自ChatGPT点燃全球AI热潮以来，技术突破与算力竞争主导着大模型领域的角逐，而谁能率先登陆资本市场备受市场关 注。 1月8日，北京智谱华章科技股份有限公司（以下简称"智谱"）在港交所敲锣上市。上市首日，智谱收盘价为131.5港元，较 116.2港元的发行价上涨约13.17%。 规模化增长态势显著 对尚处发展初期的大模型行业而言，资本市场不仅看重技术故事，更关注商业模式的可行性与成长性。 从首个开源千亿模型、首个对话模型、首个多模态模型，到全球首个设备操控智能体，再到登顶全球榜单的GLM-4.7，智 谱始终坚持全栈自主研发的技术路线。 智谱的收入主要来自大模型业务收入，采用MaaS（模型调用服务）模式，通过API（应用程序编程接口）调用向开发者和 企业输出通用智能能力。财务数据清晰展现出其商业化进程已驶入快车道：2022年至2024年，公司收入从5740万元增长至3.124 亿元，年复合增长率达130%；2025年上半年，收入进一步增至1.91亿元，同比增长325%，规模化增长态势显著。 在收入高速增长的同时，智谱的毛利率始终保持行业头部水平，2022年至2024年及2025年上半年分别为54.6%、64. ...

AIPress.com.cn报道 当大模型不再仅仅满足于预测下一个汉字，而是试图预测世界的下一个状态时，人工智能才真正开始理解因果，触摸现实。这是未来，也是2026年AI即 将发生的变化。 本文结合智源研究院提出的AI十大趋势预测，梳理了AI在2026的将有之变，相信能够为我们勾勒了一幅从虚拟走向实体、从单体走向群智的未来图景。 图说：智源研究院 2026十大AI技术趋势 趋势一：世界模型确立认知新范式 行业对于智能的理解，正经历一场静水流深的转变，共识正从单一的语言模型，转向能够理解物理规律的多模态世界模型。 Next-State Prediction（NSP）范式的确立，标志着AI不再仅仅满足于在文本中预测下一个词汇，它开始尝试预测世界的下一个状态。 正如智源悟界所验证的那样，当机器掌握了时空连续性与因果关系，它便跨越了感知的边界，触碰到了真正的认知与规划。 趋势二：具身智能的产业"出清"与落地 趋势五：新"BAT"格局下的垂直突围 C端超级应用的"All in One"入口成为兵家必争之地。海外有OpenAI与Google引领风骚，国内字节、阿里、蚂蚁等巨头亦依托生态积极布局。 我们可以看到，蚂蚁推出的 ...

The Tesla autonomy debate has always been part engineering argument, part social‑media war, so when a clip that changed the tone surfaced on X, I was not surprised. Nvidia CEO Jensen Huang said he thinks “the Tesla stack is the most advanced autonomous vehicle stack in the world” and that he is “fairly certain they were already using end‑to‑end AI,” as cited at CES 2026 in his Bloomberg interview. Sawyer Merritt, a widely followed Tesla investor, posted the clip with the quote on X (formerly Twitter) and ...

Group 1: Generative AI Developments - OpenAI launched ChatGPT Health, which connects to electronic medical records and Apple Health, with over 230 million weekly users consulting health issues [1] - The feature interprets health data and is developed with feedback from over 260 doctors across 60 countries, currently available to a limited number of users in the U.S. [1] - Claude Code released major updates, enhancing development efficiency and fixing security issues, with new features like Skills hot reloading and multi-language response configuration [2] Group 2: Market Movements and Company Performance - Zhiyu Technology became the first global company listed on the Hong Kong Stock Exchange focused on AGI, achieving a market capitalization of HKD 52.8 billion on its first trading day [3] - The company aims for a revenue growth rate of 130% from 2022 to 2024, with projected revenues of CNY 57.4 million, CNY 124.5 million, and CNY 312.4 million respectively [3] - Alphabet's market capitalization reached USD 3.885 trillion, surpassing Apple, with Gemini's market share exceeding 20% while ChatGPT's dropped below 65% [6] Group 3: Technological Advancements and Research - DeepSeek updated its R1 paper, expanding it from 22 to 86 pages, detailing technical specifications and training costs, showcasing competitive performance against OpenAI's models [4] - Epoch AI's report indicates that China's AI models lag behind the U.S. by an average of 7 months, with the gap narrowing over the next few years [9] - An a16z investor highlighted that by 2026, AI tools will shift focus from execution to exploration, emphasizing the need for software-driven approaches in product development [10][11]

高额跨境收购引起商务部关注，Manus这次来到了合规讨论的中心。 2025年最后一天，AI应用Manus 的公司蝴蝶效应宣布，被美国科技巨头Meta以数十亿美元收购。这是Meta成立以来的第三大收购，也是近年少见的中国背 景 AI 应用被整体并购案例。 Manus 主打在电脑端调用多种工具的智能体能力，能完成分析股票、写简历、订行程等任务，由中国创始人肖弘及其团队开发。产品在国内走红后，公司主 体、员工和技术团队迁往新加坡。 此次跨境并购随之带来了更复杂的合规议题。1月8日，商务部在新闻发布会上回应了 Meta 收购 Manus 是否需要审查的问题。商务部明确表示，将会同相关 部门对此项收购与出口管制、技术进出口、对外投资等相关法律法规的一致性开展评估调查。 作为被不少人视为成功范例的 AI 初创公司，这一收购案的后续走向，将会成为一个具有参考意义的合规样本。 一位接近Manus的人士向21记者透露，当初公司在迁移至新加坡时，应该已经意识到了这并不意味着能跟国内监管完全切割。毕竟在当前的背景下，监管越 来越有技术穿透性，"规避"可能更难持续。 2025年3月，Manus邀请码在二手平台流行（图源：互联网） 反 ...

从招股书看，智谱的技术底气和商业野望均不容小觑，其大模型基础能力，要去全球评测打榜；主攻B 端场景的变现策略，要与大厂分庭抗礼。 1月8日，智谱登陆港股，首日股价较发行价上涨超13%，成功拿下"大模型全球第一股"。从北京到香 港，从高校实验室到股票交易所，智谱在资本市场的一小步，也是大模型在公众视野的一大步。 尽管大模型成熟期言之尚早，其成长历程不可能只有上市这一次考试，但大考毕竟广受关注。 大考智谱走过的路，是中国大模型产业从技术探索走向商业验证的缩影。智谱董事长刘德兵说，IPO最 重要的是对"技术逻辑能否跑通商业逻辑"的实战检验。 智谱上市踩中了人工智能资本化的风口期。近段时间，国产GPU摩尔线程和沐曦股份上市接连引爆市 场。国产大模型赛道的智谱敲钟，再度推高市场对AI企业的商业化预期。 资本盛宴开启，但鸣锣开市仅仅是一个起点，当大模型的神秘面纱一次次被揭去，企业需要接受来自公 开市场最严苛、最持续的审视——技术和算力必须落地生根，数据和业绩才是最终的答卷。 资本市场既是最好的加速器，也是无情的试金石。它奖励前瞻性，但绝不会为纯粹的梦想长期买单。即 便没上市，曾经的行业领军者已分化出不同的路径选择，不断尝 ...

此次MiniMax备受头部基金的青睐。这些下单的机构中不乏众多的长线基金及国家主权基金。包括新加坡主权基金、南非以及中东、加拿大等多家主权基 金认购金额超过10亿美金。与此同时，这些长线基金的认购订单总额超过60亿美金。 来源:腾讯财经 文:谢照青 即将于1月9日敲钟上市的大模型公司MiniMax，创下近年来港股IPO机构认购历史记录。此次参与MiniMaxIPO认购的机构超过460家，超额认购达70多 倍。 此前的认购记录属于宁德时代，其在2025年登陆港股市场时，剔除基石后超额认购30倍。 与此同时，此次参与MiniMax国配订单的需求达到了320亿美元，最后超过460家机构实际下单了190亿美元。剔除基石部分，此次MiniMax的国配认购超 额到79倍左右。 | | | 截至12月31日止年度 | | 截至9月30日止九個月 | | | --- | --- | --- | --- | --- | --- | | | 2022年 | 2023年 | 2024年 | 2024年 | 2025年 | | | 美元 | 美元 | 美元 | 美元 | 美元 | | | | | | (未經審計) | | | | | ...

Core Viewpoint - The acquisition of Manus by Meta for several billion dollars raises compliance concerns, particularly regarding cross-border regulations and potential antitrust issues, making it a significant case for future reference in the industry [1]. Group 1: Acquisition Details - Manus, an AI application company, was acquired by Meta, marking Meta's third-largest acquisition since its inception and a rare instance of a Chinese AI application being fully acquired [1]. - The company, founded by Xiao Hong, has shifted its operations to Singapore after gaining popularity in China, indicating a strategic move to navigate regulatory challenges [1]. - Following the acquisition, Manus will cease its operations in China, and its founder will take on a role as Vice President at Meta, highlighting the importance of the founding team in the acquisition [3]. Group 2: Regulatory Concerns - The acquisition has prompted the Ministry of Commerce to evaluate its compliance with laws related to export controls, technology transfer, and foreign investment [1]. - There is a noted regulatory vacuum regarding antitrust and foreign acquisition reviews, as Manus's revenue of approximately $100 million (around 700 million RMB) does not meet the thresholds for mandatory reporting under Chinese antitrust laws [4][5]. - The shift of Manus's operational entity to Singapore may further complicate compliance with Chinese regulations, particularly concerning data and technology export controls [5]. Group 3: Data Compliance Issues - The acquisition raises questions about data compliance, especially if Manus has user data from China, which could complicate data export regulations [6]. - Manus's products have primarily targeted overseas markets, but the handling of any existing Chinese user data remains uncertain [6][7]. - Compliance with China's data export regulations may require re-evaluation following the acquisition, particularly if data is transferred to new third parties [10]. Group 4: Export Control Risks - The core technology of Manus may fall under China's export control regulations, necessitating careful assessment to avoid violations [12]. - The technology's classification and whether it requires prior approval for export is a critical concern, especially given the potential implications for AI companies operating internationally [13][14]. - Companies are advised to conduct thorough compliance evaluations regarding export controls, as overlooking these regulations can lead to significant legal repercussions [14].

"ChatGPT健康"将在OpenAI内设立独立空间，所有文件、对话和连接的应用将与其他聊天记录分开存 储。OpenAI承诺，这些对话不会被用于训练其基础模型，健康空间内的信息和记忆也不会流向外部。 OpenAI表示，该功能将首先向小部分早期用户开放以收集反馈，并在未来几周内扩大访问范围。 （央视财经《天下财经》）当地时间7日，美国开放人工智能研究中心，也就是OpenAI正式推 出"ChatGPT健康"，这标志着这家AI初创公司正式进军医疗健康领域。 据OpenAI的介绍，每周有超过2.3亿人在该公司平台上询问健康和保健问题，凸显医疗健康市场庞大的 用户需求。OpenAI最新推出的"ChatGPT健康"，并非用于诊断和治疗，也不能取代医疗护理，它旨在帮 助用户处理日常健康问题，并通过整合用户自身健康信息使ChatGPT的回复更具相关性。 编辑：令文芳 转载请注明央视财经 据报道，目前，越来越多的公司正瞄准利润丰厚的医疗健康市场，押注人工智能的进步能够帮助解析用 户健康数据中的规律，从而提供个性化的医疗建议。据市场研究机构Fortune Business Insight预计，2026 年，全球人工智能医疗健康市 ...

Core Insights - Nebius Group N.V. (NASDAQ:NBIS) is highlighted as a significant investment opportunity in the AI sector, particularly following the release of its updated AI cloud platform, Nebius AI Cloud 3.1 [1] Group 1: Product Launch and Features - Nebius AI Cloud 3.1 introduces next-generation NVIDIA Blackwell Ultra compute capabilities, facilitating the transition from AI experiments to large-scale production deployments [2] - The update enhances compute power, capacity management, and transparency, along with various developer and ecosystem features [2] - The GB300 NVL72 platform now supports rack-scale operations, functioning like a single giant GPU, which is essential for large-scale model training and high-throughput inference [3] - The platform expands its ecosystem with native integration with Dstack for streamlined workflows and simplified access to NVIDIA BioNeMo NIM microservices [3] Group 2: Security and Compliance - The update includes strengthened security features such as object storage data-plane audit logs for HIPAA compliance, per-object access controls, and enhanced identity management through Microsoft Entra ID integration [4] Group 3: Company Overview - Nebius Group N.V., based in Amsterdam, specializes in developing and operating AI cloud infrastructure tailored for training and inference of advanced machine learning models, offering full-stack GPU clusters and networking solutions [5]