10月21日，粤港澳大湾区国际脑机接口产业创新大会暨第六届国际脑科学前沿与产业大会在深圳脑解析与脑模拟重大科技基础设施举办。会议间隙，羊城晚报记者专访了中国科学院院士、中国神经科学学会理事长、广东省智能科学与技术研究院研究员张旭，围绕类脑智能的概念边界、侵入式芯片最新进展、类脑个人计算机上市时间表等展开交流。他透露，类脑个人计算机首批商品预计今年年底至明年1至2月份进入市场；广东省智能科学与技术研究院与珠海横琴脑虎半导体有限公司共建脑机接口与人机交互联合实验室，两款侵入式脑机接口核心芯片已流片，据其介绍可媲美国外最新产品。
On October 21, the Guangdong-Hong Kong-Macao Greater Bay Area International BCI Industry Innovation Conference and the 6th International Brain Science Frontier and Industry Conference were held at the Shenzhen Brain Analysis and Brain Simulation Major Science and Technology Infrastructure. On the sidelines, Yangcheng Evening News interviewed Zhang Xu, Academician of the Chinese Academy of Sciences, President of the Chinese Society for Neuroscience, and researcher at the Guangdong Institute of Intelligent Science and Technology, on the conceptual boundaries of brain-inspired intelligence, the latest progress in invasive chips, and the timetable for brain-inspired personal computers. He revealed that the first batch of brain-inspired personal computer products is expected to enter the market from the end of this year to January-February next year; the Guangdong Institute of Intelligent Science and Technology and Zhuhai Hengqin Naohu Semiconductor Co., Ltd. jointly built a BCI and human-computer interaction joint laboratory, with two invasive BCI core chips already taped out, which he says can rival the latest foreign products.
谈概念与芯片时，张旭说，从神经科学技术研究一直到超级计算，是一个整体的生态，有几个环节，承上启下、相互关联，不能割裂。从科学角度讲，类脑智能不仅包括类脑计算，也包括脑机接口、类脑组织工程等技术类别。它跟创新链密切相关，从基础理论开始，一脉相承，最后实现通用型类脑计算系统，也跟产业链高度吻合。相反，脑机接口如果没有信号采集、信号解码、反馈执行这三个层次，形不成闭环，就不能称之为典型的脑机接口。所以类脑智能与脑机接口不是并列关系，是内涵关系。
On concepts and chips, Zhang Xu said that from neuroscience technology research to supercomputing is a整体 ecosystem with several links that connect and relate to each other, not to be fragmented. From a scientific perspective, brain-inspired intelligence includes not only brain-inspired computing but also BCI, brain-inspired tissue engineering, and other technology categories. It is closely related to the innovation chain—from basic theory, in one continuous line, ultimately realizing general-purpose brain-inspired computing systems—and highly aligned with the industrial chain. Conversely, if BCI lacks the three levels of signal acquisition, signal decoding, and feedback execution, it cannot form a closed loop and cannot be called typical BCI. So brain-inspired intelligence and BCI are not parallel relationships but an inclusion relationship.
目前国内侵入式脑机接口核心芯片方面，广东省智能科学与技术研究院和珠海横琴脑虎半导体有限公司共建脑机接口与人机交互联合实验室，一起研制脑机接口核心芯片。已有两款商用量产化脑机芯片，还有一款生物仿生性闭环脑机芯片已流片，实现片上神经尖峰序列识别与闭环仿生电刺激，可媲美国外最新产品。张旭说，侵入式脑机接口芯片尽管现在还在临床研究阶段，但长远看对医学会有贡献，比如偏瘫患者、神经损伤患者等。侵入式和非侵入式脑机接口两者会紧密结合应用，这是更长远的目标。非侵入式可以说是普惠主流，广东省智能科学与技术研究院与燧人（珠海）医疗科技有限公司联合研发，于2025年7月推出的梦境调控枕头梦邻，定位为首款多模态梦境脑机接口设备，就是非侵入式脑机接口设备，已经进入大健康领域。
On domestic invasive BCI core chips, the Guangdong Institute of Intelligent Science and Technology and Zhuhai Hengqin Naohu Semiconductor jointly built a BCI and human-computer interaction joint laboratory to develop BCI core chips together. There are already two commercial mass-production BCI chips, and one biomimetic closed-loop BCI chip has been taped out, achieving on-chip neural spike sequence recognition and closed-loop biomimetic electrical stimulation, rivaling the latest foreign products. Zhang Xu said invasive BCI chips, though still in clinical research stage, will contribute to medicine long-term, such as for hemiplegic patients and nerve injury patients. Invasive and non-invasive BCI will be tightly combined in application—a longer-term goal. Non-invasive can be called the inclusive mainstream; the Guangdong Institute of Intelligent Science and Technology and Suiren (Zhuhai) Medical Technology jointly developed the dream regulation pillow "Menglin," launched in July 2025, positioned as the first multimodal dream BCI device—a non-invasive BCI device that has entered the big health field.
谈《行动计划》时，张旭说，广东省《脑机接口科技与产业协同发展行动计划（2026—2030年）》明确到2030年在医疗领域建成200个脑机接口病房，这个《行动计划》他也参与了其中一些进程，我觉得还是要坐实。侵入式脑机接口还处在临床试验、研究阶段，医院加入是好的，可以做强临床研究基地，加速技术进步。但也要充分认识到，脑机接口技术不是一夜之间在医疗应用上就能那么快落地，要通过国家相关部门和市场检验。普通老百姓设想的脑机接口，都是有创的、侵入式的脑机接口，这不是脑机接口的全部，不要误会了。侵入式与非侵入式脑机接口是相辅相成、共同发展的一个阶段。
On the Action Plan, Zhang Xu said Guangdong's "BCI Science and Technology and Industry Coordinated Development Action Plan (2026–2030)" specifies building 200 BCI wards in the medical field by 2030; he participated in some of its processes and believes it must be "made solid." Invasive BCI remains in clinical trial and research stages; hospitals joining is good, strengthening clinical research bases and accelerating technological progress. But it must be fully recognized that BCI technology will not land in medical applications overnight—it must pass national regulatory and market tests. Ordinary people imagine BCI as invasive, implantable BCI, but that is not all of BCI—do not misunderstand. Invasive and non-invasive BCI are complementary and developing together at this stage.
谈类脑个人计算机时，张旭说，技术成为商品有个过程。原始技术、原理性技术不可能马上兑现成受欢迎的商品，有很多跟市场相关的技术问题要解决。我们预计大概今年年底到明年1至2月份，头批商品应该能进入市场。价格方面，这个还不便于说，肯定比造个超算中心便宜多了，比租算力也便宜多了，用它可以不断训练你自己的专业模型。成为商品最关键的是好用，不一定专家才能用，这是我们在解决的问题。未来普通家庭、中小型实验室和科研机构、大小医院、中小企业都可以部署轻量化的类脑智能计算设备，自主开发适配自身需求的垂直大模型，真正把智能算力交到大众手中。
On brain-inspired personal computers, Zhang Xu said technology becoming a commodity takes a process. Original technology and principle-based technology cannot immediately convert into popular commodities; many market-related technical problems must be solved. We expect the first batch of products to enter the market around the end of this year to January-February next year. On price, it is not convenient to say yet, but it will certainly be much cheaper than building a supercomputing center and much cheaper than renting compute—you can use it to continuously train your own professional models. What matters most for becoming a commodity is being easy to use—not necessarily only experts can use it—and this is what we are solving. In the future, ordinary households, small and medium laboratories and research institutions, hospitals of all sizes, and small and medium enterprises can all deploy lightweight brain-inspired intelligent computing devices, independently develop vertical large models adapted to their own needs, truly putting intelligent compute into the public's hands.
谈扎根广东时，张旭说，来到珠海，看重的不仅是国家战略目标，还有推动澳门经济多元化、发展高科技研发与产业这一区域性国家目标，珠海和澳门是粤港澳大湾区国际科技创新中心建设的重要环节。最近国家有关方面还专门提到珠澳创新点，我非常高兴。没有白来嘛。从专业角度讲，还希望能借助澳门、香港对接国际市场，在人才、成果转化和全球供应链对接等方面优势明显。国内脑科学高端人才缺口相当大。神经科学里把类脑智能也纳入了，包括类脑计算、脑机接口等，但这是后起的研究领域、开发领域，对综合学科、交叉学科人才的需求特别大。一般学神经生物学的，不经过后期交叉学科学习，很难进入类脑智能、类脑计算、脑机接口这些方向。这是很大的问题，要多培养青年人。缺的是综合人才，纯粹搞芯片的人也不行，必须既懂神经科学也懂芯片，算法模型等也要懂一些。
On rooting in Guangdong, Zhang Xu said coming to Zhuhai values not only national strategic goals but also the regional national goal of promoting Macao's economic diversification and developing high-tech R&D and industry; Zhuhai and Macao are important links in building the Guangdong-Hong Kong-Macao Greater Bay Area international science and technology innovation center. National authorities recently specifically mentioned the Zhuhai-Macao innovation point—he is very pleased. "It wasn't a wasted trip." From a professional perspective, he hopes to use Macao and Hong Kong to connect with international markets, with obvious advantages in talent, achievement transformation, and global supply chain connection. The domestic high-end brain science talent gap is considerable. Neuroscience has incorporated brain-inspired intelligence, including brain-inspired computing and BCI, but these are latecomer research and development fields with particularly large demand for comprehensive and interdisciplinary talent. Those who study neurobiology generally struggle to enter brain-inspired intelligence, brain-inspired computing, and BCI directions without later interdisciplinary learning. This is a big problem—more young people must be cultivated. What is lacking is comprehensive talent; pure chip people alone won't do—they must understand both neuroscience and chips, plus some algorithm models.
