会议室桌面上，黑白配色的Ropet轻轻晃动身体，绒毛在灯光下泛起柔和的光泽，访谈中，它不时眨动 双眼，偶尔发出小动物般轻柔的叫声，给人带来一种不打扰的"陪伴感"。这款已进入全球50多个国家的 AI（人工智能）萌宠，直到2025年"双11"才在国内首次正式亮相。并且，令人意外的是，当几乎所有AI 硬件都在竞逐语音交互时，Ropet却主动放弃了对话功能。这种"反其道"的决策无疑是大胆且冒险的。 近日，Ropet的开发商北京萌友智能科技有限公司（以下简称萌友智能）联合创始人兼CEO（首席执行 官）何嘉斌接受了《每日经济新闻》"对话未来商业"栏目记者独家专访。 这位曾任职于微软亚洲研究院、百度、物灵科技、字节跳动的创业者，也是一位潮玩爱好者。他向记者 自述过去养猫的经历让他深刻洞察了宠物带来的情绪价值。这让他更坚定地认为，无论是静态的潮玩摆 件，还是真实饲养的宠物，核心价值都在于提供"情绪价值"与"存在感"。 该理念催生了Ropet的设计理念和定位。成立近4年，萌友智能已经完成三轮融资，最新一轮数千万元人 民币A1轮融资，由北京市人工智能产业投资基金领投，峰瑞资本跟投。 现在，这个主动放弃"说话"的AI萌宠，正试图叩 ...

■行业属性：AI硬件 ■核心竞争力：非语音交互设计、全球市场 ■未来关键词：端侧智能、成长性陪伴 会议室桌面上，黑白配色的Ropet轻轻晃动身体，绒毛在灯光下泛起柔和光泽。在访谈中，它不时眨动 双眼，偶尔发出像小动物一般轻柔的叫声，给人带来一种不打扰的"陪伴感"。 这款已进入全球50多个国家的AI（人工智能）萌宠，直到今年"双11"在国内才首次正式亮相。并且，令 人意外的是，当几乎所有AI硬件都在竞逐语音交互时，Ropet却主动放弃了对话功能。 这种"反其道"的决策无疑是大胆且冒险的。 近日，Ropet的开发商北京萌友智能科技有限公司（以下简称萌友智能）联合创始人兼CEO何嘉斌接受 了《每日经济新闻》"对话未来商业"栏目（更多内容，详见专题|对话未来商业记录、预见、让未来发 生）独家专访。 这位曾任职于微软亚洲研究院、百度、物灵科技、字节跳动的创业者，也是一位潮玩爱好者。他向记者 自述过去养猫的经历让他深刻洞察了宠物带来的情绪价值。这让他更坚定地认为，无论是静态的潮玩摆 件，还是真实饲养的宠物，核心价值都在于提供"情绪价值"与"存在感"。 这一理念催生了Ropet的设计理念和定位。成立近4年，萌友智能已经完成 ...

近年来，AI的发展遵循着一个简单直接的规则——模型越大、能力越强。这个被称为"规模法则"的路 径，催生了今天我们看到的各种强大的AI大模型。但同时，随着模型参数量的增大，模型训练的成 本、使用的成本也随之飙升，一定程度上制约了大模型的产业应用进程。 清华大学的一项最新研究成果——大模型"密度法则"20日在国际期刊《自然·机器智能》上作为封面文 章发表，为AI发展指出了一个新方向：AI大模型变强，不应只靠"体型"，更要靠"密度"。 "过去，我们往往关心一个AI模型的'块头'有多大，即参数量有多大，其实我们更应该关心模型的'能力 密度'，即每个单位的参数能展现出的智能水平。"论文第一作者、清华大学计算机科学与技术系助理研 究员肖朝军说，"这就好比评价一个武林高手，不是看他块头有多大，而是看他一招一式里蕴含了多少 功力。" 研究人员表示，实现这样的"能力密度"，不能靠"模型压缩"。研究发现，强行压缩一个大模型，就像是 把一本厚厚的字典强行塞进小本子里，很多内容会丢失，导致"智力"下降，因此需要采用更先进的"数 据+算力+算法"体系，才能培养出"高密度"的精干小模型。 通过研究过去几年发布的51个开源大模型，研究人员 ...

21世纪经济报道记者 骆轶琪 不同于早期强调手机上搭载大模型的跑分数据，今年国内头部手机厂商争相将竞争焦点转向端侧智能的 实际应用，也即离线状态下手机能实现的丰富功能。 近期，多名受访者都对21世纪经济报道记者表示，DeepSeek为代表的开源大模型快速发展，让端侧模 型小型化进展迅速。这成为今年手机端侧能实现更为丰富能力的根本原因之一。 在此基础上，手机厂商在进一步探索长上下文、多模态的应用落地能力。 这也意味着，AI手机在质疑与期待中正悄然跨越从"炫技"到"刚需"的临界点。 虽然处在一种新技术生态的发展早期，但让AI手机逐渐具备Agent（智能体）能力是必然趋势，同时， 生态模式的重构也意味着将有一系列挑战。 作为AI大模型落地的关键一役，手机厂商对AI应用的落地和思考走到了哪里？ AI底层进化 同时端侧模型对手机内存的占用也显著缩小。通常，端侧大模型会率先部署在手机品牌的高端旗舰机型 中，原因之一就是高端手机的内存配置通常高于中低端手机，对大模型能力可以有更多负载支撑。 近两年来，AI大模型在技术层面的多维度突破，为手机端侧应用提供了基础条件，例如模型小型化、 多模态化和长上下文扩展等。 接受21世纪经 ...

文/新浪财经上海站 陈秀颖 11月12日，在上海证券交易所国际投资者大会"AI时代的中国产业新格局"圆桌论坛上，芯原股份董事长兼总裁戴伟民表示，随着生成式AI和空间计算的快速发展，AI定制化芯片（AI ASIC） 他指出，芯原股份在AI定制芯片领域已有较深技术积累，公司上市时被誉为"中国半导体IP第一股"，随着公司业务在业界获得更广泛的关注度和更深的认知度，目前公司已被业界誉为"AI AS 在谈及未来技术趋势时，戴伟民认为，AI的演进正从"云端计算"转向"端侧智能"。他举例称，对AI眼镜这类便携性智能终端设备而言，实时算力和能耗控制将成为关键突破方向。"真正的创新 他提到，公司将持续探索AI眼镜、AI玩具等新形态应用，而未来垂域模型在教育、医疗、金融等应用领域的"端上微调"也将成为主流，"这将是AI产业从云到端的重要跃迁。" 责任编辑：常福强 专题：专题|2025上海证券交易所国际投资者大会 ...

他提到，公司将持续探索AI眼镜、AI玩具等新形态应用，而未来垂域模型在教育、医疗、金融等应用 领域的"端上微调"也将成为主流，"这将是AI产业从云到端的重要跃迁。"新浪声明：此消息系转载自新 浪合作媒体，新浪网登载此文出于传递更多信息之目的，并不意味着赞同其观点或证实其描述。文章内 容仅供参考，不构成投资建议。投资者据此操作，风险自担。 专题：专题|2025上海证券交易所国际投资者大会 文/新浪财经上海站 陈秀颖 他指出，芯原股份在AI定制芯片领域已有较深技术积累，公司上市时被誉为"中国半导体IP第一股"，随 着公司业务在业界获得更广泛的关注度和更深的认知度，目前公司已被业界誉为"AI ASIC龙头企业"。 在谈及未来技术趋势时，戴伟民认为，AI的演进正从"云端计算"转向"端侧智能"。他举例称，对AI眼镜 这类便携性智能终端设备而言，实时算力和能耗控制将成为关键突破方向。"真正的创新不只是把模型 放在云上，而是让模型在端上运行，让设备更懂场景、更懂人。" 11月12日，在上海证券交易所国际投资者大会"AI时代的中国产业新格局"圆桌论坛上，芯原股份董事长 兼总裁戴伟民表示，随着生成式AI和空间计算的快速发展，A ...

他指出，芯原股份在AI定制芯片领域已有较深技术积累，公司上市时被誉为"中国半导体IP第一股"，随 着公司业务在业界获得更广泛的关注度和更深的认知度，目前公司已被业界誉为"AI ASIC龙头企业"。 在谈及未来技术趋势时，戴伟民认为，AI的演进正从"云端计算"转向"端侧智能"。他举例称，对AI眼镜 这类便携性智能终端设备而言，实时算力和能耗控制将成为关键突破方向。"真正的创新不只是把模型 放在云上，而是让模型在端上运行，让设备更懂场景、更懂人。" 他提到，公司将持续探索AI眼镜、AI玩具等新形态应用，而未来垂域模型在教育、医疗、金融等应用 领域的"端上微调"也将成为主流，"这将是AI产业从云到端的重要跃迁。" 专题：专题|2025上海证券交易所国际投资者大会 炒股就看金麒麟分析师研报，权威，专业，及时，全面，助您挖掘潜力主题机会！ 文/新浪财经上海站 陈秀颖 11月12日，在上海证券交易所国际投资者大会"AI时代的中国产业新格局"圆桌论坛上，芯原股份董事长 兼总裁戴伟民表示，随着生成式AI和空间计算的快速发展，AI定制化芯片（AI ASIC）需求正显著增 长，且AI的演进正从"云端计算"转向"端侧智能"。 责任 ...

Core Insights - The company reported a record revenue of 5.071 billion yuan for the first three quarters of 2025, representing a year-on-year growth of 9.29%, and a net profit attributable to shareholders of 698 million yuan, up 17.51% year-on-year [1] - The company is experiencing strong performance driven by increased penetration of edge intelligent technology and the successful launch of new products [1][2] - The company maintains a multi-product line strategy, with significant new product launches in 2025 that have quickly gained market acceptance, contributing to overall sales growth [2] Financial Performance - For the first three quarters of 2025, the company achieved a gross margin of 36.23%, 37.29%, and 37.74%, indicating a continuous improvement trend [1] - The company managed to reduce its sales expense ratio to 0.95%, down 0.35 percentage points year-on-year, and the management expense ratio to 2.33%, down 1.87 percentage points year-on-year [1] Product Strategy - The company's SoC chips are widely used across various sectors including home, automotive, education, and industrial applications, enhancing its competitive edge [2] - The strategic new products launched in 2025 have quickly surpassed the million-unit commercial threshold, significantly boosting the company's overall sales scale [2] Future Projections - Revenue forecasts for 2025-2027 are projected at 7.597 billion, 9.253 billion, and 10.610 billion yuan, with net profits expected to reach 1.066 billion, 1.442 billion, and 1.876 billion yuan respectively [3] - The company is expected to maintain a "buy" rating based on projected price-to-earnings ratios of 32.69, 24.16, and 18.57 for the years 2025, 2026, and 2027 [3]

Summary of Zhongke Chuangda Conference Call Company Overview - Zhongke Chuangda reported a revenue of 5.148 billion yuan in the first half of 2023, representing a year-on-year growth of 39.34% [2][4] - The net profit attributable to shareholders was 229 million yuan, with a year-on-year increase of 50.72%, indicating significant performance growth [2][4] Core Strategic Focus - The company's core strategy is the transformation towards an AI-driven Operating System (AIOS), collaborating with major model manufacturers like ByteDance and Mianbi Intelligence to build a global ecological platform [2][6] - Zhongke Chuangda has achieved edge deployment of 14B and 7B multimodal models in the automotive sector based on Qualcomm and NVIDIA platforms, launching products like Airbox to accelerate automotive intelligence [2][5] Business Segments IoT Business - The IoT business is experiencing rapid growth, driven by technological iterations and increasing user demand, particularly in domestic and international markets across various business scenarios [2][8] - The company expects continued high-speed growth in the IoT sector, focusing on enhancing technical capabilities and exploring new market opportunities [8] AIOS Development - Zhongke Chuangda has been dedicated to operating system development since its establishment in 2008, gradually shifting towards AIOS as AI technology evolves [5][6] - The company integrates AI natively into operating systems based on Linux and Android, transitioning from traditional control-based OS to decision-making OS [5] Automotive Sector - The automotive business, established in 2013, is moving towards a stable development phase, with AIOS positioned to leverage the complex applications in this sector [10][11] - The company believes that the transition from software-defined vehicles to AI-defined vehicles will present significant industry opportunities [11] Robotics Sector - Zhongke Chuangda has launched several robotic products aimed at industrial and logistics applications, integrating industrial IoT and visual solutions to enhance its product offerings [18] Research and Development - The company continues to increase R&D investment, particularly in AIOS platform development and multi-chip capabilities, while flexibly adjusting personnel to meet new business and technology demands [3][19] Market Strategy - Zhongke Chuangda's strategy combines platform-based development with vertical scene applications, ensuring robust growth across various sectors including smartphones, IoT, smart vehicles, and robotics [9][15] - The company emphasizes collaboration with major chip manufacturers like Qualcomm and NVIDIA to support cross-platform operating systems [17] Future Outlook - The company maintains an optimistic outlook for 2026, focusing on technology development and global strategic partnerships to capture emerging industry opportunities [7][16] - The IoT business is expected to play an increasingly important role in overall performance, driven by continuous technological advancements and user demand [8] Key Product Launches - The AI Box, launched in September at the International Motor Show in Germany, addresses the computational power shortage associated with large models in vehicles, providing a comprehensive AI computing solution [12] Challenges and Opportunities - The fragmentation of the IoT market presents challenges, but Zhongke Chuangda views it as an opportunity to leverage its universal platform to meet diverse customer needs [13][14]

Financial Performance - The company achieved a revenue of 1.848 billion CNY in the recent quarter, representing a growth of 42.87% year-on-year [4] - Net profit attributable to shareholders reached 70.568 million CNY, an increase of 48.26% compared to the previous year [4] - For the first three quarters, total revenue was 5.148 billion CNY, up 39.34% year-on-year, with net profit of 229 million CNY, reflecting a growth of 50.72% [4] Product and Technology Development - The company is focusing on AI integration across various sectors, including smartphones, smart vehicles, and smart hardware, with a goal to transition from traditional operating systems to AI-based operating systems (AIOS) [4] - AIOS is designed to revolutionize the OS landscape by integrating traditional interfaces with AI capabilities, enabling new forms of interaction such as generative UI and multi-modal experiences [4] - The company is actively developing AIOS for automotive applications, exemplified by the launch of the Drip OS platform in collaboration with Geely at the IAA Mobility 2025 [4][6] AIOS Strategy and Implementation - AIOS is structured into four levels: M1 (prototype), M2 (small-scale), M3 (expandable), and M4 (mass production), with a focus on effective AI strategy execution [5] - The Drip AIOS leverages Qualcomm's Snapdragon Ride platform, achieving real-time operation of 7B large models in vehicles, enhancing the AI cockpit experience [5][6] - The company is expanding its AIOS applications to include AI glasses and other IoT devices, aiming for a comprehensive AI ecosystem [5] Market Expansion and Collaboration - The company is enhancing its overseas automotive ecosystem, facilitating the entry of foreign brands into China and integrating with local middleware and software ecosystems [5][7] - Strategic partnerships with chip manufacturers like Qualcomm and NVIDIA are pivotal for the development of AIOS and related technologies [6][9] - The company is also collaborating with major AI model firms to optimize applications in smart automotive environments, enhancing user experience through advanced AI interactions [8][9] Future Outlook - The company anticipates continued growth in its smart software and AI sectors, driven by advancements in AI mobile technology and heterogeneous computing [8] - The focus will remain on building a robust AI ecosystem that supports smart homes and AI hardware, ultimately aiming for a fully interconnected AI environment [8][10] - The company is committed to maintaining a strong global presence with R&D teams across 16 countries, ensuring responsiveness to local market needs [10]