没想到，2月21日它就整了波大活，发了封致歉信，承认GLM-5发布后犯了三个错。 前两天，才和大家聊了聊在港股创下暴涨神话的智谱AI，就在前一个交易日，这家上市仅一个多月的"全球大模型第一股"，凭着单日暴涨42.72%的行 情，把市值干到3232亿港元，超越一堆老牌互联网大厂，成了开年国内最猛的科技黑马。 一是GLM-5算力消耗高于老模型，公司设置高峰3倍、非高峰2倍的扣费倍率，却未提前完整公示，引发用户反馈用量异常、消耗过快。 二是GLM-5上线后流量远超预期，叠加黑灰产、黄牛恶意占用算力，公司算力扩容没跟上，只能按用户等级分阶段限流，很多付费用户无法正常使用核 心模型。 三是老用户升级机制设计粗糙，不少老用户误升级新套餐后权益受损，公司既无提前风险提示，也未设置回滚通道。 对此，智谱同步推出整改与补偿方案。 整改上，完整公示扣费规则，用量看板刷新缩至10分钟，明确模型分级开放节奏，承诺打击黑灰产、扩容算力。 补偿上，2026年1月1日-2月21日的Lite、Pro付费用户可全额退款，误升级老用户支持一键回滚，全量订阅用户统一延长15天使用期。 如今智谱已搭建起完整产品体系，包括面向普通用户的AI助手智谱清 ...

Core Viewpoint - The article discusses the competitive landscape of AI models during the 2026 Spring Festival, highlighting significant investments from major tech companies and the emergence of new AI products, indicating a shift in the industry towards multi-modal AI models and increasing token consumption [4][5][6]. Group 1: AI Model Releases and Market Dynamics - Major tech companies are engaging in a "red envelope war" with substantial financial commitments, including Baidu's 500 million, Tencent's 1 billion, and Alibaba's 3 billion [4]. - A series of AI model releases, including ByteDance's Doubao 2.0 and others, signifies a strategic positioning in the AI market, with each release aiming to reshape industry dynamics [4][6]. - The rapid release of multi-modal AI models is expected to increase the demand for tokens, leading to a structural rise in token consumption [6][7]. Group 2: Token Consumption Trends - The daily token consumption for major AI models has seen exponential growth, with Volcano Engine's model reaching 63 trillion tokens daily by the end of 2025, up from 20 trillion at the end of 2024 [7][8]. - The overall daily token consumption in China is projected to escalate from 100 billion in early 2024 to 180 trillion by February 2026, indicating a significant increase in AI usage [7][9]. Group 3: Pricing and Market Implications - The global cloud computing market is experiencing a price increase, with major players like Amazon and Google raising their prices, prompting domestic companies to follow suit [10]. - Zhizhu's recent price hike for its GLM Coding Plan, with an increase of at least 30%, reflects the strong market demand and the need for enhanced service quality [10][11]. - The shift in the industry from measuring success by user engagement to token consumption indicates a new pricing strategy, allowing model providers to leverage their bargaining power [11]. Group 4: Investment Directions - Investment opportunities are identified in cloud service providers and infrastructure, as AI-driven IT spending continues to rise, benefiting from increased demand for GPU computing and storage [12]. - Large model providers that can maintain subscription retention and expand their enterprise presence in high ROI scenarios are positioned to thrive despite competitive pressures [12]. - The need for AI safety and governance tools is expected to grow as companies integrate AI into workflows, creating a demand for AI security platforms [13].

编者荐语： 日经中文网与36氪开展内容交换合作。将精选36氪的精彩独家财经、科技、企业资讯，与读者分享。 以下文章来源于36氪Pro ，作者剡沛 智谱的上市是众望所归，也算是中国AI的成 人礼。 文 ｜ 剡沛 封面来源 ｜ 日经中文网 2026年1月8日，中国大模型第一股诞生——智谱，正式挂牌港交所。 上市第5天（12号午盘），市值一度超了1100亿港元，比挂牌当天翻了一倍还多（511亿港元左右）。 虽然这次敲钟，是中国AI行业苦等3年的结果，但很多股民仍然十分诧异： 中国大模型第一股，为什 么不是DeepSeek、Kimi或者豆包，而是智谱？ 它甚至还被许多媒体，冠上了"中国OpenAI"的称号， 所以智谱到底是谁，又来自哪里？ 定义这家公司的核心标签，一定是清华。 创始人唐杰及创始团队，都来自清华大学计算机系。知识工 程研究室（KEG），这是国内最早开展NLP（自然语言处理），和知识图谱研究的实验室之一。Kimi 月之暗面的创始人杨植麟，也是这个实验室的成员，而唐杰，还当过杨植麟的老师。 36氪Pro . 36氪旗下官方账号。深度、前瞻，为1%的人捕捉商业先机。 2024年10月，智谱发布了AutoGLM ...

Group 1 - The core point of the article is that the stock price of Zhiyu surged by 41%, reaching a market capitalization of over 190 billion HKD, driven by two key factors: the price increase of the GLM Coding Plan and the official open-sourcing of the GLM-5 model [1] - The company announced a structural adjustment to the pricing system of the GLM Coding Plan, with an overall price increase starting from 30%, effective from February 12, while maintaining prices for existing subscribers [1] - The GLM-5 model has achieved state-of-the-art performance in coding and agent capabilities, closely approaching the performance of Claude Opus4.5 in real programming scenarios [1] Group 2 - Zhiyu confirmed that the previously popular model "Pony Alpha" on the OpenRouter platform is indeed the GLM-5, which is now available on the chat.z.ai platform [2] - The company has completed deep adaptation and operator-level optimization of GLM-5 with major domestic chip platforms, ensuring stable and efficient operation on domestic computing clusters [2] - Since its listing on the Hong Kong Stock Exchange on January 8, 2026, the company's stock price has increased from 120 HKD to 422 HKD, a rise of 251.67% [2] Group 3 - Zhiyu is currently preparing for an A-share listing, having submitted its IPO counseling record to the China Securities Regulatory Commission, with CICC as the counseling institution [2][3]

Core Insights - The competition in the AI agent sector is intensifying, with a focus on "machine execution capability" as AI agents evolve from passive responders to proactive collaborators in various fields [1] - Financial AI agents face unique challenges due to the industry's high-risk and heavily regulated nature, necessitating high reliability and accuracy [1] Financial AI Agents: Unique Characteristics and High Barriers - Financial AI agents must achieve 100% execution accuracy and seamless integration with complex systems to handle sensitive tasks like risk control and financial reconciliation [1] - The market for financial AI agents can be categorized into different "capability quadrants," highlighting the varying value propositions and applicable scenarios [1] Professional AI Agents: Breaking Through Complex Business Scenarios - Jinzhihui's AI Agent platform stands out by focusing on high-value, business-relevant deep empowerment, achieving significant automation in core business processes [2] - The platform enhances the "Reflection" capability, allowing for repeated validation of task results, ensuring precision in critical financial scenarios [2] - Jinzhihui's unique "Browser Use" model improves script development efficiency by over 70%, facilitating a shift from "how to execute" to "defining business" [2] Execution-Focused AI Agents: Meeting Specific Automation Needs - Douzi, developed by ByteDance, offers a low-code, visual workflow design, enabling quick assembly of marketing automation and intelligent customer service applications [4] - While ideal for marketing teams seeking rapid deployment, Douzi lacks the necessary accuracy and compliance for complex core business processes in regulated industries [4] Tencent Yuanqi: Leveraging WeChat Ecosystem - Tencent Yuanqi maximizes the WeChat ecosystem's potential, integrating various Tencent services for quick deployment of intelligent services in specific scenarios [6] - However, it does not meet the financial-grade requirements for handling complex core business processes [6] Zhiyu Qingyan: Expertise in Professional Text Understanding - Zhiyu Qingyan excels in deep semantic understanding and generation of professional texts, making it a valuable tool for research and legal institutions [7] - It faces limitations in linking complex systems and meeting the reliability and accuracy standards required in the financial sector [7] Baidu Lingjing: Intelligent Management of Enterprise Knowledge - Baidu Lingjing focuses on the intelligent management of enterprise knowledge assets, providing precise answers through smart retrieval [7] - It is crucial for large organizations aiming to integrate and activate scattered knowledge but is not suited for core business automation in finance [7] Dify: Open-Source LLM Application Development Framework - Dify offers a comprehensive toolchain for developing applications, appealing to teams with strong technical capabilities seeking customization [7] - However, it lacks the reliability and precision needed for financial-grade core business automation [7] AI Agent Selection: Linked to Enterprise Digital Maturity - The choice of AI agents is a significant decision tied to the level of digital maturity within an enterprise, impacting productivity liberation [9] - For enterprises with lower digital maturity seeking quick trials, execution-focused products may suffice, while those in regulated industries with higher maturity require robust solutions like Jinzhihui to enhance core competitiveness [9]

Core Insights - The report emphasizes the strong potential of the AI industry and high-growth segments such as gaming, recommending continued investment in these areas [2][13][16]. Group 1: Internet Sector - E-commerce: Alibaba is catalyzing AI-related developments, introducing the "Tongyun Ge" concept, which integrates large models, cloud computing, and chips as a key support for its technology strategy [2][16]. - Social Entertainment Media: Bilibili and Tencent are expected to see strong advertising momentum, with Tencent's gaming fundamentals improving and Bilibili preparing to release new games [2][17]. - Internet Healthcare: JD Health and Alibaba Health are leveraging their platform advantages to deepen collaborations with upstream pharmaceutical manufacturers, leading to sustained revenue and profit growth [2][17]. - Short Video: Kuaishou is maintaining a stable core business, with AI technology enhancing user engagement and commercial conversion [2][18]. - Trendy Toys + IP: Pop Mart announced the establishment of its European headquarters in London, aiming to expand its market presence [2][19]. - Long Video: Multiple platforms are releasing quality series, suggesting investment opportunities in iQIYI and Mango TV [2][20]. - Music Streaming: Tencent Music and NetEase Cloud Music show stable performance, although concerns about competition have led to valuation adjustments [2][20]. Group 2: Media Sector - Gaming: The report maintains a positive outlook on the gaming sector, with expectations of continued industry prosperity into 2026. Key recommendations include Tencent, NetEase, and companies with strong product pipelines like Century Huatong and Giant Network [2][21]. - Advertising: Adjustments in the advertising landscape are not expected to impact the operational trends of Focus Media, with increased investment from internet advertisers anticipated [2][22]. - Publishing: Some publishing companies are facing challenges due to educational reforms, but firms with strong fundamentals and high dividend yields are recommended [2][22]. - Film and Television: Attention is drawn to companies with robust project pipelines, such as Huace Film & TV and Mango TV, as well as cinema chains like Wanda Film and Hengdian Film [2][22]. - IP Derivatives: Companies involved in IP derivatives are highlighted for potential investment, including Huayi Brothers and Shanghai Film Group [2][22]. Group 3: AI Developments - The report notes the rapid advancement of AI applications, particularly with the emergence of Clawdbot, which has gained significant attention in the industry [2][23]. - Recommendations include major cloud players like Google and Amazon, as well as domestic giants like Alibaba and Tencent, focusing on their self-developed models and ecosystems [2][23]. - Specific applications in AI across various sectors are suggested for investment, including AI in gaming, marketing, and healthcare [2][23].

Group 1 - The launch of GLM-4.7 has led to a rapid increase in users for GLM Coding Plan, causing temporary strain on computing resources, with users experiencing throttling errors and slower model speeds during peak hours [1] - In 2024, the computing service expenses for the company are projected to be 1.553 billion yuan, accounting for over 70% of R&D expenditures, a trend expected to continue into the first half of 2025 [1] - Major tech companies like Microsoft and Google are optimistic about their capital expenditure guidance for 2026, with a total expected to reach 596.4 billion USD, a 47% year-on-year increase, indicating a sustained rise in AI computing and infrastructure investments [1][2] Group 2 - The PCB industry is entering a new cycle driven by AI, with both demand and prices expected to rise, while cooling technologies for AI GPU racks are increasingly favoring liquid cooling solutions [2] - Companies like Qingyun Technology and Chuangye Heima are collaborating with Zhiyu to provide computing resources and explore new AI service models, leveraging their respective strengths in customer resources and advanced model technologies [3]

Investment Rating - The report rates the electronic industry as "Outperform" compared to the market [1]. Core Insights - The successful IPO of Beijing Zhipu Huazhang Technology Co., Ltd. marks a significant milestone as it becomes the world's first publicly listed company focused on general large models, indicating a new growth momentum for the domestic AI industry [4][7]. - Zhipu has established itself as one of the largest independent large model vendors in China, with a revenue of 312 million yuan in 2024, reflecting a year-on-year growth of 149.60% and a compound annual growth rate of 133.96% from 2022 to 2024 [27][29]. - The company has made substantial investments in R&D, with expenditures of 529 million yuan in 2023 and 2.195 billion yuan in 2024, ensuring continuous product innovation and competitive strength [31]. Summary by Sections 1. Zhipu's Development History - Founded in 2019, Zhipu focuses on developing intelligent large models, leveraging technology from Tsinghua University's Knowledge Engineering Lab [8]. - The company released its self-developed pre-training framework GLM in 2021, overcoming technical bottlenecks of BERT and GPT [10]. 2. GLM-4.7 Model Performance - GLM-4.7, Zhipu's flagship open-source large language model, has shown significant enhancements in programming and reasoning capabilities, achieving an 84.9% score in the LiveCodeBench V6 benchmark [13]. - The model has improved its reasoning and tool invocation abilities, surpassing GPT-5.1 in the HLE test with a score of 42.8% [13]. 3. Zhipu's Market Position - Zhipu is recognized as the largest independent large model vendor in China, with a revenue ranking second among all general large model developers [27]. - The company’s revenue growth is driven by its integrated MaaS platform, which rapidly penetrates various sectors such as technology and finance [29].

Core Insights - The AI sector witnessed significant developments in 2025, highlighted by the launch of the DeepSeek-R1 model, which achieved over 22.15 million daily active users within 21 days, sparking widespread AI adoption during the Chinese New Year [1] - Major transactions included Meta's acquisition of Manus for between $2 billion and $3 billion, and the IPOs of Chinese AI unicorns Zhiyuan and MiniMax on the Hong Kong Stock Exchange, marking them as the first and second publicly listed companies in the global large model space [1] Group 1: Manus and Meta - Manus, founded by Xiao Hong, initially gained attention with a successful test video and $75 million in funding, but faced challenges including layoffs and a potential crisis before being acquired by Meta [4][6] - Xiao Hong will take on a significant role at Meta, focusing on AI strategy and product development, following Meta's history of acquiring key talent in the AI field [6][10] - Meta's strategy appears to be shifting towards commercialization, as evidenced by the acquisition of Manus and the integration of its team into Meta's AI initiatives [11][12] Group 2: Zhiyuan and MiniMax IPOs - Zhiyuan, a project stemming from Tsinghua University, successfully transitioned from academic research to commercial success, becoming the first independent company to develop a large model with over 1 trillion parameters [15][16] - MiniMax, founded by former SenseTime executive Yan Junjie, focuses on multi-modal AI models and has attracted significant investment from major gaming companies, positioning itself as a key player in the AI landscape [17][18] - Both companies' IPOs reflect a broader trend of AI firms seeking to capitalize on their technological advancements and market potential, despite facing challenges such as legal disputes over intellectual property [19][20]

Core Insights - The successful IPOs of Zhipu and MiniMax mark a significant transition for China's AI industry from technology validation to commercialization, indicating a new phase of capital market engagement for the large model sector [1][3][21] - Both companies have shown strong market performance post-IPO, with Zhipu's market cap reaching approximately HKD 110 billion and MiniMax's exceeding HKD 1.2 trillion at their peaks [1][3] Company Performance - Zhipu's stock price increased by over 13% on its listing day, with a market cap of about HKD 579 billion, later peaking at over HKD 1.1 trillion before settling around HKD 800 billion [1][3] - MiniMax's stock surged nearly 110% on its first day, with a market cap surpassing HKD 1.05 trillion, later adjusting to approximately HKD 1.128 trillion [1][3] Market Sentiment - The IPOs reflect a warming attitude of the Hong Kong market towards tech companies, with both firms raising significant capital—Zhipu HKD 4.348 billion and MiniMax HKD 5.540 billion, placing them in the upper tier of recent tech IPOs [3][6] - The subscription amounts for the IPOs were substantial, with MiniMax attracting HKD 253.3 billion and Zhipu HKD 197.7 billion, indicating strong investor interest [6][10] Industry Dynamics - The current valuation of MiniMax, with a market cap significantly exceeding its annual revenue of approximately HKD 4.17 billion, suggests a positive market outlook for AI development in China [10] - Both companies represent different market strategies within the AI sector, with Zhipu focusing on G and B-end markets and MiniMax targeting C-end applications, showcasing the diverse paths of AI development in China [10][11] Future Implications - The long-term performance of Zhipu and MiniMax will serve as a reference point for the valuation framework of the Chinese large model industry, potentially leading to a more mature assessment of AI companies [11][21] - The successful IPOs may encourage other AI firms to pursue public listings, as seen with comments from other industry leaders regarding their future IPO plans [17][20]