当外界还在争论脑机接口改变世界究竟还有多远，10月21日，清华大学教授洪波在粤港澳大湾区国际脑机接口产业创新大会暨第六届脑科学前沿与产业大会上预测，脑机接口的GPT时刻将在2030年到来，届时它不仅能帮助渐冻症、脊髓损伤患者重建新的通路，更可能彻底改变人类与人工智能交互的方式。这场汇聚院士、专家、医生、企业家与投资人的大会逐渐形成一个共识，中国脑机接口的机会，都系于全链条闭环，而广东正是全国少数能把研发、生产、生态完全自闭环的省份。
As the outside world still debates how far brain-computer interfaces are from changing the world, on October 21, Tsinghua University Professor Hong Bo predicted at the Greater Bay Area International BCI Industry Innovation Conference and the 6th Brain Science Frontier and Industry Conference that BCI's "GPT moment" will arrive in 2030, when it will not only help ALS and spinal cord injury patients rebuild new pathways but may completely change how humans interact with AI. The conference gathering academicians, experts, doctors, entrepreneurs, and investors gradually formed a consensus: China's BCI opportunity hinges on a "full-chain closed loop," and Guangdong is one of the few provinces nationwide that can fully self-close R&D, production, and ecosystem.
洪波在主旨报告中道出了这一技术的根本价值。他说，脑机接口的想法非常了不起，初心是帮助那些有运动残疾或者脑疾病的病人建立一个新的通路。他说，这条通路将造福十万渐冻症患者、百万脊髓损伤患者，甚至未来惠及千万上亿的老年痴呆患者。落地成果正在从实验室走向病房。洪波团队将传感电极置于大脑硬膜外，把想法变成手部控制命令，帮助脊髓损伤瘫痪病人重新喝水、吃饭。其NEO脑机接口今年3月获国家药监局批准为三类有源植入医疗器械上市，7月在华山医院开出全球首张处方，彭博社、路透社和《自然》杂志均以中国人跑在了前面予以报道。
Hong Bo stated the technology's fundamental value in his keynote. He said the BCI idea is remarkable, with the original intention of helping patients with motor disabilities or brain diseases establish a new pathway. This pathway will benefit 100,000 ALS patients, a million spinal cord injury patients, and potentially tens of millions to hundreds of millions of Alzheimer's patients in the future. Deployment results are moving from laboratory to ward. Hong Bo's team placed sensing electrodes outside the dura mater, turning "thoughts" into hand control commands, helping spinal cord injury paralysis patients drink and eat again. Its NEO BCI obtained NMPA Class III active implantable medical device approval in March this year and wrote the world's first prescription at Huashan Hospital in July; Bloomberg, Reuters, and Nature all reported that "China is running ahead."
上海脑虎科技创始人、CEO兼首席科学家陶虎指出，脑机接口作为深入大脑、定点治病的手段，其真正价值在于修复大脑。中国科学院深圳先进技术研究院脑认知与脑疾病研究所所长王立平表示，深圳先进院18年前便开始布局，围绕跨物种动物模型建立、脑认知的基本原理解析以及光遗传脑机接口等技术体系的研发，建立了多学科交叉的研究体系建设。历经十余年，建成脑解析与脑模拟重大科技基础设施，已服务超过370个团队，并依托脑设施建立全国首个多物种柔性脑机接口研发与验证平台。深圳先进院在建院之初即布局脑科学与脑机工程研究，在国际上率先提出了声遗传学，开启了超声脑机接口的研究，并成功研制出世界首台超声脑调控仪器。
Shanghai NeuroXess founder, CEO, and Chief Scientist Tao Hu pointed out that BCI, as a means of going deep into the brain and treating diseases at specific points, has its true value in repairing the brain. Wang Liping, director of the Shenzhen Institute of Advanced Technology's Brain Cognition and Brain Disease Institute, said SIAT began布局 18 years ago, building a multidisciplinary research system around cross-species animal models, basic principles of brain cognition, and optogenetic BCI technology. Over more than a decade, it built the Brain Analysis and Brain Simulation major science infrastructure, serving over 370 teams, and established the country's first multi-species flexible BCI R&D and validation platform. SIAT布局 brain science and brain-machine engineering from its founding, pioneered sonogenetics internationally, initiated ultrasound BCI research, and successfully developed the world's first ultrasound brain modulation instrument.
洪波认为，脑机接口从1.0迭代到2.0，重建再到未来的3.0，人机共生，已经展现了一个非常美好的前景。他预计脑机接口的GPT时刻在2030年，在未来这几年是关键时期。他强调，数据和脑机接口的解码模型是决定突破的关键。产业爆发的牵引力来自应用场景的拓展。洪波展示了一位北京的截瘫患者用脑信号驱动人形机器人，将一瓶水递给他人。他说，整个过程都是通过脑信号在驱动，我们已经看到了这个未来。他进而展望，今天每个人都有一部智能手机，未来会不会每个人都有一个自己的脑机接口。上海阶梯医疗创始人赵郑拓认为，脑机接口将成为人机交互的核心底层基础设施，借助大模型解读思想、让机器人成为自然人身体的延伸。
Hong Bo believes BCI iterating from 1.0 to 2.0, reconstruction, and then to future 3.0, human-machine symbiosis, has already shown a very beautiful prospect. He predicts BCI's GPT moment in 2030, with the coming years being critical. He emphasized data and BCI decoding models are key to breakthroughs. The driving force of industry explosion comes from scenario expansion. Hong Bo showed a Beijing paraplegic patient using brain signals to drive a humanoid robot to hand a bottle of water to another person. He said the entire process is driven by brain signals—"we have already seen this future." He further envisioned: today everyone has a smartphone; will everyone have their own BCI in the future. Shanghai Stairmed founder Zhao Zhengtuo believes BCI will become the core underlying infrastructure of human-machine interaction, using large models to interpret thoughts and making robots extensions of natural human bodies.
前景虽好，短板同样明显。洪波将中国脑机接口现状概括为四句话，技术上跟跑很及时，但是基础研究深耕不足，关键技术积累不够，创新链条还没有完全打通。他点出两大断点，首先是大学科研院所工程能力差，其次是企业临床转化能力差，导致产业呈现橄榄型，原始创新少、标志性产品少。中国工程院院士高天明提醒行业一定要沉下心来，不要敲锣打鼓，否则一阵子热就过去了，同时呼吁一切能成产品的尽量往市场上走。深创投周伊则从资本端提醒，脑机接口需要耐心、长期、稳定的资本注入，投资人应关注企业里程碑而非追逐热点。
The prospects are good, but shortcomings are equally obvious. Hong Bo summarized China's BCI status in four sentences: "technology follows quickly, but basic research lacks depth, key technology accumulation is insufficient, and the innovation chain is not fully connected." He pointed out two breakpoints—first, universities and research institutes have poor engineering capability; second, enterprises have poor clinical translation capability, causing the industry to be "olive-shaped"—few original innovations and few landmark products. Chinese Academy of Engineering academician Gao Tianming reminded the industry to "calm down, don't beat gongs and drums, otherwise the heat will pass." Shenzhen Capital Group's Zhou Yi reminded from the capital side that BCI needs patient, long-term, stable capital injection; investors should focus on corporate milestones rather than chasing hot spots.
他还点出制约产品性能的四大天花板，高通量柔性电极、低功耗专用芯片、医疗级封装、高质量神经数据集与解码算法，其中高质量标准化神经数据集尤为稀缺，没有这个东西，数据不够，大模型算不出来。面对机遇，深圳被寄予厚望。浙江强脑科技合伙人何熙昱锦表示，深圳是我们一直在投入并且将来继续准备加大投入的地区，这里的先进制造、产业链、供应链、人才与临床资源具有优势。建议大湾区借助本地优势打通一两个真实的应用场景，让技术到产品、产品到服务在真实世界中反复打磨，从单点场景突破做出样板案例，再向全国复制。陶虎看来，广东深圳有条件做出别处做不到的全链条布局，但难的是作为一个系统工程，它需要系统级产品级的突破，这些年我们都是单点突破，但是要串起来，这是广东的破局，也是中国破局的地方。
He also pointed out four ceilings constraining product performance—high-throughput flexible electrodes, low-power dedicated chips, medical-grade packaging, and high-quality neural datasets and decoding algorithms, with high-quality standardized neural datasets particularly scarce: "Without this, data is insufficient, and large models cannot compute." Facing the opportunity, Shenzhen is highly expected. BrainCo partner He Xiyujin said Shenzhen is a region "we have been investing in and will continue to increase investment in," with advantages in advanced manufacturing, supply chain, talent, and clinical resources. She suggested the Greater Bay Area use local advantages to "open up one or two real application scenarios," letting technology-to-product and product-to-service be repeatedly refined in the real world, making benchmark cases from single-point breakthroughs, then replicating nationwide. In Tao Hu's view, Guangdong and Shenzhen have conditions to achieve full-chain布局 that others cannot, but the difficulty is that as a systems engineering project, it needs system-level and product-level breakthroughs—"these years we have all been making single-point breakthroughs, but stringing them together is Guangdong's breakthrough and China's breakthrough."
