近日，具身智能大算力芯片企业辉羲智能宣布，随着Pre-B+及B轮融资相继完成，公司2026年内累计股权融资规模已超25亿元。本次融资由北京市人工智能产业投资基金领投，北京机器人产业发展投资基金、隐山资本、正心谷资本、金浦投资、建投投资、华泰宝利、慧目圆轮、上河动量基金、国家人工智能产业投资基金等政府基金及市场化投资机构共同参与；顺为资本、元生资本、凯辉基金、商汤国香资本、博华产投、国汽投资、朗玛峰创投、三七互娱、盛奕资本、星连资本、卓源亚洲资本等老股东持续加码。
Recently, embodied intelligence large-compute chip company Huixi Intelligent announced that with Pre-B+ and Series B rounds successively completed, its cumulative equity financing in 2026 has exceeded 2.5 billion yuan. This round was led by the Beijing AI Industry Investment Fund, with government funds and market-oriented institutions including Beijing Robotics Industry Development Investment Fund, Yinshan Capital, Zhengxin Gu Capital, Jinpu Investment, Jiantou Investment, Huatai Baoli, Huimu Yuanlun, Shanghe Momentum Fund, and National AI Industry Investment Fund participating; existing shareholders including Shunwei Capital, Yuansheng Capital, Cathay Capital, SenseTime Guoxiang Capital, Bohua Industrial Investment, Guoqi Investment, Longma Peak Ventures, 37 Interactive Entertainment, Shengyi Capital, Xinglian Capital, and Zhuoyuan Asia Capital continued to increase.
国家级与地方产业基金、产业资本及市场化投资机构的共同支持，体现了长期资本对具身智能核心算力技术自主创新及产业化前景的持续看好。本轮融资将重点用于新一代具身智能大算力芯片研发，并加快规模化交付及产业生态建设。公司将持续深化软硬件协同创新，提升具身智能计算平台的性能、效率与易用性，推动自主可控算力产品的规模化应用。随着具身智能加速走向真实应用场景，端侧大模型高效推理、低时延实时控制与低成本工程化部署，成为具身智能规模化应用亟待突破的关键挑战。
Joint support from national and local industrial funds, industrial capital, and market-oriented investment institutions reflects long-term capital's continued optimism about embodied intelligence core compute technology independent innovation and industrialization prospects. Funds will focus on next-generation embodied intelligence large-compute chip R&D and accelerating scaled delivery and ecosystem building. The company will continue deepening software-hardware collaborative innovation, improving the embodied intelligence computing platform's performance, efficiency, and usability, and promoting scaled application of independently controllable compute products. As embodied intelligence accelerates toward real application scenarios, edge large-model efficient inference, low-latency real-time control, and low-cost engineering deployment have become key challenges urgently needing breakthrough for scaled embodied intelligence application.
针对上述痛点，辉羲智能坚持数据闭环定义芯片的技术理念，推出基于通用可编程架构的高性能AI SoC光至R1及配套计算平台，并于2026世界机器人大会期间发布Huixi Embodied算力产品系列，持续完善具身智能算力产品布局。同时，公司构建了覆盖模型框架、编译器、算子库、Runtime、中间件、操作系统、仿真及实时控制的软件体系，通过软硬件协同优化，降低机器人企业从算法开发到实机部署的工程门槛，加速产品迭代与商业化落地。
Addressing these pain points, Huixi Intelligent adheres to the technical philosophy of "data loop defines the chip," launching the high-performance AI SoC Guangzhi R1 based on a general programmable architecture and supporting computing platform, and releasing the Huixi Embodied compute product series during the 2026 World Robot Conference, continuously improving embodied intelligence compute product布局. Meanwhile, the company built a software system covering model frameworks, compilers, operator libraries, Runtime, middleware, operating systems, simulation, and real-time control, reducing robot enterprises' engineering threshold from algorithm development to real-machine deployment through software-hardware collaborative optimization, accelerating product iteration and commercialization.
这个融资节奏和前面讨论的芯际无限、智元机器人入股芯片企业、蓝思科技参投49亿AI与具身智能基金，是同一个脉络。具身智能公司正在往芯片和算力方向布局，而芯片公司也在往具身智能方向靠。辉羲智能的股东名单里同时出现了国家级基金、地方产业基金、市场化机构和产业资本，这个组合说明，具身智能大算力芯片这个赛道，已经不只是创业公司和财务投资人的游戏，国家队和产业方都在往里押注。前面讨论过辉羲智能创始人徐宁仪的背景，清华电子系本硕博、微软亚洲研究院硬件计算组组长、百度昆仑芯片主任架构师，也讨论过联合创始人章健勇在蔚来的自动驾驶量产经验。这个团队组合加上25亿的年内融资，指向的是同一个判断，具身智能的算力底座，正在成为资本和技术同时押注的方向。
This funding rhythm follows the same thread as the previously discussed Xinji Wuxian, AgiBot investing in chip companies, and Lens Technology participating in a 4.9 billion AI and embodied intelligence fund. Embodied intelligence companies are布局 toward chips and compute, while chip companies are leaning toward embodied intelligence. Huixi Intelligent's shareholder list simultaneously includes national-level funds, local industrial funds, market-oriented institutions, and industrial capital—this combination indicates the embodied intelligence large-compute chip track is no longer just a game for startups and financial investors; national teams and industrial parties are all betting on it. Earlier we discussed Huixi Intelligent founder Xu Ningyi's background—Tsinghua Electronics bachelor's, master's, and doctorate, Microsoft Research Asia Hardware Computing Group head, Baidu Kunlun chip chief architect—and co-founder Zhang Jianyong's NIO autonomous driving mass production experience. This team combination plus 2.5 billion in-year financing points to the same judgment: embodied intelligence's compute foundation is becoming a direction where capital and technology bet simultaneously.
