（文/陈济深 编辑/张广凯） 12月19日，北京智谱华章科技股份有限公司（智谱）正式向港交所递交招股书，打响了中国大模型企业上市的第一 枪。 这不仅意味着智谱有望成为"全球大模型第一股"，也标志着中国大模型赛道正式迈入资本化新阶段。 作为全球AI新势力中的重要玩家，招股书也第一次揭示了这家明星独角兽商业化的"冰火两重天"。 过去三年，公司收入分别为0.57亿元、1.25亿元、3.12亿元。2025年上半年持续加速，短短6个月就入账1.9亿元，同比 增长325%。 收入增长的背后是智谱惊人的烧钱速度：过去三年，智谱经调整净亏损分别为0.97亿、6.21亿以及24.66亿元。 不过值得注意的是，智谱在B端业务占据主导的情况下，面对传统B端大厂和互联网巨头的竞争，长期维持住了超过 50%的毛利率。 随着2025年字节火山引擎、阿里云、百度等互联网巨头纷纷下场加码AI大战，智谱的这份招股书提供了一个观察中国 大模型企业在全球AI淘汰赛下真实状态的绝佳切片。 巨头夹缝中的"独立样本" 智谱的核心团队成员，基本都来自清华大学计算机系知识工程实验室，长期从事大规模预训练模型、语义大数据分 析、智能问答等领域的研究。 有这样 ...

Core Insights - The AI market in China is rapidly evolving, with Huoshan Engine emerging as a leading player, particularly in the model-as-a-service (MaaS) sector, where it holds the largest market share domestically and ranks third globally [2][3] - The daily token usage of the Doubao model has surged to over 50 trillion, marking a tenfold increase compared to the previous year [1][4] - The focus for 2025 in the AI market will be on multimodal capabilities and agents, with Huoshan Engine launching several new products centered around these themes [3][6] Market Position and Growth - Huoshan Engine has established itself as a significant force in the AI sector, with projected revenues exceeding 200 billion in 2024, reflecting a growth rate of over 60% [6][23] - The company aims to simplify model usage by integrating multiple capabilities into a single API, contrasting with competitors who offer separate models for different functions [26][27] Product Innovations - The newly launched Seedance 1.5 pro video generation model emphasizes immediate usability, capable of producing synchronized audio-visual content without extensive post-production [8][15] - The model's advancements include improved lip-sync accuracy and enhanced immersion, making it particularly suitable for diverse content creation [13][21] Competitive Landscape - The AI video model market is characterized by rapid iteration, with companies focusing on producing fully publishable works rather than just raw video segments [7][9] - Huoshan Engine's approach to model training and optimization has led to a tenfold increase in inference speed, significantly reducing costs and enhancing performance [31][30] Future Directions - The company is exploring innovative billing models, such as the "AI Savings Plan," which offers tiered discounts to help businesses reduce costs by up to 47% [32][33] - Huoshan Engine is committed to building a comprehensive AI infrastructure that enables businesses to easily adopt advanced AI capabilities, aiming to make AI assistants as ubiquitous as websites and apps [38][39]

Core Viewpoint - The AI market is rapidly evolving, with major players like Volcano Engine leading the way in model consumption and innovation, particularly in the areas of multi-modal capabilities and AI agents [3][5][51]. Group 1: Market Growth and Trends - As of December, the daily token usage of Doubao model has surpassed 50 trillion, representing a growth of over 10 times compared to the same period last year [3]. - By 2025, the token usage is projected to reach 16.4 trillion, indicating significant growth potential in the AI market [4]. - The competition among cloud vendors for "AI cloud supremacy" is intensifying, with major updates from companies like Google and OpenAI [4]. Group 2: Product Innovations - Volcano Engine has released key products focusing on multi-modal capabilities and AI agents, including the Doubao flagship model 1.8 and the video generation model Seedance 1.5 pro [5][6]. - The Seedance 1.5 pro model emphasizes the ability to produce "publishable complete works," showcasing advancements in video generation technology [10][11]. - The model's improvements in voice and image synchronization have made it a standout in the market, achieving high levels of usability with minimal input [11][18]. Group 3: Business Model and Strategy - Volcano Engine aims to simplify model usage by integrating multiple capabilities into a single API, reducing complexity for clients [38][39]. - The company is focusing on enhancing the efficiency of model training and deployment, with the Seedance 1.5 pro achieving over a 10-fold increase in inference speed [46]. - A new billing model, "AI Savings Plan," has been introduced to help enterprises save up to 47% on costs, reflecting a shift towards value-based pricing [47][48]. Group 4: System Engineering and Infrastructure - The competition in AI infrastructure has shifted from merely comparing model capabilities to a broader system engineering challenge [51]. - Volcano Engine is developing a comprehensive AI infrastructure that includes both the core model (Doubao) and operational tools (AgentKit) to facilitate easier deployment for enterprises [53]. - The goal is to enable every enterprise to have its own AI assistant, akin to having a website or app, supported by a complete ecosystem [54].

事实上，豆包手机助手背后的技术基础是Agent（智能体）能力的落地。而在大会发布中，谭待也强调 了豆包1.8在Agent能力方面"Tool Use（工具使用）能力，复杂指令遵循能力、OS Agent（操作系统智能 体）能力都实现了大幅增强。" 谭待特别演示了豆包1.8进行"全网比价选购耳机"的能力，贝壳财经记者注意到，该能力在此前的豆包 手机发布视频中也有出现。根据演示，模型接到任务后，首先规划了任务处理流程，然后理解屏幕，并 调用10多个工具，选好匹配的耳机；最后在多个电商平台上进行比价，找到了最低价的耳机。 值得注意的是，豆包1.8还展示了长视频理解能力，如通过1小时4分钟的监控画面找到"把车刮花的肇事 者"，谭待表示，同样的能力还能够广泛地应用到在线教育、安全巡检、产品质检等场景中，帮助企业 提升业务效率，降低运营成本。 12月18日，在因现场人数过多而不得不关闭大门、限制人流的Force原动力大会上，火山引擎总裁谭待 公布了一组数据：截至今年12月，豆包大模型的日均tokens调用量已经超过50万亿，相比去年12月，实 现了超过十倍的增速。挟此"爆发"之势，火山引擎宣布豆包大模型升级至1.8，面向多模 ...

Core Insights - The AI sector is experiencing a significant capital surge in 2025, with companies like Zhipu and MiniMax vying for the title of "first stock of large models," highlighting the industry's growing prominence [1] - A value gap exists where companies invest heavily in AI but many remain stuck in pilot phases without generating tangible financial impacts [1] - The market is shifting towards the "second half" of model value realization, with companies facing the dilemma of high investment costs versus the fear of missing out on technological advancements [1] Group 1: Market Dynamics - The transition from "selling model parameters" to "delivering MaaS (Model as a Service)" allows companies to focus on business value rather than the risks of model iteration [2] - The competition in the AI "second half" is characterized by a shift from demo showcases to a battle of foundational models as the basis for enterprise AI deployment [4] - A dramatic market reshuffle is occurring, with Anthropic's Claude series leading the enterprise-level LLM API market with a 32% usage share, while OpenAI's share has dropped from 50% to 25% [4][9] Group 2: Financial Growth and Strategy - Anthropic's "enterprise-first" strategy has led to a remarkable increase in annual recurring revenue (ARR), soaring from $1 billion to $5 billion within months [9] - Traditional cloud giants like Alibaba Cloud are adopting a "build kitchen" strategy, offering a full-stack solution from IaaS to MaaS, while engaging in price wars to attract customers [10][11] - Smaller firms are finding opportunities by focusing on niche markets and differentiating their offerings rather than competing directly with giants [12][14] Group 3: Performance and Efficiency - As of 2025, companies are prioritizing model performance and efficiency over mere token price reductions, indicating a shift in focus towards effective AI solutions [13] - Zhipu's new models, GLM-4.5 and GLM-4.6, have seen a rapid increase in token usage, particularly in coding tasks, attracting significant developer interest [14][27] - The demand for high-performance models in critical applications, such as coding and financial analysis, is driving companies to pay premiums for improved accuracy and reliability [18][21] Group 4: Future Trends and Implications - The emergence of MaaS is not just a commercial choice but a technological necessity, as companies must navigate the complexities of AI deployment strategies [17] - The market is witnessing a shift where foundational models are becoming the primary applications, with the potential for models to evolve into autonomous agents [22][24] - The valuation of AI companies is changing, with a growing recognition that foundational models represent a new form of labor rather than just software, leading to a potential revaluation of independent firms in the sector [26][28]

Core Viewpoint - The emergence of large models is transforming the landscape of China's cloud computing industry, with companies like Volcano Engine gaining significant traction in the AI application development platform market [1][2][15]. Group 1: Market Position and Performance - Volcano Engine ranked first in the "Challenger" quadrant of Gartner's Magic Quadrant for AI application development platforms, showcasing its strong capabilities in model services [2][6]. - As of mid-2025, Volcano Engine is projected to hold a 49.2% market share in China's public cloud large model service market, indicating its leading position [7]. - The company aims to achieve a revenue target of over 20 billion RMB this year, reflecting a growth rate exceeding 100% [16]. Group 2: Strategic Focus and Investments - Volcano Engine has aggressively invested in AI over the past three years, positioning itself to capitalize on the rapid growth of the AI sector [7][8]. - The company has shifted its focus towards Model as a Service (MaaS), recognizing the potential for high margins and growth in this area [6][11]. - The integration of AI capabilities into existing cloud services is seen as a critical strategy for competing against established players in the market [16]. Group 3: Competitive Landscape and Challenges - The cloud computing market is characterized by high entry barriers due to established players having strong customer ties and high data migration costs [8][9]. - Volcano Engine faces competition from major cloud providers like Alibaba Cloud, which are also investing heavily in AI and large model services [16]. - The overall market for MaaS is still in its early stages, with a projected size of only 1.29 billion RMB by mid-2025, despite a staggering growth rate of 421.2% [15]. Group 4: Future Outlook and Innovations - The company is exploring new growth avenues, particularly in the development of intelligent agents, which are expected to create significant economic value beyond traditional applications [15]. - Volcano Engine's strategy includes leveraging AI as a lever to penetrate the existing market, with a significant portion of its revenue expected to come from large model services [16]. - The company is also adjusting its sales strategies to prioritize MaaS products, which offer higher returns compared to traditional cloud services [11].

Core Viewpoint - The article discusses the upcoming IPO of Zhipu AI, highlighting its strategic transformations and challenges as it prepares for the public offering in 2025. The company is undergoing significant internal changes, including management restructuring and a shift in business model to enhance efficiency and competitiveness in the AI sector [5][25]. Group 1: IPO Preparation - Zhipu AI is set to enter its IPO preparation phase, aiming for completion by October 2025, with considerations for dual listings in Hong Kong and A-shares [5]. - The company has recently completed a new round of financing and launched new models, indicating proactive strategic moves ahead of its IPO [5]. Group 2: Internal Restructuring - Following the departure of former COO Zhang Fan, Zhipu AI has undergone substantial organizational changes, merging teams and redefining roles to streamline operations [6][8]. - The restructuring reflects a deeper transformation within the company, aiming to integrate G-end and B-end resources for better synergy and efficiency [14][15]. Group 3: Financial Backing and National Support - Zhipu AI has secured significant funding from various national and local government-backed entities, enhancing its "national team" profile within the AI sector [9][11]. - The company has raised substantial amounts in recent financing rounds, including a notable 30 billion yuan in June 2024 and 200 billion yuan in September 2024 [10]. Group 4: Business Model Transformation - Zhipu AI is shifting its business model from heavy delivery to a "light delivery" approach, aiming to avoid the pitfalls of previous AI companies that became overly reliant on customized solutions [18][19]. - The introduction of the CoCo intelligent agent represents a move towards standardization, allowing for more efficient service delivery and cost reduction for clients [19][20]. Group 5: Strategic Focus on MaaS - The company is prioritizing the development of its MaaS (Model as a Service) platform, which is crucial for maintaining competitiveness against larger tech firms [22][24]. - Zhipu AI's strategy to enhance its MaaS capabilities is seen as essential for providing a comprehensive solution from model deployment to service integration, which is vital for its survival in a competitive landscape [24][25]. Group 6: Industry Context and Challenges - The article highlights the broader challenges faced by AI unicorns in China, including intense competition and the need for sustainable business models amid fluctuating market conditions [25][26]. - Zhipu AI's adjustments reflect a common trend among AI companies to balance technological aspirations with commercial realities, as they navigate the complexities of the evolving AI landscape [25][26].

Core Insights - Zhiyu is set to enter its IPO sprint phase in one month, with plans to complete its counseling program by October 2025, aiming for a dual listing in both Hong Kong and A-shares [2][3] - The company has undergone significant internal restructuring over the past six months, particularly following the departure of its former COO, Zhang Fan, which has led to a reorganization of teams and responsibilities [3][4][10] - Zhiyu is shifting its business model from a reliance on B-end clients to a more integrated approach that combines G-end and B-end resources, aiming to enhance operational efficiency and revenue generation [11][12] Financing and Investment - Zhiyu has secured substantial funding from various sources, including a $3 billion Series C round in June 2024 and a $20 billion Series D round in September 2024, indicating strong investor confidence [6][7] - The company has also received strategic financing from multiple state-owned enterprises, enhancing its "national team" attributes and solidifying its market position [5][8] Organizational Changes - The departure of key personnel, including Zhang Fan and Chief Ecosystem Officer Liu Jiang, reflects broader strategic adjustments within the company as it prepares for its IPO [9][10] - The restructuring aims to consolidate G-end and B-end operations under a unified management strategy, enhancing collaboration and resource sharing [11][12] Business Model Transformation - Zhiyu is transitioning towards a "light delivery" model to avoid the pitfalls of previous AI companies that became overly reliant on customized solutions, which led to cash flow issues [14][15] - The introduction of the CoCo intelligent agent represents a move towards standardization, allowing for more efficient service delivery and reduced operational costs [14][15] Strategic Focus on MaaS - The company is prioritizing the development of its MaaS (Model as a Service) platform, which is crucial for maintaining competitiveness against larger tech firms like Alibaba and ByteDance [17][19] - This strategic pivot aims to create a comprehensive solution that integrates model capabilities with service delivery, addressing the challenges posed by external competition [18][20] Conclusion - Zhiyu's recent transformations reflect the broader challenges faced by AI unicorns in balancing technological aspirations with commercial realities, as it seeks to establish a sustainable business model ahead of its IPO [20][21] - The company's ability to navigate these changes will be critical in determining its future success and positioning within the competitive landscape of the AI industry [21]

Group 1 - He Guoshuai, the General Manager of the Intelligent Terminal Division at Zhipu AI, is leaving the company and will join a startup founded by former COO Zhang Fan, focusing on commercialization [2][3][4] - He Guoshuai played a significant role in Zhipu's commercialization efforts, notably improving the performance of the South China region from last to first place within six months [3] - Zhipu AI's COO Zhang Fan also left the company in June to pursue entrepreneurship in the AI Agent field, with his new project receiving investment support from Zhipu [6] Group 2 - The management structure of Zhipu's commercialization department underwent changes earlier this year, with a reorganization that affected how business responsibilities were divided [6] - There are indications that Zhipu's management may experience frequent adjustments in July, following the recent departures and restructuring [6]

Core Insights - The Alibaba Modao community has been recognized as China's largest AI open-source community, achieving significant growth in model availability and user engagement [1][3][4] Group 1: Community Growth and Achievements - The Modao community has surpassed 70,000 open-source models, marking a 200-fold increase since its inception [1] - User base has expanded to 16 million, representing a nearly 16-fold growth compared to April 2023 [1] - Over 500 institutions have contributed to the community's development [1] Group 2: Platform Capabilities and Offerings - Modao community serves as a comprehensive AI platform, supporting developers in model experience, download, tuning, training, inference, and deployment across various fields [3] - The community provides over 4,000 types of MCP services and debugging tools, enhancing developer support [1][4] - Free computing power is offered to developers to facilitate model usage [3][4] Group 3: Open Source and Collaboration - The community emphasizes an open, neutral, and non-profit organizational structure, promoting inclusive AI technology [4] - The introduction of the MCP Square provides thousands of MCP services and hosting options, with open interfaces for third-party integration [4] - A developer badge incentive program has been launched to reward outstanding contributions, offering free GPU computing power and other benefits [4] Group 4: Future Directions and Innovations - The launch of models like DeepSeek and Qwen highlights the growing importance of open-source in AI innovation [3] - The CTO of Alibaba Cloud emphasized the significance of "cloud collaboration" and the need for further research on integrating edge models with cloud capabilities [3]