证券研究报告 | 公司点评 gszqdatemark 2026 02 03 年 月 日 寒武纪-U（688256.SH） 25 年业绩预告全年转盈，规模效应加速凸显 寒武纪发布 2025 年年度业绩预告。预计 2025 年收入端实现 60-70 亿 元，同比增长 410.87-496.02%；实现归母净利润 18.5-21.5 亿元，同比扭 亏为盈；实现扣除非经常性损益后的归母净利润为 16-19 亿元。 盈利拐点确立，规模效应已现。AI 算力需求持续攀升叠加云端产品线技术 成熟度高，带动公司业务大规模商业化落地，在运营商、金融、互联网及 其他垂直行业等领域持续发力，营业收入较上年同期大幅增长，带动公司 整体经营业绩提升，净利润实现扭亏为盈，规模效应凸显。 单季度收入中枢环比增长，供应链保障有力。根据业绩预告，公司 25Q4 营收为 13.93-23.93 亿元，营收中枢环比 25Q3 有所增长。与此同时，25Q3 末公司存货余额为 37.29 亿，环比提升 39%，存货重回快速增长通道，供 给端改善显著，为订单爆发做好供应链保障。 豆包大模型每日 token 使用量翻倍，CSP 大厂需求持续引爆增长。202 ...

Core Insights - The domestic AI chip industry is rapidly advancing, driven by both policy support and market demand, creating significant opportunities for local computing ecosystems [1][9] - The competition in the AI chip sector is increasingly centered around ecosystem development, with a focus on creating alternatives to NVIDIA's dominant CUDA platform [2][4] Group 1: Industry Dynamics - The AI chip market has been historically dominated by NVIDIA, which has established a near-monopoly through its CUDA software ecosystem [2] - Domestic companies, such as Cambricon, are making strides in developing competitive software platforms like Cambricon NeuWare, which aims to provide compatibility with mainstream frameworks [4][6] - The shift towards domestic chips is being accelerated by supply chain uncertainties and geopolitical factors, prompting local AI firms to seek alternatives [9] Group 2: Technological Developments - Cambricon NeuWare has achieved compatibility with the latest versions of PyTorch and Triton, facilitating easier migration of models from NVIDIA GPUs to MLU [4][6] - The platform includes a comprehensive software stack and tools for performance analysis, enhancing the usability for developers [5] - Cambricon is also focusing on supporting a wide range of models for both training and inference, demonstrating its capability in handling complex AI tasks [6][7] Group 3: Market Opportunities - The domestic AI chip market is projected to grow significantly, with estimates suggesting a rise from 142.54 billion yuan in 2024 to 1,336.79 billion yuan by 2029, reflecting a compound annual growth rate of 53.7% [10] - The increasing penetration of domestic chips in AI server markets is expected, with local suppliers projected to capture 40% of the market share by 2025 [9][10] - The supportive policies from the government, such as the "Artificial Intelligence +" initiative, are expected to further bolster the growth of the domestic AI chip ecosystem [9]

Group 1: Core Insights - The article highlights the top 50 keywords in AI for the week, providing a comprehensive overview of the latest developments in the industry [2][3][4]. Group 2: Computing Power - Key developments in computing power include collaborations between Cambricon and NeuWare, OpenAI and AWS, and the introduction of space AI servers by NVIDIA and Google's space AI initiatives [3][4]. Group 3: Models - Significant model advancements include the release of Composer-1 by Cursor, linear attention by Kimi, and the preview of Gemini 3 Pro by Google, among others [3][4]. Group 4: Applications - Various applications are emerging, such as OpenAI's Bug agent, Canva's creative operating system, and AI smart glasses by Rokid, showcasing the diverse use cases of AI technology [3][4]. Group 5: Opinions - Notable opinions expressed include Elon Musk's views on intelligence elevation, Stanford University's insights on speech therapy, and discussions on AI's impact on labor by Geoffrey Hinton [4].

Group 1: Generative AI Developments - Cambricon has launched the Cambricon NeuWare foundational software platform, fully compatible with the latest version of PyTorch and Triton operator development language, enabling rapid migration of user models and custom operators [1] - OpenAI has tightened its usage policy, stating that ChatGPT will no longer assist in providing professional advice in high-risk fields such as healthcare, law, and finance, due to rising legal risks and global compliance pressures [2] - Meituan has open-sourced its multimodal model LongCat-Flash-Omni, which has a total parameter count of 560 billion and an active parameter count of 27 billion, achieving state-of-the-art results in multimodal benchmark tests [3] Group 2: AI Applications and Innovations - Baidu's Wenxin app has introduced a "Magic Comic" feature that allows users to generate multi-page AI comics from a single sentence or photo within two minutes, supporting custom character designs and various artistic styles [4] - Cartesia has launched the new Sonic-3 voice model, supported by a $100 million investment from Nvidia, which can generate voice in 42 languages and over 500 tones, with a response time of under 190 milliseconds [5][6] - Turbo AI, founded by two 20-year-old college dropouts, has seen its user base grow from 1 million to 5 million in six months, generating annual recurring revenue in the eight figures while serving clients like Goldman Sachs and McKinsey [7] Group 3: AI Tools and Market Trends - A review of mainstream AI browsers indicates a division between progressive browsers (Chrome/Edge) and radical browsers (ChatGPT Atlas/Perplexity Comet/Dia), each with unique strengths and weaknesses [8] - Rokid has partnered with BOLON to launch the BZ5000 AI smart glasses, which weigh only 38 grams and feature a 12-megapixel camera, emphasizing localized services through its YodaOS system [9] - AI expert Fei-Fei Li has called for universities and non-profit organizations to reclaim the mission of advancing AI as a public good, highlighting the shift from open research to closed commercial competition [10][11] Group 4: Data and Market Opportunities - a16z partners emphasize the importance of building "data moats" in fragmented, sensitive, or hard-to-access fields, with examples like VLex and OpenEvidence showcasing proprietary data systems as competitive advantages [12]

Core Viewpoint - The article highlights the remarkable journey of Chen Tian Shi, the CEO of Cambricon Technologies, who transformed from a gaming-obsessed student to a leading figure in the AI chip industry, culminating in the company's recent stock surge and market recognition as a "chip king" in China's capital market. Group 1: Early Life and Education - Chen Tian Shi was born in June 1985 in Nanchang, Jiangxi, into an intellectual family, with parents who encouraged a broad education [1] - Despite his intelligence, he struggled academically during his early years, often distracted by video games, which led to a reputation as a "poor student" [2][4] - His turning point came during his undergraduate studies at the University of Science and Technology of China, where he shifted his focus to serious research in artificial intelligence [4] Group 2: Research and Development - After completing his PhD in 2010, Chen joined the Chinese Academy of Sciences, where he and his brother proposed the concept of AI-specific chips, despite initial skepticism from the academic community [8] - Their perseverance led to significant breakthroughs, including the development of the world's first deep learning processor prototype, "Cambricon" [8] Group 3: Entrepreneurship and Growth - In 2016, Chen founded Cambricon Technologies, which quickly became a unicorn startup in the AI chip sector, launching several innovative products [9] - The company gained significant recognition after its collaboration with Huawei, integrating its chips into Huawei's flagship devices, leading to a rapid increase in valuation [9] Group 4: Challenges and Strategic Shifts - In 2019, Cambricon faced a major setback when Huawei shifted to self-developed architectures, threatening its primary revenue stream [10] - Chen made a strategic decision to diversify the company's offerings, focusing on cloud AI chips and building a comprehensive product line, which ultimately led to a successful IPO in 2020 [14] Group 5: Market Recognition and Future Outlook - Following its IPO, Cambricon experienced significant challenges, including being placed on the U.S. Entity List, which restricted access to critical resources [15] - Despite these challenges, the company has shown resilience, with revenues increasing significantly and achieving profitability for the first time in 2025 [15] - Chen's vision remains focused on establishing Cambricon as a foundational player in the AI chip industry, emphasizing the importance of perseverance and long-term commitment [18][20]