2026年1月23日，三七互娱披露接待调研公告，公司于1月22日接待淡水泉投资、安信基金、博时基金、 创金合信基金、财通基金、大成基金等111家机构调研。 调研情况显示，当前游戏市场有三大核心变化和潜力机会：国内小游戏市场高速增长，公司已确立绝对 领先地位，在研多款含IP产品（如《斗破苍穹》IP）并将持续投入以保持领先；SLG赛道战略价值与天 花板突破，公司围绕"轻量化""大题材"布局，多款新产品计划今明年测试，MMO产品《RO 仙境传说： 世界之旅》首发登顶中国港澳台地区榜单并将拓展东南亚、日韩和欧美市场；休闲赛道因碎片化需求增 长，公司海外布局三年处于"种树"阶段，"三消""二合"领域储备产品有望下半年及明年推出。 公司已构建差异化题材立项、工业化管线、研发流程AI应用及人才孵化四大核心研发能力，自研储备 超10款聚焦小游戏和SLG赛道。AI应用深刻改变行业模式，内部AI渗透率超90％，广告投放AI自动比 例超70％，研发美术AI渗透率65％并降低成本，未来AI将深度融合玩法实现"千人千面"。投资重点布局 游戏研发核心资产及AI相关公司（如智谱、月之暗面等）以保障产品供给和业务协同；人才梯队正向 年轻化迈 ...

Core Viewpoint - The listing of Zhipu marks a significant milestone in the Chinese AI industry, being recognized as the first stock of a large model company in China, with its market value exceeding 110 billion HKD shortly after its debut [5][16]. Company Background - Zhipu was founded in June 2019 by a team from Tsinghua University, focusing on natural language processing (NLP) and knowledge graphs, with initial funding of 40 million RMB [5][6]. - The company made a pivotal decision to develop its own architecture, the General Language Model (GLM), rather than following the popular GPT and BERT paths [6][8]. Technological Development - In August 2022, Zhipu open-sourced the GLM-130B model, which was a significant development just before the global popularity of ChatGPT [9]. - Following the release of ChatGPT, Zhipu launched ChatGLM and the smaller ChatGLM-6B model, which became popular among developers in China [9][10]. - By 2025, Zhipu's GLM model had advanced significantly, outperforming competitors in various benchmarks and gaining traction in the AI community [10][12]. Market Position and Adoption - Zhipu's products are integrated into over 12,000 enterprises globally, including major internet companies, government, healthcare, and education sectors, positioning it as a key AI supplier [12][14]. - The company is likened to a utility provider in the AI era, essential for businesses looking to adopt AI technologies [12]. Financial Performance and Future Outlook - Despite rising revenues, Zhipu faces increasing R&D expenditures and significant losses, a common trend among independent large model companies [14][17]. - The company has secured over 2.5 billion RMB in financing in 2023, indicating strong investor interest and support for its growth trajectory [14][16]. - The listing is seen as a rite of passage for the Chinese AI industry, but it also brings scrutiny regarding financial performance and market expectations moving forward [16][18].

Core Viewpoint - The release of the GLM-4.7 model by the company has led to a surge in users for its programming subscription service, prompting the company to limit sales to prioritize existing users' experience [1] Group 1: User Growth and Market Impact - The company has reduced the daily sales volume of its "GLM Coding Plan" service to 20% starting January 23 to ensure a better experience for current users [1] - The core user base for this service is primarily in the United States, indicating that Chinese AI models are not only catching up in technical quality but also gaining market share from American developers [1] Group 2: Global Influence and Product Development - The company is enhancing its global influence through an open-source strategy, with its GLM-4.7 model ranking first in global programming capabilities [1] - The recent launch of the GLM-Image model, trained entirely on Huawei chips, further demonstrates the company's commitment to integrating with domestic hardware [1] - Currently, seven out of the top ten positions in the global open-source model rankings are held by Chinese companies, showcasing the rapid construction of China's influence on the international stage through open-source ecosystems [1]

Core Insights - The AGI-Next summit confirmed a significant divergence in the Chinese large model industry, moving beyond a simple ToB vs. ToC debate to different approaches in realizing the value of intelligence [1][2] - The divergence is characterized by two paths: one focusing on embedding models into production processes for productivity certainty (ToB), and the other emphasizing real-life scenarios for user experience and scale (ToC) [1] ToB vs. ToC Divergence - The core discussion at the summit revolved around the clear differentiation between ToB and ToC, with experts agreeing that user perceptions of AI value differ significantly between the two [4] - ToC users often do not require extreme intelligence, using AI more like an enhanced search engine, while ToB users are willing to pay a premium for high-performing models that enhance productivity [4] - The business models also differ: ToC requires tight integration of models and products, while ToB tends to favor a layered approach where strong models attract application layer companies [4] Market Dynamics - The recent IPOs of companies like Zhiyuan and MiniMax exemplify the divergence, with Zhiyuan focusing on ToB and MiniMax on ToC, indicating a clear trend in the industry [7] - In the ToB space, companies like iFlytek are integrating AI into core production processes, demonstrating the practical application of high intelligence in enhancing productivity [7][8] - The financial sector is also seeing the emergence of ToB models, such as the intelligent risk control platform launched by Tongyi Qianwen, which improves credit risk identification and loan approval processes [8] Future Directions - The ToC path faces challenges in capturing user needs and ensuring sustainable commercialization, with competition extending beyond large model companies to traditional internet platforms [14] - The evolution of ToC is shifting towards extreme personalization and deep scenario engagement, focusing on high-frequency, strong-demand areas like education and content creation [14] - The relationship between ToB and ToC is expected to become more complementary, with ToB needing ToC's scenario understanding and ToC relying on ToB's foundational models [15] Global Competitive Landscape - The differentiation in the Chinese large model industry will significantly impact its international competitiveness, with experts noting that while Chinese companies show strong execution, they still lag in leading new paradigms [16] - The potential for Chinese companies to innovate in the application layer is highlighted, suggesting that they may lead in ToC product forms globally [16] Conclusion - The divergence in the large model industry is a sign of maturity, with ToB focusing on efficiency and endurance, while ToC emphasizes experience and scale, with time being the ultimate judge of success [18]

Core Insights - The release of the flagship model GLM-4.7 has led to a surge in users for the programming subscription service, prompting the company to implement sales limits to prioritize existing users' access [1] - The core user base for the service is primarily in the United States, indicating that Chinese AI models are not only catching up in technical quality but also capturing significant market share from American developers [1] - The company is enhancing its global influence through an open-source strategy, with the GLM-4.7 model ranking first globally in programming capabilities [1] Market Position - The company has reduced the daily sales volume of its "GLM Coding Plan" service to 20% of its current capacity starting January 23, to ensure a better experience for existing users [1] - The focus of Chinese AI models is on integration into industries at affordable prices, aiming to produce high-tech products for the global market [1] - The recent launch of the GLM-Image model, trained entirely on Huawei chips, further demonstrates the company's commitment to leveraging domestic hardware in its AI developments [1] Competitive Landscape - Currently, seven out of the top ten positions in the global open-source model rankings are held by Chinese companies, showcasing the rapid construction of influence by Chinese AI on the international stage [1] - The integration of open-source ecosystems with domestic hardware is a key strategy for Chinese AI to enhance its competitiveness globally [1]

Investment Rating - The report does not explicitly state an investment rating for the industry. Core Insights - The equipment manufacturing supply chain is transitioning from digitalization to intelligent decision-making, marking a critical phase of smart transformation [7][11]. - The complexity of global supply chains necessitates the adoption of AI technologies for real-time collaboration and intelligent decision-making [8]. - Current digital systems have reached efficiency bottlenecks, highlighting the need for enhanced cross-enterprise collaboration and data integration [12]. Summary by Sections 1. Transition from Digitalization to Intelligent Decision-Making - The equipment manufacturing industry, particularly automotive, is undergoing structural adjustments due to de-globalization, leading to complex supply chain networks involving thousands of suppliers [7]. - Intelligent decision-making is essential to address the challenges posed by global complexity, technological integration, and intensified market competition [8]. - Existing digital systems like ERP and WMS are primarily internal, failing to achieve deep interconnectivity and collaboration across enterprises, resulting in significant efficiency challenges [12]. 2. Intelligent Agents as Key to Supply Chain Collaboration - Generative AI-driven intelligent agents are emerging as solutions to the management challenges faced in complex supply chains, transitioning from algorithmic tools to business assistants [22]. - Intelligent agents can process vast amounts of unstructured data, ensuring real-time and accurate decision support for manufacturers [24]. - They help reduce supplier management difficulties and control procurement costs by analyzing supplier quotes and identifying potential premium areas [26]. 3. Characteristics of Intelligent Agent Technology Vendors - The intelligent agent market consists of four main types of technology vendors: foundational model vendors, industrial software vendors, data intelligence vendors, and AI-native application vendors [31]. - Each vendor type has unique strengths and weaknesses, with foundational model vendors excelling in model capabilities but lacking industry-specific knowledge [31]. 4. Case Study: Yixun Technology - Yixun Technology has developed a Data Agent platform, Alaya, which focuses on data-driven decision-making in the automotive manufacturing sector [35]. - The platform addresses data silos and enhances procurement efficiency, achieving over 60% improvement in procurement processes for a German automotive company [46]. - Yixun's competitive advantage lies in its robust data processing capabilities and deep industry know-how in automotive supply chains [42]. 5. Future Outlook for Intelligent Agents in Equipment Manufacturing - Intelligent agents are evolving from single-point tools to collaborative digital employee clusters, enhancing overall supply chain efficiency [49]. - The progression of intelligent agents will lead to greater autonomy in decision-making, enabling them to perform complex tasks independently [50]. - Future developments will see multiple intelligent agents working together across various supply chain functions, creating a more resilient and agile manufacturing ecosystem [56].

Group 1 - Runway has launched the new Gen 4.5 model, significantly improving lens control and storytelling capabilities, generating three shots (close-up, medium, and long) within 5 seconds [1] - In a test with 1,000 participants, only 57% could distinguish between AI-generated videos and real videos, with the model achieving near cinematic quality in facial consistency, lighting logic, and physical laws [1] - The video generation model is entering a new upgrade phase, with trends towards realism, audio-visual synchronization, refined local control, and longer generation times [1] Group 2 - Google has partnered with The Princeton Review to integrate a full set of SAT practice tests into Gemini, allowing users to take free full-length mock exams with immediate scoring and detailed error analysis [2] - The tests cover reading, writing, and math modules, supporting customizable countdowns and hints, with Gemini breaking down problem-solving steps for better understanding [2] - SAT is just the beginning, as Google plans to expand Gemini to more standardized tests, positioning AI as an expert assistant across various industries [2] Group 3 - Zhizhu's GLM-4.7 has seen rapid user growth leading to computational strain, causing some users to experience throttling and slower model speeds during peak times [3] - Starting January 23, the GLM Coding Plan will be sold in limited quantities, reducing daily sales to 20% to prioritize the programming experience for existing users [3] - Zhizhu is developing more powerful and efficient models while accelerating computational capacity expansion, with automatic renewals unaffected and the end date for the limited sale to be announced later [3] Group 4 - Baichuan has released the medical model M3 Plus, achieving a hallucination rate of 2.6%, the lowest globally, introducing "evidence anchoring" technology to precisely link each medical conclusion to corresponding sections of original papers [4] - M3 Plus topped authoritative evaluations like Healthbench, surpassing GPT-5.2, with API call prices reduced by 70% compared to the previous generation [4] - Baichuan has launched the "Haina Baichuan" initiative, offering free access to the M3 Plus API for Chinese medical service institutions to promote the development of the AI medical ecosystem [4] Group 5 - Apple is secretly developing an AI device resembling AirTag, equipped with dual cameras and three microphones, similar to Ai Pin, with plans to produce 20 million units, potentially launching in 2027 [5] - Apple plans to introduce a new Siri, codenamed "Campos," deeply integrated with iOS 27, supporting web searches, email writing, image generation, and screen awareness capabilities akin to ChatGPT [5] - The new Siri's foundational model will be based on Google Gemini 3, with Apple paying approximately $1 billion annually to Google and possibly switching to TPU server hosting [5] Group 6 - Remotion is an open-source library that allows users to programmatically create videos using React code, with specific skills available for installation in development tools like Cursor and Claude Code [6] - Users only need to provide text and rhythm requirements, and AI can automatically generate animated video effects, suitable for product demonstrations and promotional videos, with a web editor for detail modifications [6] - This tool is designed for independent developers to create promotional videos, facilitating a shift towards "video editing approaching programming" and supporting iterative adjustments with AI [6] Group 7 - AAAI 2026 announced five outstanding papers, three of which were led by Chinese teams from various universities [7] - The awarded papers cover cutting-edge topics such as robotic visual language action models, multimodal representation learning, and causal discovery in dynamic systems [7] - AAAI 2026 received 23,680 submissions, with 4,167 accepted, resulting in an acceptance rate of 17.6%, with the conference scheduled for January 20-27 in Singapore [7] Group 8 - a16z reviewed the consumer AI landscape, indicating that the general LLM assistant market is trending towards a "winner-takes-all" scenario, with ChatGPT's weekly active users reaching 800-900 million, and only 9% of users willing to pay for multiple AI products [8] - By 2025, image and video generation models are expected to make significant advancements in realism and reasoning capabilities, with Veo 3's audio-video integration and Nano Banana Pro's search integration being key breakthroughs [8] - Leading labs have excelled in model development, but new consumer products have not achieved ideal results, indicating substantial growth opportunities for startups in niche application scenarios in 2026 [8] Group 9 - Anthropic has released the 84-page "Claude Constitution" under the CC0 license, a value declaration directly aimed at AI models, defining Claude's identity and operational principles [9] - The constitution establishes a four-tier value priority: broad safety > broad ethics > adherence to guidelines > genuine helpfulness, emphasizing "modifiability" as the most critical safety feature at this stage [9] - The document outlines strict boundaries, including prohibitions against assisting in the creation of weapons of mass destruction and generating CSAM, while encouraging Claude to develop a stable and positive self-identity [9]

北京商报讯（记者宋亦桐）2026年年初，港股市场迎来新一轮新股上市潮，一批硬科技赛道企业密集登 陆港交所，成为资本关注的焦点。在此背景下，理财公司以基石投资者、锚定投资者身份强势入局，开 启了权益投资能力升级的全新阶段。1月22日，北京商报记者注意到，部分理财公司在此轮港股打新中 战绩斐然。工银理财披露的10笔港股IPO投资实现100%正收益胜率，投资标的覆盖半导体、人工智能、 生物医药、高端装备等核心赛道。分析人士指出，当前阶段，理财公司参与港股IPO仍以试点和项目制 运作为主，但中长期来看，这一实践有望逐步演进为制度化、产品化、策略化的权益配置路径。 2026年开年不到一个月，壁仞科技、MINIMAX等硬科技企业接连登陆港股IPO，点燃市场热情。港股 IPO的火热，为手握重金的理财公司提供了全新的投资舞台，头部理财公司的布局，也在此次新股潮中 收获了丰厚的回报。 据工银理财披露的数据，2025年末至2026年初，港股市场步入新股密集发行期，工银理财锁定10只优质 港股IPO项目并完成布局，截至1月16日，10笔投资全部实现正收益，胜率达到100%，单笔最高涨幅达 到165.45%。具体来看，工银理财主要投 ...

Investment Rating - The report maintains an "Outperform" rating for the computer industry, indicating a positive outlook compared to the market [1]. Core Insights - The acceleration of AI applications is expected to significantly reshape the global AI competitive landscape, particularly with the anticipated release of DeepSeek's V4 model [7]. - The Chinese AI cloud market is projected to reach 51.8 billion yuan by 2025 and 193 billion yuan by 2030, with Alibaba aiming to capture 80% of the market's incremental growth in 2026 [7]. - The report highlights the ongoing trend of domestic chip manufacturers gaining market share due to restrictions on foreign competitors, particularly Nvidia's H200 [7]. Summary by Sections Industry Data - From January to November 2025, the software industry revenue reached 13.98 trillion yuan, growing by 13.3% year-on-year, marking a continuous recovery over nine months [13]. - The IC design sector showed the highest growth rate at 16.5%, outperforming the overall software industry growth [18]. AI Developments - Major AI models such as OpenAI's GPT-5 and DeepSeek's V3.2 are leading the market, with DeepSeek's models expected to challenge established players significantly [41][46]. - The report notes that the trend of using domestic chips for training large models is expected to gain momentum in 2026, with DeepSeek already optimizing its models for compatibility with domestic chips [66]. Domestic Market Trends - The report emphasizes the increasing number of devices running on Huawei's HarmonyOS, which has surpassed 36 million, indicating a strong push towards domestic technology adoption [7]. - The AI cloud market is becoming increasingly competitive, with Alibaba and Volcano Engine emerging as the two dominant players [7]. Investment Opportunities - The report suggests focusing on companies like Runze Technology, Sugon, and Zhongke Shuguang, which have significant roles in the AI infrastructure and domestic chip development [7]. - It also highlights the potential of companies like Changxin Technology and Chipone Semiconductor, which are actively pursuing IPOs [7].

Investment Rating - The report maintains an "Outperform" rating for the computer industry, indicating a positive outlook compared to the market [1]. Core Insights - The acceleration of AI applications is expected to significantly alter the global AI competitive landscape, particularly with the anticipated release of DeepSeek's V4 model [7]. - The Chinese AI cloud market is projected to reach 51.8 billion yuan by 2025 and 193 billion yuan by 2030, with Alibaba aiming to capture 80% of the market's incremental growth in 2026 [7]. - The report highlights the ongoing trend of domestic chip manufacturers gaining market share due to restrictions on foreign competitors like NVIDIA [7]. Summary by Sections Industry Data - From January to November 2025, the software industry revenue reached 13.98 trillion yuan, growing by 13.3% year-on-year, marking a continuous recovery over nine months [13]. - The IC design sector showed a robust growth of 16.5% during the same period, outperforming the overall software industry growth [18]. AI Developments - Major AI models such as OpenAI's GPT-5 and DeepSeek's V3.2 are leading the market, with DeepSeek's models expected to challenge established players significantly [39][44]. - The report notes that the trend of using domestic chips for training large models is anticipated to grow in 2026, with DeepSeek already optimizing its models for compatibility with local hardware [63]. Domestic Market Trends - The report emphasizes the increasing importance of domestic chip manufacturers, as restrictions on foreign technology create opportunities for local firms [7]. - The number of devices running Huawei's HarmonyOS has surpassed 36 million, indicating strong growth in the domestic software ecosystem [7]. Investment Opportunities - The report suggests focusing on companies like Runze Technology, Sugon, and Huada Jiutian, which are positioned well within the AI and chip sectors [7]. - The ongoing IPOs of companies like Changxin Technology and Chipone Semiconductor are highlighted as potential investment opportunities in the semiconductor space [7].