Brain-computer interfaces are missing a "Lobster Pool"! Tsinghua University's Gao Xiaorong responds to a reporter from the 21st Century Business Herald: The large-scale brain electroencephalogram model foundation is most likely to be developed in China.

The Beijing Daily’s reporter | Zhang Rui    The Beijing Daily’s editor | Wei Guanhong

Since this year began, warm policy signals for brain-computer interfaces have been coming in frequently. For the first time, the Government Work Report includes “brain-computer interface.”

On March 13, the National Medical Products Administration approved the world’s first invasive brain-computer interface medical device—an “Implantable brain-computer interface hand motor function compensation system” from Borekon Medical Technology (Shanghai) Co., Ltd.—for listing.

During the “2026 Zhongguancun Forum Annual Conference” held from March 25 to 29, at the “Brain-Computer Interface Innovation Development Forum,” Deputy Minister of Industry and Information Technology Ke Qixin made it clear that brain-computer interfaces are currently at a critical stage of moving from technological R&D to large-scale application. It is necessary to further pool strengths and work together to help brain-computer interfaces accelerate from the laboratory to real-world applications.

During the forum, reporters from The Daily Economic News interviewed multiple experts and industry figures around issues such as brain-computer interface commercialization, technical routes, and future prospects.

Policy empowerment: Commercialization accelerates, entering the “brain renovation” stage

With “brain-computer interface” written into the Government Work Report for the first time, does this mean commercialization will speed up?

In response, multiple interviewees said, “That is definitely the case.”

Gao Xiaorong, a long-term professor at Tsinghua University and one of the main founders of the disciplines of neural engineering and brain-computer interfaces, has conducted brain-computer interface research in China since 1998. In an interview with reporters from The Daily Economic News, he said that this means “we are going to enter the ‘brain renovation’ stage now; the whole world is becoming an era where people need to ‘renovate their brains.’”

When discussing the listing of the world’s first invasive brain-computer interface medical device, he believes it is “very meaningful.” “Brain-computer interfaces have been around for 50 years since the concept was proposed, and finally there is a product coming to market.”

Jie Fu, CEO (Chief Executive Officer) of Shanghai Jinquansheng Biotechnology Co., Ltd., told reporters from The Daily Economic News that only market demand can truly drive the industry’s development. Once the medical side opens application channels, it will become the “low-lying area” that attracts all kinds of technology transformation and implementation; resources, capital, and technology will naturally converge in that direction. If all parties cannot see a clear path to monetization over the long term, all the R&D work done before will lack an effective outlet. “We can see that the government is already pushing this very actively and pragmatically.”

In May last year, Beijing Tiantan Hospital affiliated with Capital Medical University opened a brain-computer interface specialized outpatient clinic. Zhao Zizong, an academician of the Chinese Academy of Sciences and a professor at Beijing Tiantan Hospital affiliated with Capital Medical University, told reporters from The Daily Economic News that, “At present, the outpatient clinic is very hot, and Dr. Yang Yi—who goes out to see patients there—often can’t get off work once her schedule is up.”

Zhao Zizong explained that opening an outpatient clinic has two purposes: first, recruiting patients for research; second, preparing for future promotion, which requires building a case database. At present, the focus is mainly on three groups: people with hemiplegia, paraplegia, and amyotrophic lateral sclerosis.

“Unlike ordinary outpatient care, it’s necessary to assess many issues, including family circumstances, income, and the relationship between spouses, etc.” Zhao Zizong said. “In the past we didn’t pay attention to these—we thought the patient came, and that was enough. Actually the problems are much more complex. They are not simply disease-related issues. Long-term paralysis often leads to family problems, social problems like poverty caused by illness, and so on.”

He said that opening an outpatient clinic means brain-computer interfaces have entered the view of ordinary patients, but whether they can be done is another matter. It is still in the clinical trial stage, supported by research funding.

**Ecosystem to be built: Currently, there is no brain-electroencephalogram (EEG) large-model foundation platform

At present, AI (artificial intelligence) development is booming. Zhao Zizong believes that in the development process of brain-computer interfaces, AI technology is needed. Adding AI can help accelerate equipment upgrades and iteration and also assist with post-implantation training—for example, whether AI can be used to produce templates with stronger suitability, so that patients with different conditions can use them.

In Gao Xiaorong’s view, what is most lacking in the brain-computer interface field today is foundation-platform construction. An ecosystem similar to CUDA (a parallel computing platform and programming model developed by NVIDIA) has not been built yet. “It’s like building a ‘lobster pond’—once we build the ‘lobster pond,’ everyone can ‘raise lobsters.’”

Gao Xiaorong said that what we need to do now is precisely this foundation-platform construction. But currently nobody is willing to do such “dirty work and hard work,” and it requires handling massive amounts of data. “We started doing brain-computer interface competitions in 2010 and have accumulated a large amount of data. Now we are working on foundation models and basic computing power, and we will also cooperate with relevant institutions to invest resources in building infrastructure. Just like the development path of large models—someone needs to lay the groundwork first.”

Gao Xiaorong said that to build this “lobster pond,” there is a lot of work to do. “In simple terms, it requires data, algorithms, and computing power, and also application scenarios. Only after preparing all of these can we build this ‘lobster pond’—that is, the foundation platform for an EEG large model.” Gao Xiaorong said he believes that the foundation platform for an EEG large model will most likely be created in China, because our work is relatively ahead of schedule. “Just like language models have foundation models, EEG also needs a foundation model.”

Dispute over the route: Benefiting the many products** must be non-invasive****

Brain-computer interfaces are roughly divided into two categories. One is invasive, which requires surgical implantation of electrodes. The other is non-invasive, which collects signals through external devices such as head-worn setups.

In Zhao Zizong’s view, non-invasive approaches are the easiest to promote. Invasive signals are of better quality, but the technical requirements are high and the difficulty is greater. In addition, implanted devices may cause issues over the long term, such as immune reactions, fiber encapsulation, signal attenuation, and so on.

“Many companies in China are working on an external-type cap, but most of what they do is improving sleep, helping students focus, and so forth. It may be better to truly focus on motor-function rehabilitation, but the downside is that signal quality is not as good as invasive methods.” In Zhao Zizong’s view, “the simplest is the best.” Whether it is semi-invasive or fully invasive, it still requires craniotomy. Implantation can’t be 100% free of side effects, and that depends on the individual.

The global proportion of non-invasive versus invasive brain-computer interfaces is roughly 8:2. Is that because non-invasive is not as difficult?

In response, Fu Jie said, it’s not because the difficulty is lower. The core of brain-computer interfaces lies in the real-time “reading” and “writing” of signals. Currently, most companies focus on multi-modal data acquisition (reading) and neural modulation (writing), and these two are often separated. As industry heat rises, these directions are collectively called the brain-computer interface track. “At present, about 80% of companies are still at the stage of signal acquisition or one-way writing. But to achieve truly non-invasive brain-computer products with closed-loop control and personalized adjustment capabilities, I think the industry still needs to go through a relatively long development process.”

Fu Jie admitted that she is more optimistic about non-invasive approaches. Because chronic brain health problems are becoming a global “silent epidemic.” The real value of non-invasive brain-computer interfaces is not flashy technology, but responding to a social reality: children trapped in poor attention, delayed sleep, and anxiety; middle-aged people overwhelmed by pressure, insomnia, and comorbidities layered upon one another; and the elderly shifting from poor sleep to neurodegenerative diseases, with the social caregiving burden at a ratio of 1:2.5.

“From each person’s lifetime perspective, chronic brain diseases are unavoidable. Most of these chronic brain problems are not suitable to be solved with invasive methods. In terms of cost-effectiveness and risk, the risk-reward ratio is not a good match. So non-invasive brain-computer interface solutions targeted at chronic brain diseases will surely become that beam of light,” she said.

In Gao Xiaorong’s view, what is needed now is for products that can benefit the many to come out. It can’t be that only millionaires can “renovate their brains,” while others can’t. The world’s first implanted product currently on the market is not a mass-market, broadly accessible product. “Broad accessibility must be non-invasive—everyone can afford it. Invasive is more expensive and more complex than non-invasive.”

Promising outlook: In the “15th Five-Year Plan” period, there is hope to roll it out nationwide, but multiple challenges remain

When asked how far brain-computer interface technology will develop during the “15th Five-Year Plan” period, Zhao Zizong said that by the “15th Five-Year Plan” stage it may be possible to roll it out nationwide, but it must be conducted in hospitals with proper qualifications— not every institution can do it.

Zhao Zizong emphasized that brain-computer interface technology is still in the trial stage, not a replacement for traditional treatment methods, but rather provides an additional rehabilitation pathway. The promotion of the technology also needs to solve multiple issues, such as the shortage of professional training personnel and the development of standards.

He gave an example: after device implantation, professional staff are needed for training, and there is a shortage of talent in this area. At present, it is mostly computer professionals who help with decoding and training patients, and the training time is long. Patients don’t just stay in the hospital for a few days and then leave; instead, they need to first learn how to operate a computer, and understand what different signals represent as instructions.

“Currently, our plan is this: after surgery, the patient stays in the hospital for one month. After discharge, they stay near the hospital for two more months, and then only after that can they return home. Because it’s still in the research stage, the number of patients is limited. It is basically follow-up-based: if there is any problem, they can come back at any time for resolution. If we want to roll this out nationwide, who will do this work? That’s why we can only do it one by one, not because there are no devices— the equipment and instruments are all there, and implantation is also very simple— but because the training work after implantation can’t keep up,” he said.

In an explanation to reporters from The Daily Economic News, Zhao Zizong said that after extracting signals, it is necessary to analyze which signals correspond to which actions. Many signals have no value, or they are not signals from the controlling hand. To extract valid signals, it is currently computer professionals who do it. They guide patients to move the cursor, tell them when it goes too high, too low, left, or right, and then it needs adjustment. “The adjustment process is the training process—training them how to move.”

The reporter learned that because the patient mix at outpatient clinics includes a relatively high proportion of groups such as farmers and workers, this portion of patients’ rehabilitation training first needs to learn how to use a computer.

Does training necessarily need to be done through a computer? In the future, could it be replaced with a phone? Gao Xiaorong said that a computer and a phone should not be too different, and in the future it will definitely become a phone. “We are already moving our lab research toward a direction of glasses.”

Besides that, funding is also very important. Zhao Zizong mentioned that last year, the United States said the average cost per patient was 5,000 USD. “I felt very confident then,” which is about 30,000 to 40,000 yuan. Chinese patients can still accept it. But this year the United States mentioned a figure close to 50,000 USD, and that price remains a high barrier for ordinary patients.

Regarding when large-scale commercial applications might appear, Fu Jie believes that the key still needs to return to the brain-health problem—or the disease itself. Establishing a set of scientifically sound dialogue logic between clinical and approval authorities, and proving what advantages this technology has over existing treatment methods—for example, a certain therapy was effective for 50% of patients in the past, but it could be improved to 75% now, and in the future it may reach 90%. “I think the limiting step is approval itself. The market opening has already been cracked open, but that doesn’t mean you can directly fly to the commercialization endpoint. You still need to validate it step by step with solid clinical data.”

Cover image source: The Economic Daily media resources database