近日，清华系RISC-V芯片企业芯际无限完成Pre-A轮融资，本轮融资由智元机器人联合安徽省汽车创新中心、国元股权、合肥包河创投等机构共同投资，万创投行担任融资财务顾问。其中，智元机器人是国内人形机器人第一梯队企业，此次战略入股意在为具身智能业务锁定自主可控的国产算力底座。安徽省汽车创新中心及地方国资的加入，则将助力企业对接区域汽车产业集群资源与场景订单。
Recently, Tsinghua-affiliated RISC-V chip company Xinji Wuxian completed a Pre-A funding round, jointly invested by AgiBot, Anhui Provincial Automotive Innovation Center, Guoyuan Equity, and Hefei Baohe Venture Capital, with Wanchuang Investment Bank as financial advisor. Among them, AgiBot is a first-tier domestic humanoid robot enterprise; this strategic investment aims to lock in an independently controllable domestic compute foundation for embodied intelligence business. The addition of Anhui Provincial Automotive Innovation Center and local state capital will help the company connect with regional automotive industry cluster resources and scenario orders.
芯际无限核心技术源自清华大学集成电路学院高性能处理器实验室，核心团队自2016年启动RISC-V芯片研发，是国内最早从事RISC-V芯片研发的团队之一。公司于2023年成立，2026年完成清华大学技术转移并更为现名。公司首席科学家何虎现任清华大学集成电路学院副教授、博士生导师、上海清华国际创新中心副主任，曾牵头负责省部级大算力智能芯片项目。目前，公司车规级芯片已实现规模量产，智能电机控制MCU出货量突破100万颗、2026年新签订单达520万颗，与江淮、奇瑞等主流车企达成深度合作。
Xinji Wuxian's core technology originates from Tsinghua University's School of Integrated Circuits High-Performance Processor Lab; the core team began RISC-V chip R&D in 2016, among China's earliest teams engaged in RISC-V chip R&D. The company was founded in 2023, completed Tsinghua technology transfer in 2026, and changed to its current name. Chief Scientist He Hu is currently an associate professor and doctoral supervisor at Tsinghua's School of Integrated Circuits and deputy director of Shanghai Tsinghua International Innovation Center, having led provincial-level large-compute intelligent chip projects. Currently, the company's automotive-grade chips have achieved scaled mass production, intelligent motor control MCU shipments exceeded 1 million units, with 2026 new orders reaching 5.2 million units, and deep cooperation with mainstream automakers including JAC and Chery.
从技术体系看，芯际无限定位车规加具身智能双赛道RISC-V算力底座提供商，构建了覆盖RISC-V CPU IP、AI加速NPU IP、SoC芯片设计、系统软件栈、Chiplet先进封装的全栈自研体系，向上输出标准化IP与完整SoC芯片，向下配套自研实时操作系统与开发工具链。车规侧，XC32系列完整覆盖车身、动力、底盘、集中式域控场景，主力产品XC32MC车身电机控制MCU已批量上车，达到ASIL-D最高功能安全等级的XC32LS双核锁步域控芯片已完成点亮，进入多家车企与Tier1测试阶段。具身智能侧，公司布局XC64AE系列64位大小脑芯片，采用国内先进工艺与Chiplet芯粒互联技术。车载MCU与机器人关节控制芯片底层需求高度同源，其车规级高可靠设计能力可在机器人场景形成降维优势。目前，公司机器人关节控制芯片已实现量产，面向中端算力的XC64AE芯片已启动与头部具身智能企业的量产验证，面向高端算力、采用Chiplet集成架构的V1000芯片进入流片阶段。
From a technical system perspective, Xinji Wuxian positions itself as a dual-track automotive-grade plus embodied intelligence RISC-V compute foundation provider, building a full-stack in-house system covering RISC-V CPU IP, AI acceleration NPU IP, SoC chip design, system software stack, and Chiplet advanced packaging, outputting standardized IP and complete SoC chips upward and supporting self-developed real-time operating systems and development toolchains downward. On the automotive side, the XC32 series fully covers body, powertrain, chassis, and centralized domain control scenarios; the main product XC32MC body motor control MCU has been mass-installed in vehicles; the XC32LS dual-core lockstep domain control chip, achieving ASIL-D highest functional safety level, has completed lighting up and entered testing at multiple automakers and Tier1s. On the embodied intelligence side, the company布局 XC64AE series 64-bit big-brain/small-brain chips using domestic advanced processes and Chiplet interconnect technology. Automotive MCUs and robot joint control chips share highly homologous underlying requirements; its automotive-grade high-reliability design capability can form a "dimensional reduction" advantage in robot scenarios. Currently, the company's robot joint control chips have achieved mass production; the mid-range compute XC64AE chip has started mass production validation with leading embodied intelligence enterprises; the high-end compute V1000 chip using Chiplet integration architecture has entered tape-out.
本轮融资将重点用于具身智能全栈RISC-V芯片的研发及产业化落地。从PC时代的x86、移动时代的ARM，到AI与智能终端时代，指令集架构正迎来新一轮变革，开源、轻量化、可定制的RISC-V成为全球产业共识的核心方向。谷歌已全面适配RISC-V系统生态，英伟达、Meta、高通等国际巨头通过兼容适配、技术收购、战略入股等方式加速布局。截至2023年底，RISC-V IP核累计出货量已达130亿颗。据SHD Group预测，2031年全球RISC-V SoC出货量将突破200亿颗，市场渗透率提升至25.7%，其中汽车领域占比达31%。国内赛道热度同步攀升，头部企业密集登陆资本市场，一级市场融资密集落地。当前多数玩家聚焦通用计算、消费级MCU等场景，车规级认证与具身智能适配的高端RISC-V芯片标的稀缺，芯际无限正是国内少数同时打通双场景技术与量产壁垒的企业。
This funding round will focus on R&D and industrialization of full-stack RISC-V chips for embodied intelligence. From x86 in the PC era and ARM in the mobile era to the AI and intelligent terminal era, instruction set architecture is undergoing a new round of transformation; open-source, lightweight, customizable RISC-V has become a core direction of global industry consensus. Google has fully adapted its RISC-V system ecosystem; international giants including NVIDIA, Meta, and Qualcomm are accelerating布局 through compatibility adaptation, technology acquisition, and strategic investment. As of end-2023, cumulative RISC-V IP core shipments reached 13 billion units. According to SHD Group forecasts, global RISC-V SoC shipments will exceed 20 billion units by 2031, with market penetration rising to 25.7%, of which automotive accounts for 31%. Domestic track heat is rising synchronously—leading enterprises are密集 listing on capital markets, with dense primary market financing. Currently most players focus on general computing and consumer MCU scenarios; high-end RISC-V chip targets with automotive-grade certification and embodied intelligence adaptation are scarce. Xinji Wuxian is one of the few domestic enterprises simultaneously breaking through dual-scenario technology and mass production barriers.
万创投行作为芯际无限融资财务顾问，基于对半导体与具身智能产业链的长期跟踪判断认为，人形机器人的算力需求远超传统消费电子，全身多关节实时控制、毫秒级响应与高安全可靠性，对底层芯片的定制化能力、稳定性与实时性提出极致要求。当前全球机器人芯片呈现特斯拉自研FSD、英伟达通用Jetson两大路线，芯际无限以RISC-V定制化走出第三条突围路径，向量扩展架构支撑多关节并行运算，模块化指令集可针对具身智能场景精准裁剪，叠加2.5D先进封装与Chiplet技术，突破单芯片算力与成本瓶颈。智元机器人的战略入股，将为芯际无限导入具身智能场景的终端需求、量产数据与落地渠道，加速车规级高可靠技术向机器人场景迁移。依托清华系技术壁垒、车规级量产验证与产业场景资源的多重壁垒，芯际无限有望在汽车、具身智能与AI三大高景气场景中占据关键生态位。
Wanchuang Investment Bank, as Xinji Wuxian's financial advisor, based on long-term tracking of the semiconductor and embodied intelligence supply chains, believes humanoid robot compute demand far exceeds traditional consumer electronics—full-body multi-joint real-time control, millisecond response, and high safety reliability impose extreme requirements on underlying chips' customization capability, stability, and real-time performance. Currently global robot chips present two major routes—Tesla's in-house FSD and NVIDIA's general Jetson; Xinji Wuxian takes a third breakthrough path with RISC-V customization—vector extension architecture supports multi-joint parallel computation, modular instruction sets can be precisely tailored for embodied intelligence scenarios, and 2.5D advanced packaging and Chiplet technology break through single-chip compute and cost bottlenecks. AgiBot's strategic investment will import embodied intelligence scenario terminal demand, mass production data, and deployment channels for Xinji Wuxian, accelerating migration of automotive-grade high-reliability technology to robot scenarios. Relying on multiple barriers including Tsinghua technical barriers, automotive-grade mass production validation, and industrial scenario resources, Xinji Wuxian is expected to occupy a key ecological niche in the three high-prosperity scenarios of automotive, embodied intelligence, and AI.
