Core Viewpoint - MiniMax has experienced a remarkable stock price surge since its debut on the Hong Kong stock market on January 9, reaching a market capitalization of over HKD 200 billion following the release of its new model M2.5, which has drawn significant attention and investment returns for its backers, particularly miHoYo, which has seen a substantial profit from its early investment [1][3][6]. Group 1: MiniMax's Performance and Market Response - MiniMax's stock price increased by over 13% on February 13, 2023, following the announcement of its new model M2.5, which is said to match the performance of Claude Opus 4.6 [1][2]. - The market responded quickly, with MiniMax's market capitalization exceeding HKD 200 billion [3]. - The company has achieved a unique position in the Chinese large model landscape, having gone from establishment to IPO in just four years [2]. Group 2: Investment and Returns - miHoYo, a gaming company, was one of the angel investors in MiniMax, investing at a post-money valuation of USD 200 million, and has since seen its investment return exceed 100 times [4][6]. - miHoYo's stake in MiniMax increased to 7.34% during subsequent funding rounds, and it currently holds approximately 5.24% of the company post-IPO, translating to a book value of nearly RMB 10 billion [5][6]. - Other notable investors in MiniMax include Hillhouse Capital, IDG Capital, Sequoia China, Alibaba, Tencent, and several others, indicating a broad and significant backing for the company [6]. Group 3: miHoYo's Financial Performance - miHoYo has disclosed that it has paid a total of RMB 15 billion in taxes from 2021 to 2025, reflecting its substantial revenue growth [11]. - In 2021, miHoYo reported a revenue of RMB 27.34 billion and a net profit of RMB 16.145 billion, with tax contributions increasing from RMB 1.8 billion in 2021 to a projected RMB 15 billion by 2025 [11][12]. - The company has made significant investments in various sectors, including technology and gaming, and has been actively involved in the IPO process of other companies [12][14].

Core Viewpoint - The company positions itself not merely as a "shovel vendor" or "accelerator," but as a comprehensive provider of services that facilitate the deployment and operation of large models in AI applications [2]. Group 1: Company Services - The company utilizes its MaaS platform to complete multi-model debugging and deployment, offering services such as model invocation and elastic scaling, which provide a convenient pathway for the widespread adoption of large models [2]. - The company adapts to heterogeneous computing power from providers like Nvidia and Ascend, constructing high-power data centers and intelligent computing clusters, which serve as the foundational support for the stable operation of AI applications [2].

Core Viewpoint - MiniMax has launched its new generation text model MiniMaxM2.5, showcasing significant improvements in programming capabilities and performance metrics compared to its predecessor [2][4]. Group 1: Model Performance - In programming capabilities, MiniMaxM2.5 achieved scores of 80.2% on the SWE-BenchVerified and 51.3% on the Multi-SWE-Bench, marking a significant enhancement over the previous generation [2]. - The model outperformed Opus4.6 in multi-language complex environments, demonstrating its "native Spec capability" by actively decomposing architecture and functional planning before coding [2]. Group 2: Tool Utilization and Search Capabilities - MiniMaxM2.5 can automatically handle complex tasks, achieving a 20% performance improvement in tasks such as BrowseComp and WideSearch compared to the previous model [4]. - In office scenarios, the model showed significant capability enhancements in high-level tasks involving Word, PPT, and Excel financial modeling, achieving an average win rate of 59.0% in the GDPval-MM evaluation framework against mainstream models [4]. Group 3: Speed and Cost Efficiency - The M2.5-lightning version supports an output speed of over 100 TPS, approximately double that of mainstream models, with input costs around $0.3 per million tokens and output costs about $2.4 per million tokens [4]. - Calculating costs based on a continuous output of 100 tokens per second, running M2.5 for one hour would cost approximately $1, while at 50 tokens, the cost would be around $0.3 [4]. - MiniMax believes that when performance and cost are no longer constraints, the economic model for large-scale deployment of agents will fundamentally change [4].

Core Insights - The AI video generation model Seedance 2.0 from ByteDance and the new model GLM-5 from Zhiyu have sparked ongoing excitement in the market, leading to significant stock price increases for related companies [2] - The demand for computing power has surged, with the semiconductor equipment index rising by 2.02% and specific companies like Fuchuang Precision and Jingyi Equipment seeing gains of 11.6% and 5.92% respectively [2] Group 1: Market Reactions - On February 13, cultural media theme index saw Light Media rise by 15.3% and Zhangyue Technology by 10%, while AI application index recorded Shen Technology up by 10.02% and Fengyuzhu up by 9.98% [2] - Despite declines in the Hang Seng Index and Hang Seng Tech Index, Zhiyu and MiniMax saw stock increases of 13.78% and 9.86%, with Zhiyu's market capitalization reaching a new high since its listing [2][3] Group 2: Pricing Strategies - Zhiyu announced a price increase for its GLM Coding Plan subscription service, with an overall price hike starting from 30% due to strong market demand and the need for enhanced service quality [4] - The CEO of ListenHub noted that the price increase reflects the growing scarcity and value of tokens as model capabilities improve, confirming earlier expectations of rising costs [5] Group 3: Model Advancements - The GLM-5 model has doubled its parameters from 355 billion to 744 billion and increased its pre-training data from 23 terabytes to 28.5 terabytes, enhancing its capabilities significantly [5] - Multi-modal models like Seedance 2.0 and Nano Banana Pro are consuming tokens at rates far exceeding pure text, indicating a substantial increase in demand for computational resources [6] Group 4: Competitive Landscape - The recent advancements in domestic models like Seedance 2.0 and GLM-5 indicate a shift towards practical applications in work and life, with a focus on video generation and coding [8] - Zhiyu's GLM-5 is directly competing with Anthropic's Claude Opus 4.5, although it still lags behind in certain advanced capabilities [9] - Chinese AI models are rapidly closing the gap with their U.S. counterparts, with significant contributions from companies like DeepSeek and Alibaba, showcasing strong global competitiveness [9]

发布新模型后，智谱"A+H"上市布局有新进展？2月13日，南都N视频记者从中国证监会官网获悉，智谱撤回了 其于2025年4月提交的辅导备案，并办理了新的辅导备案登记。根据最新备案登记信息，智谱拟首次公开发行股 票并在上海证券交易所科创板上市，IPO辅导机构变更为两家券商，由原来的中金公司变更为国泰海通和中金公 司两家。 南都N视频记者向智谱方面求证，截至发稿暂无回应。值得注意的是，智谱H股股价已连续五个交易日上涨，截 至发稿，智谱2月13日股价涨超17%，报473.8港元/股，总市值达2112亿港元，为上市首日的3倍多。 智谱股价连续多个交易日上涨。 智谱去年已开启A股上市历程。2025年4月，智谱在北京证监局办理辅导备案，当时由中国国际金融股份有限公 司担任辅导机构。 今年1月8日，智谱（02513.HK）正式在香港联合交易所挂牌上市，成为"全球大模型第一股"。 智谱不断推进大模型上新进程。2月12日，南都记者获悉，此前一段时间内，让开发者社区广泛关注的代 号"Pony Alpha"匿名模型，系智谱新模型GLM-5。2月12日，新模型已在chat.z.ai平台上线，模型在Coding与 Agent能力上取得 ...

Core Viewpoint - MiniMax has achieved a market capitalization exceeding 200 billion HKD since its listing on the Hong Kong stock exchange, driven by the launch of its new model M2.5, which has garnered significant market attention and investor returns [2][3]. Group 1: MiniMax's Market Performance - MiniMax's stock price surged over 13% on February 13, 2023, following the announcement of its M2.5 model, which is comparable to Claude Opus 4.6 in performance [3]. - The company reached a market valuation of over 200 billion HKD shortly after its IPO, marking a significant milestone in the tech industry [2][3]. Group 2: Investment Landscape - MiHoYo, a gaming company, was a key angel investor in MiniMax, investing at a post-money valuation of 200 million USD, and has seen its investment return exceed 100 times [4][5]. - Other notable investors in MiniMax include Hillhouse Capital, IDG Capital, Sequoia China, Alibaba, Tencent, and several others, indicating a strong backing from prominent venture capital firms [5]. Group 3: MiHoYo's Financial Contributions - MiHoYo has disclosed that it will contribute a total tax of 150 billion CNY over five years from 2021 to 2025, showcasing its significant financial impact [10][11]. - The company's tax contributions have increased from 1.8 billion CNY in 2021 to a projected 150 billion CNY by 2025, reflecting its rapid revenue growth [10]. Group 4: MiHoYo's Investment Strategy - MiHoYo has diversified its investment portfolio, backing various innovative projects, including brain-computer interfaces and private space ventures, indicating a forward-thinking investment approach [12][13]. - The company has positioned itself as a significant player in the venture capital space, actively engaging with young, tech-driven entrepreneurs [13].

Investment Rating - The report gives an investment rating of "Positive" for the computer industry [2]. Core Insights - AI programming is reshaping the core productivity methods in software development, with large model technologies empowering programming tools. The value of AI programming lies in enhancing software development efficiency and quality, lowering technical barriers, and accelerating project iteration cycles [2][11]. - The demand for AI programming is driven by both professional developers upgrading their skills and the empowerment of non-professionals. The global AI code tools market is projected to grow from USD 6.11 billion in 2024 to USD 26.03 billion by 2030, with a compound annual growth rate (CAGR) of 27.1% [2][26]. - The overseas application of AI programming is scaling up, with significant revenue growth validating its explosive potential. Major products like GitHub Copilot and Cursor have seen substantial annual recurring revenue (ARR) growth, indicating a robust market response [2][34]. - Domestic companies are actively entering the AI programming space, with significant product launches and user growth, such as ByteDance's Trae IDE and Alibaba's Tongyi Lingma [2][3]. Summary by Sections AI Coding: Reshaping Software Development - AI programming enhances software development efficiency by automating coding tasks, with IDC data indicating a 35% productivity increase for developers using AI coding tools [11][14]. - The market potential for AI programming is vast, with a projected growth in the global AI code tools market from USD 6.11 billion in 2024 to USD 26.03 billion by 2030, reflecting a CAGR of 27.1% [26][27]. - The technology is evolving from Copilot to Agent models, indicating a shift towards more autonomous programming environments [23][24]. Overseas AI Programming Applications - GitHub Copilot has surpassed 20 million users, demonstrating the effectiveness of its platform ecosystem [42][59]. - Cursor, a leading AI programming IDE, achieved a valuation increase from USD 90 billion to USD 293 billion within six months, highlighting its market potential [60][63]. Domestic Company Developments - ByteDance's Trae IDE has gained over 6 million users globally, while other domestic products like Snapdevelop and EasyDevelop are also making significant strides in the market [3][34]. - The domestic AI programming market is expected to grow from RMB 6.5 billion in 2023 to RMB 33 billion by 2028, with a CAGR of 38.4% [26][28].

Core Insights - MiniMax officially launched its new generation text model, MiniMax M2.5, on February 12, 2026, positioning it as a "native Agent production-level model" in the competitive landscape referred to as the "AI gods battle" [2] Group 1: Model Performance - In programming capabilities, MiniMax M2.5 achieved a score of 80.2% on the SWE-BenchVerified and 51.3% on the Multi-SWE-Bench, showing significant improvement over its predecessor [2] - The model surpassed Opus 4.6 in multi-language complex environments, reaching the industry's best performance [2] - The model exhibits "native Spec capabilities," actively decomposing architecture and functional planning before coding, closely resembling the work patterns of real architects [2] Group 2: Tool Utilization and Search Capabilities - The model can automatically handle complex tasks, achieving better results with lower round consumption in tasks like BrowseComp and Wide Search, showing a 20% improvement over the previous generation [3] - In office scenarios, MiniMax M2.5 demonstrated significant capability enhancements in high-level tasks involving Word, PPT, and Excel financial modeling, achieving an average win rate of 59.0% in the GDPval-MM evaluation framework compared to mainstream models [3] Group 3: Ecosystem Development - MiniMax M2.5 was launched on MiniMax Agent on February 12 and globally open-sourced for localized deployment on February 13, with over 10,000 experts built by users worldwide within a day [3] - The company aims to build a sustainable and scalable Agent ecosystem, referred to as Agent Universe, to enhance model capabilities, generalization, and cost-effectiveness, facilitating the penetration of Agents into various aspects of work and life, from programming to entertainment [3]

Investment Rating - The report indicates a positive outlook for the industry, suggesting a potential for significant growth in the coming months [6][44]. Core Insights - The report highlights a rapid release of computing power demand driven by the dual forces of training and inference, with 2026 identified as a pivotal year for this transition [6]. - Major internet companies are intensifying their competition in AI, leading to a surge in demand for high-quality, multi-modal models, which in turn is expected to drive substantial growth in computing power requirements [11][25]. - The supply side is expected to improve structurally, with domestic computing power resources becoming more abundant, thus supporting the anticipated demand explosion [6][32]. Summary by Sections 1. Rapid Release of Computing Power Demand - The "arms race" in large models continues unabated, with leading internet firms like ByteDance, Alibaba, and Tencent releasing new models with trillions of parameters, enhancing their competitive edge [11][12]. - The demand for inference computing power is rising at an unexpected rate, with significant user growth reported for AI applications, particularly the Doubao app, which reached 226 million monthly active users by December 2025, marking a year-on-year increase of over 200% [6][25]. 2. Supply Side Improvements and Domestic Production Acceleration - The approval of NVIDIA's H200 AI chips for the Chinese market is expected to alleviate computing power shortages for major internet firms, facilitating faster model iterations [32]. - Domestic computing power chips have reached a performance level that is now considered "good enough," with significant advancements in local chip development and deployment [33]. 3. Full-Chain Inflation in Domestic Computing Power - The report predicts that the computing power industry will enter a "full-chain inflation" cycle in 2026, with growth expected across various segments including AIDC, cloud services, and supporting power equipment [38]. - Major tech companies are projected to increase their capital expenditures significantly, with estimates suggesting that the four largest tech firms in Silicon Valley will spend up to $650 billion in 2026 [40]. 4. Related Companies - The report lists several companies as relevant to the industry, including Dongyangguang, Hanwha, Haiguang Information, Wangsu Technology, and others, indicating a broad spectrum of potential investment opportunities [4][44].

Core Insights - Meitu's "Kaipai" has announced the first integration of Seedance 2.0, expected to launch by the end of February 2023, positioning itself as the leading tool in China's voice-over video production market [1][2] Group 1 - "Kaipai" is a voice-over video production tool launched by Meitu in 2023, currently holding the largest user base in China's voice-over video sector [1] - The integration of Seedance 2.0 will embed its generative capabilities directly into the Kaipai workflow, allowing users to utilize advanced models without needing separate applications or additional learning [1] - Concerns regarding the impact of large models on application markets were previously raised, but industry analysts view Kaipai's swift integration as part of Meitu's "model container" strategy, enhancing product functionality and user experience [2]