--- title: "Beinao-2: China's 1,024-channel invasive brain-computer interface moves from monkey to human" date: 2026-10-25 category: Brain–Computer Interface site: NeuroAI canonical: https://neuroai.site/a/na-brain-beinao2-invasive-bci language: en --- # Beinao-2: China's 1,024-channel invasive brain-computer interface moves from monkey to human > Beijing's self-developed high-channel invasive BCI has held a stable implant in a macaque for a year and is now edging toward human trials. A macaque with both arms bound reached for a strawberry it could only picture in its head. The fruit sat behind glass; the animal never touched it. Yet a robotic arm a meter away closed its grip on the berry — driven by nothing but the monkey's intent. Somewhere inside that monkey's skull sat a thread finer than a human hair, quietly translating electricity from single neurons into a command. This is Beinao-2 (北脑二号), a high-performance invasive brain-computer interface (脑机接口) that China built largely from its own parts. ## Key takeaways - Beinao-2 (北脑二号) is a high-performance invasive brain-computer interface (脑机接口) developed by the Beijing Institute for Brain Disorders (北京脑科学与类脑研究所) together with Beijing Xindazhi NeuroTech (北京芯智达神经技术有限公司). - Its centerpiece is a self-developed 1,024-channel flexible microwire electrode whose strands are roughly one-tenth to one-hundredth the width of a human hair. - In macaque trials the implant stayed stable for close to a year and let the animal steer a robotic arm to intercept moving 2D targets — described by the developers as the first time a macaque achieved that level of cursor control by thought. - A semi-invasive sibling, Beinao-1 (北脑一号), has already been placed in three human patients, with plans for about ten more implants and a roughly 50-patient formal clinical trial after regulators approve. - The long-term aim is to give people with spinal-cord injury, stroke, amyotrophic lateral sclerosis (ALS, 渐冻症) or locked-in states a way to move or communicate again. ## What the system is made of The headline number is 1,024 channels. In plain terms, that is how many points along the electrode can listen to the brain at once. More channels generally mean a finer read on which neurons fire and when, which is what you need to turn a flicker of intention into a smooth arm movement rather than a crude twitch. The electrode itself is the harder problem. Rigid implants irritate tissue and drift; Beinao-2 instead uses flexible microwire made from biocompatible material, with strands said to be one-tenth to one-hundredth the diameter of a hair. According to the team's disclosures, a version of this electrode sat stably in a macaque's brain for nearly a year while still recording the activity of 500–600 individual neurons at sub-millisecond resolution. ## Why a monkey grabbing fruit matters The demonstration that drew attention was not the implant alone but what it enabled. A macaque, with its arms restrained, was trained to "want" to catch a moving cursor on a screen. Once the electrode was in place, the system read the corresponding neural signal and drove a robotic arm to complete the catch. The Beijing team describes this as the first time a macaque achieved dexterous brain-controlled interception of a 2D moving target — a stepping stone toward useful, real-time control, not just a one-off trick. ## A parallel track already in people Beinao-2 is the invasive flagship, but it is not the only program out of the same Beijing pipeline. Beinao-1 (北脑一号) is a semi-invasive, wireless system that sits closer to the brain surface rather than deep inside it. Reporting by Reuters in early 2025 said three patients had already received Beinao-1 implants, with the team planning about ten more implants that year and a formal clinical trial of roughly 50 patients once regulators signed off. That matters because the fastest route to helping patients may run through the semi-invasive design: slightly lower signal quality, but less risk and fewer complications. Beinao-2 is the higher-ceiling bet — more channels, deeper read — and the one aimed squarely at the kind of performance that invasive leaders like Neuralink are chasing. ## How it compares, honestly Stating the comparison plainly: the Beijing team says its electrode performance and decoding precision are on par with, and in several metrics ahead of, publicly disclosed figures from international invasive BCI efforts. Independent, head-to-head benchmarking is scarce, and "ahead on a spec sheet" is not the same as "ahead in patients helped." Neuralink had implanted its device in three human patients as of the same Reuters reporting; Synchron, which uses a different, blood-vessel-based approach, had the longest human track record at the time with ten patients. According to Li An, chief scientist at BrainNet (脑机网), the realistic near-term value of high-channel invasive systems in China is less about racing foreign labs and more about building a domestic clinical supply chain — electrodes, chips, algorithms — that hospitals can actually deploy at scale. ## Honest limitations A few things this article does not establish. The 1,024-channel figure and the monkey results come from the developers and Chinese forum/official coverage; we did not find independent peer-reviewed publication of the human-scale performance numbers. The "50-patient trial" and the wireless Beinao-2 timeline are stated plans, not completed enrollments. Animal results rarely transfer cleanly to humans, and long-term stability in people — years, not months — remains unproven for this platform. We have also not verified pricing, manufacturing volume, or any overseas regulatory filing. ## What readers can do now - If you follow BCI as an investor or clinician, track the Beijing Xindazhi (芯智达) pipeline separately from Neuramatrix, NeuroXess and other Chinese invasive players — they are different companies with different electrodes. - Patients and families should treat "human trials starting soon" as exactly that: not an available therapy. Ask any domestic hospital offering BCI whether the procedure is in a registered clinical trial (临床试验) with ethics approval, not an informal implementation. - Watch for peer-reviewed papers or regulator (NMPA, 国家药监局) filings as the real signal that Beinao-2 has moved from impressive demo to approved medicine. --- Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-brain-beinao2-invasive-bci Free to quote with attribution and a link to the original.