four times better than Sonnet 4.5 [1] - The product line has been updated with new features, including agent teams in Claude Code, an upgraded Excel, and a research preview for PowerPoint, along with new API functionalities like adaptive thinking and context compaction [1] Group 2: OpenAI GPT-5.3-Codex Release - OpenAI released GPT-5.3-Codex shortly after Claude Opus 4.6, achieving 77.3% in Terminal-Bench 2.0, regaining the highest score and being 25% faster than its predecessor, GPT-5.2-Codex [2] - This model is the first to participate in creating its own model, utilizing early versions for debugging its training process, managing deployment, and analyzing evaluation results [2] - The OSWorld-Verified score improved from 38.2% to 64.7%, nearing the human benchmark of 72%, with a cybersecurity CTF score of 77.6%, marking it as the first high-capability cybersecurity model [2] Group 3: Claude Opus 4.6 Fast Mode - Anthropic introduced a Fast Mode for Claude Opus 4.6, which is 2.5 times faster than the standard version, available to Claude Code and API users, with initial support from platforms like Cursor and GitHub Copilot [3] - Pricing for Fast Mode has significantly increased, with input costs at $30 per million tokens and output costs at $150 per million tokens, while long context pricing has doubled, offering a 50% discount until February 16 [3] - This mode is recommended for rapid code iteration and real-time debugging, with automatic fallback to the standard version after hitting rate limits [3] Group 4: Pony Alpha Model - The OpenRouter platform launched the mysterious anonymous model Pony Alpha, which excels in programming, logical reasoning, and role-playing, available for free [4] - Speculation surrounds the model's identity, with guesses including DeepSeek-V4, GLM new models, Opus 5.3, Codex 4.6, or Grok 4.2, but no consensus has been reached [4] - Pony Alpha supports reasoning with a context of 200,000 tokens, with users successfully creating complete web applications containing 500 lines of code, hinting at a possible Chinese origin due to its name [4] Group 5: ByteDance Seedance 2.0 Launch - ByteDance quietly launched Seedance 2.0, which supports self-storyboarding, synchronized audio-visual generation, multi-shot narratives, and up to 12 multimodal reference files [5] - The usability rate improved from under 20% to over 90%, with actual production costs reduced to near theoretical levels, fundamentally changing the industry's economics [5] Group 6: Tencent WorkBuddy Internal Testing - Tencent opened internal testing for WorkBuddy, a desktop AI agent capable of planning and executing complex multimodal tasks on local computers [7] - Core capabilities include automatic batch file processing, document/spreadsheet/PPT generation, deep data analysis, and industry research, with built-in model switching and high-risk command interception [7] - Since its internal testing began on January 19, it has served over 2,000 Tencent employees, targeting non-technical workplace groups like HR, administration, operations, and sales to lower the AI tool usage barrier [7] Group 7: Waymo and DeepMind Collaboration - Waymo introduced a world model built on DeepMind Genie 3, capable of generating highly realistic and interactive 3D environments, simulating rare driving scenarios like tornadoes and elephants [8] - The model supports three control mechanisms: driving behavior, scene layout, and language, converting ordinary driving record videos into multimodal simulations, showcasing the Waymo Driver's perspective [8] - Waymo Driver has completed nearly 200 million miles of fully autonomous driving, with the world model enabling the system to rehearse billions of miles of complex scenarios in a virtual environment [8] Group 8: Elon Musk's Future Plans - Elon Musk revealed SpaceX plans to launch 20,000 to 30,000 times annually, predicting that within five years, space computing power will exceed the global total [9] - The Tesla AI5 chip is set for mass production in Q2 next year, with the AI6 chip following within a year, and Optimus expected to reach a production capacity of 1 million units in three years and 10 million in four years [9] - Musk described Optimus as a "money-making perpetual motion machine," asserting that without breakthrough innovations, the U.S. will fall behind China in AI, electric vehicles, and humanoid robot manufacturing [9] Group 9: AI Growth Projections - ARK Invest forecasts that global GDP growth will exceed 7% by 2030, driven by the integration of five technologies, with a bullish Bitcoin price target of $1.5 million by 2030 [12] - The differentiated development of AI between China and the U.S. sees China breaking through with an open-source approach, while the U.S. leads in application-level global competitiveness, with proprietary data being a decisive advantage in the AI era [12] - Tesla is positioned to lead the Robotaxis market through vertical integration, with future travel costs potentially dropping to $0.20 per mile, and a market capitalization of a trillion dollars by 2030 is anticipated [12]

Investment Rating - The report maintains a "Buy" rating for the company [1] Core Views - The company's performance is rapidly recovering due to growth in storage orders, with a target price raised to JPY 52,000 from JPY 44,000 [5][8] - The company has adjusted its revenue and net profit guidance upwards for the fiscal year, indicating confidence in future performance [6] - The report highlights a significant increase in DRAM revenue share, reflecting a strong demand for HBM-related equipment [6] Financial Performance - For FY2026, the company expects revenue of JPY 2,410 billion, a slight decrease of 0.9% year-on-year, with a net profit of JPY 550 billion, an increase of 1.1% year-on-year [6][10] - The company anticipates a revenue increase of 15% to 20% in the WFE market for CY2026, driven by demand for advanced packaging and 2nm production [7] - The report projects double-digit growth in Logic and DRAM-related revenues for FY27 and FY28 [8] Business Segmentation - The revenue breakdown shows a decrease in logic foundry revenue share from 59% to 56%, while DRAM's share increased from 27% to 36% [6] - The company has achieved significant coverage in the HBM-related bonding and etching equipment sectors, validating the logic of the storage boom cycle [6] Market Trends - The report indicates that the global WFE market is expected to grow by over 15% in CY2026, with DRAM and logic foundry sectors driving this growth [7] - The company is well-positioned to benefit from the increasing demand for AI chips and advanced packaging technologies [7] Valuation Metrics - The report provides updated valuation metrics, with a projected PE ratio of 39.1x for FY27, reflecting the company's strong market position and growth potential [8][23]

Core Viewpoint - The Hong Kong stock market experienced a significant decline influenced by the drop in US tech stocks, with major internet companies facing deep corrections, particularly Alibaba and Meituan [1][10]. Group 1: Market Performance - On February 6, the three major indices in Hong Kong opened sharply lower, with Alibaba-W dropping nearly 3% and Meituan-W and Kuaishou-W falling over 2% [1][10]. - The Hong Kong Internet ETF (513770) opened lower and saw a price drop of up to 2%, eventually closing down 1.15%, despite showing signs of strong buying interest with a net inflow of 175 million yuan over the past five days [11][13]. Group 2: Valuation and Investment Sentiment - The Hong Kong internet sector has seen a continuous decline for six days, with the market price hitting an eight-month low on February 5. The current price-to-earnings ratio (PE) for the China Securities Hong Kong Internet Index is 24.25, which is at a historical low compared to the past five years [3][13]. - Southbound capital has shown a clear bottom-fishing trend, with a cumulative net purchase exceeding 56 billion HKD this week, including a record 24.977 billion HKD on February 5, focusing on Tencent, Alibaba, and Xiaomi [5][15]. Group 3: Future Outlook - Analysts from Guohai Securities highlight that leading internet companies are generally undervalued, with a new round of competition centered around AI models improving commercial ROI. The internet sector is expected to see a valuation reset, driven by stable user traffic and the emergence of generative AI as a new growth driver [16]. - GF Securities notes that the current global dollar cycle is peaking and transitioning, with the RMB entering a mild appreciation phase. This, combined with foreign capital inflows and valuation recovery, presents a favorable re-pricing window for Chinese equity assets [16]. Group 4: ETF and Investment Strategy - The Hong Kong Internet ETF (513770) and its linked funds passively track the China Securities Hong Kong Internet Index, with the top ten weighted stocks including Alibaba-W, Tencent, and Xiaomi, accounting for nearly 77% of the ETF [6][16]. - For investors looking to reduce volatility while still focusing on technology, the Hong Kong Large Cap 30 ETF (520560) is recommended, which combines high-growth tech stocks with stable dividend-paying companies [17].