2026.01. 29 本文字数：1311，阅读时长大约2.5分钟 据了解，DeepMind推出的基因解码工具AlphaGenome，可破解98%的基因"暗物质"，并且准确率 高达90%。 作者 | 第一财经 钱童心 1月29日，谷歌旗下人工智能公司DeepMind研究成果再登《自然》杂志封面。人工智能模型已经能 够解码人类基因。 这拓展了AI在基因领域的研究。据统计，AlphaGenome目前每天处理超过100万次API调用，用户 已经超过3000人，遍布160个国家/地区。研究人员已开始利用这一工具攻克生物学领域一些最棘手 的难题。 在今年年初举行的摩根大通医疗健康峰会上，行业高管提出，人工智能尚未完成药物开发中最具挑战 性的问题，即发现导致重大医学进步的新分子。AlphaGenome的问世有望帮助科学家们完成"最后 一块拼图"。 人类基因约有30亿个碱基，但其中只有不到2%的基因序列用于编码蛋白质，其余98%被称为非编 码区。然而，正是这2%的基因序列对调控基因的活性至关重要，并包含了大量与疾病相关的变异位 点。 据介绍，AlphaGenome能够同时对11种不同的基因调控过程进行综合预测，准确捕捉基因 ...

Market Overview - The Federal Reserve maintained the federal funds rate target range at 3.5%-3.75%, aligning with market expectations[9] - The Hang Seng Index closed at 27,827, down 2.58% for the day but up 8.57% year-to-date[3] - The S&P 500 index closed at 6,978, with a slight decrease of 0.01% for the day and a year-to-date increase of 1.94%[3] Stock Performance - Gold prices surged, with spot gold exceeding $5,280 per ounce, marking a monthly increase of over 22%[9] - Semiconductor stocks saw significant gains, with the Philadelphia Semiconductor Index rising by 2.34%[9] - Major tech stocks showed mixed results, with Intel up 11.04% and Microsoft down 6% post-earnings report[9] Sector Highlights - Oil stocks continued to rise, with Shanghai Petrochemical gaining over 5% due to geopolitical tensions driving oil prices up by 3%[9] - New consumer concept stocks performed well, with a snack retail chain soaring 69% on its debut[9] - Airline stocks faced pressure, with China Southern Airlines down 6.83% due to rising operational costs[9] Economic Indicators - The U.S. dollar index fell to 95.51, the lowest since February 2022, contributing to the rise in gold prices[9] - The Japanese yen strengthened, impacting export-related stocks negatively, with Toyota down 3.24%[13] - The Bank of Japan raised its policy rate to 0.75%, the highest in 30 years, indicating a cautious approach to future rate hikes[13]

"真武"的水平究竟有多高？对这个问题，我让Gemini大模型总结了主流英文媒体的报道，大致结论如下： 96GB显存、700GB/s片间互联，具备与H20类似的竞争力； 今天（2026年1月29日），对于中国AI产业，乃至中国芯片产业而言，是一个很有意义的日子：平头哥（阿里巴巴集团芯片业务主体）官网公布了一款名 为"真武810E"的高端AI芯片。在此之前，关于平头哥自研PPU的说法，早已在市面上广为流传，还曾登上《新闻联播》——其中出现的少量技术参数，引 发了国内芯片圈子的热烈讨论。现在，平头哥自研PPU"真武"，终于正式出现在外界视野之中了。 请注意：虽然这是"真武"的第一次公开亮相，但是在此之前，它早已投入实用了，规模还颇大：迄今已经在阿里云实现了多个万卡集群部署，服务了国家 电网、中科院、小鹏汽车、新浪微博……等各行各业的400多家客户。它不是一款停留在PPT上的芯片，也不是一款还在早期试用阶段的芯片，而是已经 扎扎实实地积累了大量客户案例、形成了一定的生态系统。早在半年多以前，我的一位阿里云的朋友就表示："我觉得平头哥自研芯片的水平不比市面上 的主流方案差。"当时我半信半疑，现在我差不多能理解他的自信了 ...

作者 | Big Technology Podcast 编译 | 王启隆 出品丨AI 科技大本营（ID：rgznai100） 如果说 Sam Altman 是 AI 时代的布道者，善于用宏大的愿景点燃公众的想象力；那么 Demis Hassabis 更像是一位在实验室里盯着显微镜的科学 家，冷静、严谨，对"炒作"有着天然的免疫力。 一年前，当整个硅谷都在因为 ChatGPT 的红利期似乎见顶而焦虑，甚至开始讨论"大语言模型（LLM）是否撞墙"时，Demis 却感到困惑。在他看来， 进步从未停止。他掌舵的 Google DeepMind 刚刚经历了 AlphaFold 3 的高光时刻，正试图将 AI 的触角从简单的聊天机器人延伸到生物学、物理学乃 至材料科学的最深处。 在达沃斯的一间木质会议室里，Demis 近期接受了 Big Technology 播客的专访。这场对话的特别之处在于，他没有回避那些尖锐的问题： 现在的 AI 是不是只有"金鱼记忆"？谷歌会不会为了财报在 Gemini 里塞满广告？所谓的 AGI 究竟是营销话术还是科学定义？ 最令人印象深刻的是他对"智能载体"的断言。在纪录片《The Think ...

黄金已经涨疯了！ 今早，COMEX黄金一度突破5600美元，伦敦金现逼近5600美元，前者在短短4个交易日内，连破7道整百关口，涨势之迅猛引发全球关注。 但斌此时转发了一篇分析"历史上黄金白银暴涨后惨烈结局"的短文，被网友反击："吃不上葡萄说葡萄酸。" 对此，他再次回应："我不酸！我对黄金没有太多研究。" 其中的观点是：黄金是保值资产，价格反映的是阶段性的供求变化，长期跑不赢生息资产。 对于近两年地缘冲突对黄金投资价值的影响，但斌认为这不好评估。 "黄金，我们其实没有太关注，就像巴菲特说的，黄金就是保值，其实过去200年黄金就是保值，因为它不是生息资产，长期跑不赢生息资产， 所以我们倒没有太买这些东西。这两年它又加了一个地缘政治问题，你想我们央行买了这么多、全世界都在买黄金，可能会导致阶段性的变 化，也行未来两年，惊涛骇浪来了呢？那它可能又涨，这个东西不好说。" 对于但斌更好看的生息资产，恰好其执掌的东方港湾海外基金最新美股持仓曝光，四季度大动作不断，最大的亮点是谷歌超越英伟达，成为第一大重仓 股。 截至2025年末，东方港湾海外基金共持有10只美股标的，持股市值合计13.16亿美元，折合人民币超91亿元。 ...

Core Viewpoint - Google has disrupted a significant residential proxy operation, IPIDEA, which was used to facilitate cybercrime by compromising millions of consumer devices globally [1][2]. Group 1: Action Against Cybercrime - Google has targeted IPIDEA, one of the largest residential proxy networks, to protect consumers from cybercriminals and state-sponsored hackers [2]. - The operation involved legal actions to seize internet domains managing compromised devices and the implementation of automatic protections for Android users via Google Play Protect [3]. Group 2: Impact on Proxy Network - The disruption has led to a significant degradation of IPIDEA's operations, reducing the pool of available devices for proxy operators by millions [3]. - IPIDEA operated at least 13 residential proxy brands, all of which were taken offline during the operation [4]. Group 3: Scale of Cybersecurity Threat - Google identified over 600 Android applications and 3,075 unique Windows files associated with the network's command-and-control infrastructure [4]. - Following the announcement, Alphabet Class A shares rose by 1.54% and Class C shares increased by 1.62% in after-hours trading [4].

Group 1: AI Chatbot Market Dynamics - OpenAI's ChatGPT currently holds 68% of the AI chatbot market, a decrease from 87.2% a year ago, while Google's Gemini has rapidly increased its market share to 18.2% from 5.4% in January 2025 [2] - The competition in the generative AI chatbot space is intensifying, particularly with the success of Gemini [1] Group 2: Investment and Valuation Trends - SoftBank is in discussions to invest up to $30 billion in OpenAI's latest funding round, which could elevate OpenAI's valuation to approximately $830 billion [2] - Anthropic has raised its revenue forecast for 2026 by 20% to $55 billion, indicating strong growth potential in the AI sector [3] Group 3: Emergence of Agentic AI - 'Agentic AI' represents a new wave of AI technology capable of achieving complex goals with minimal human oversight, exemplified by the Clawdbot (now Moltbot) [5] - The rise of agentic AI is causing traditional software stocks to face selling pressure as investors anticipate disruption [5] Group 4: Corporate Investments and Stake Valuations - Zoom has seen its shares surge following its $53 million investment in Anthropic, with its stake now valued at least $2 billion [4] - Other major tech companies, including Amazon, also hold stakes in Anthropic, reflecting the growing interest in AI technologies [4]

Core Viewpoint - The current interest in artificial intelligence (AI) is characterized as a bubble, with expectations of significant future losses for most investors [2]. Group 1: Historical Context of Technology Bubbles - Numerous technology bubbles have occurred throughout history, with revolutionary innovations often leading to greater speculative bubbles [3]. - Historical examples include the British railway mania, the internet boom, and the SPAC bubble, which share common characteristics that can be applied to the current AI situation [3][4]. Group 2: Characteristics of the Current AI Bubble - The emergence of AI has generated extravagant claims, with proponents suggesting it will solve major global issues and transform economies [9]. - There is a climate of easy money, with low interest rates encouraging speculative investments in AI, similar to past bubbles [11][12]. - Investor optimism is high, with significant disparities between investor sentiment and general public sentiment, driven by economic inequality [19][22]. - A surge in publications and media coverage promoting AI has created a narrative that fuels speculation, reminiscent of past technological manias [25][23]. Group 3: Investment Trends and Valuations - AI startups raised over $200 billion in 2025, with significant capital flowing into private markets rather than public equity [32][33]. - Valuations in the AI sector are extremely high, with companies like OpenAI and Anthropic seeing massive increases in their valuations [41]. - The current market cap to GDP ratio and cyclically-adjusted price-to-earnings (CAPE) ratios indicate that U.S. equities are overpriced, with the CAPE reaching 40 [34][40]. Group 4: Risks and Challenges - The AI technology remains immature, with many experts doubting its ability to achieve artificial general intelligence (AGI) in the near future [42][46]. - There is a significant overcommitment of capital in the AI sector, leading to concerns about future returns and profitability [49][53]. - Unscrupulous behavior and dubious accounting practices are present in the AI ecosystem, reminiscent of past bubbles [54][59]. Group 5: Potential Outcomes - The investment boom in AI may be nearing its end, with signs of financial strain among major players and rising costs impacting profitability [65][66]. - Historical patterns suggest that once the bubble bursts, significant corporate bankruptcies and a shakeout phase will follow, leading to a consolidation in the industry [61][62].

Core Insights - Google DeepMind's new AI model, AlphaGenome, expands the predictive capabilities of AI into the complex realm of the human genome, marking a significant advancement in genomic research [1] Group 1: AlphaGenome's Capabilities - AlphaGenome can simultaneously predict 11 different gene regulatory processes, accurately capturing complex interactions within genes [3][9] - The model analyzes intricate gene splicing mechanisms, identifying how a single gene can produce multiple proteins and when errors in this process lead to diseases [4] - It has successfully reconstructed pathogenic mutations related to leukemia, predicting changes in regions up to 8000 base pairs away from the gene [6][19] Group 2: Performance Metrics - AlphaGenome has achieved state-of-the-art (SOTA) performance in various tests, surpassing existing models in the field of genomic prediction [8][12] - In 24 evaluations related to genomic trajectory prediction, it secured 22 SOTA results, demonstrating its precision in capturing the effects of small genetic variations [12] - The model's predictions have been validated through rigorous benchmark tests, showcasing its ability to outperform competitors like Borzoi and Enformer in multiple rounds [12] Group 3: Technological Framework - AlphaGenome employs a hybrid architecture combining CNN and Transformer technologies, allowing it to extract local DNA sequence features while capturing long-range dependencies [23][30] - The model's input window has been expanded to 1 million base pairs, enabling comprehensive coverage of interactions between remote enhancers and promoters [28] - A two-phase training strategy was implemented, including pre-training with strict cross-validation and a distillation strategy to enhance generalization and inference efficiency [30] Group 4: Applications and Implications - AlphaGenome's ability to predict molecular phenotypes from DNA sequences enhances the understanding of non-coding regions, addressing challenges in genome-wide association studies (GWAS) [17] - The model has successfully identified regulatory directions for 49% of GWAS-related variants, significantly exceeding traditional methods [17] - Its findings provide actionable insights into the biological functions of non-coding region variations, potentially leading to breakthroughs in disease understanding and treatment [23]

挪威主权财富基金2025年录得15.1%的年度回报，在科技和金融公司强劲表现的推动下，这只规模达2.2 万亿美元的全球最大主权基金延续了对大型科技股的依赖。 管理该基金的挪威央行投资管理公司（NBIM）周四发布声明称，股票投资去年回报率达19.3%，其他 资产类别同样实现增长。尽管整体表现稳健，该基金仍落后其基准指数28个基点。 "科技、金融和基础材料板块的股票表现突出，为整体回报做出重大贡献，"首席执行官Nicolai Tangen在 声明中表示。该基金最大持仓包括英伟达、苹果、微软、Alphabet和亚马逊等科技巨头。 随着欧美关系近期因特朗普威胁占领格陵兰等事件趋于紧张，该基金超过半数资产集中在美国的配置引 发专家关注。本周一个政府任命的专家小组专门建议基金为日益加剧的地缘政治动荡做好准备。 科技股主导业绩表现 大型科技股近几个季度持续主导该基金的业绩表现。NBIM持有全球约7200家公司约1.5%的上市股份， 是全球最大的主权财富基金。其最大持仓集中在美国科技巨头，包括英伟达、苹果、微软、Alphabet和 亚马逊。 据彭博此前报道，NBIM已对其持有的最大美国科技股进行了减持调整，但这些公司仍在其投 ...