NeuroXess Completes 700 Million Yuan Series B Financing; Tri-All Brain-Computer Interface System Enters Registrational Clinical Trials

By Hu Xiangyun

Editor | Hai Ruojing

36Kr has learned that NeuroXess has completed a 700 million RMB Series B round of financing. This round was jointly led by SDIC Jili, Shenzhen Capital Group, and Qinzhi Capital, with participation from SMIC Capital, Huangpu Investment, Cornerstone Capital, IDG Capital, Xichen Capital, Junxi Capital, Huafu Securities, Zhongwei Capital, Jiangxi Financial Holdings, and other institutions, while existing shareholders such as Shanda and Zhongping Capital increased their investment. The raised funds will be used for multicenter clinical trials and registration approval of the "fully implanted, fully wireless, fully functional" ("Three Alls") brain-computer interface system, mass-production construction of the super factory, as well as R&D of next-generation products and exploration of BCI+AI integration.

In terms of institutional attributes, this round of financing was overall led by state-owned capital, while on the industrial capital side it brought together a "cross-sector" lineup of leading institutions from tracks such as semiconductors, medical devices, and materials.

Tao Hu, founder and chief scientist of NeuroXess, told 36Kr that this was done, first, because brain-computer interfaces have risen to the level of national strategy, and state-owned capital leading the investment can bring stronger policy support; and second, out of consideration for subsequent industrial collaboration. In his view, the market currently still lacks electrode companies, chip companies, or algorithm companies born specifically for brain-computer interfaces. Therefore, only by joining hands with upstream and downstream partners and first "building up the entire brain-computer interface industry" can the industry ceiling be raised higher.

In 2026, the situation of leading brain-computer interface companies being "rushed for allocation" is no longer an isolated case. In Tao Hu's perception, three or four years ago, brain-computer interfaces were still a "relatively immature sci-fi concept" in investors' eyes, and people liked to hear stories about "the first in China," "the highest channel count," or "China's version of Neuralink"; but now, brain-computer interfaces have become an industry that is seriously regarded, and the market is more concerned about clinical effects, patient benefits, independent and controllable supply chains, and the final delivered products.

In terms of technology and products, NeuroXess has self-developed a fully implanted, fully wireless, fully functional "Three Alls" brain-computer interface system. As the name suggests, fully implanted means that all core components, including the battery, are implanted in the body, with no exposed parts on the body surface; fully wireless means that both power supply and data transmission are completed wirelessly, so that even when patients shower, sleep, or socialize, they are unaffected; fully functional means that the same system can perform motor decoding as well as speech decoding and environmental control.

Tao Hu believes that the "Three Alls" solution can better avoid the infection risk problems of traditional implantation surgery and is expected to continuously and stably collect real data. In traditional solutions, wires pass through the skin to connect to external devices, serving as a channel for bacteria to enter the skull, with a relatively high infection risk. And with wired power supply, patients are in a tense state, so the collected EEG may not reflect normal life. Full implantation plus full wireless is equivalent to blocking these two entry points.

In this sense, the brain-computer interface is no longer an experimental system that requires external equipment to maintain, but has become an "implantable system capable of working independently."

To achieve this, NeuroXess, when designing the implantation solution, chose to place flexible electrodes on the surface of the cerebral cortex under the dura mater, and to place high-heat-generating units such as the battery, processor, and antenna under the skin of the chest. Tao Hu explained that inserting into the cerebral cortex may seem more "precise," but as brain tissue shifts and intracranial pressure changes, it can easily cause electrode displacement or detachment. In addition, once an immune response is triggered and nerves are wrapped in inflammation, signals will attenuate, and in severe cases a second craniotomy may be required. Placed on the cortical surface, signal quality is basically the same as deep insertion, while long-term reliability is actually better. If the number of channels is to be increased in the future, it is only necessary to increase electrode density.

Reflected in patients, this is what NeuroXess has always pursued: "not harming the brain, not burning the brain, and reproducible," that is, electrodes do not puncture the brain parenchyma, heat sources stay away from the brain, and the surgery borrows the mature DBS paradigm, so neurosurgeons can operate independently after training. "We hope this solution can bring back patients' dignity."

According to data provided by NeuroXess, as of now, this solution has completed 65 clinical trials; in July this year, the "Three Alls" brain-computer interface system officially entered the Class III medical device registration clinical stage, with Fudan University Affiliated Huashan Hospital leading the GCP registration clinical trial, Professor Mao Ying's team serving as the principal investigators, and nearly 20 top-tier hospitals across the country jointly carrying it out.

At the end of September, NeuroXess disclosed its latest exploratory case, demonstrating the capabilities of this brain-computer interface product more concretely. For example, in a real-life scenario, a subject, with the help of the simultaneous localization and mapping (SLAM) capability of AI smart glasses, completed visual perception and dynamic modeling, identified the area he was focused on, and listed controllable devices. Then, he used the brain-computer interface to confirm his intention and was able to independently adjust the hospital bed, turn lights on and off, and control other home devices.

In the entire chain, the brain-computer interface is the underlying connection, and XessOS (NeuroXess's self-developed brain-computer operating system) is responsible for translating intentions into device commands.

Tao Hu calls this direction "BCI+AI." In his view, the brain-computer interface is "not just a medical device, but an interface for the fusion of human intelligence and artificial intelligence," in which humans should be responsible for setting goals and making judgments, while AI and machines undertake execution. On the one hand, the high-quality neural data of brain-computer interfaces requires AI large models for precise decoding; on the other hand, embodied intelligence and large models lack precisely the entry point of "intention."

But Tao Hu is well aware that the current data volume is still far from sufficient. "With a reserve of dozens of clinical cases, it is still impossible to make algorithms seamlessly transfer between different patients. It is possible that the model I tuned on this patient will not work when switched to the next person." His judgment is that at least several hundred or even several thousand cases are needed to form an algorithm that can be generalized, reproduced, and transplanted.

In Tao Hu's vision, the development of brain-computer interfaces will go through two stages. The first stage is now, using the invasive "Three Alls" system to serve severe patients with high-level paraplegia, aphasia, blindness, and other conditions, allowing them to "return from patients to normal people," able to communicate independently, control devices, and live on their own.

The second stage will be in the next 3-5 years, using invasive brain-computer interfaces as the gold standard to calibrate non-invasive solutions with weak and unstable signals, so that brain-computer interfaces can move from the hospital bed into ordinary people's daily lives and let "normal people become superhuman." "Non-invasive is the future of the industry, but non-invasive signals are too poor, and high-quality invasive data must be used for calibration," Tao Hu believes.