Unitree's early investors bet on a brain-computer interface company

Anyone who has watched the movie *Avatar* must remember a classic plot: on the planet Pandora, Jack Sully, a former marine warrior paralyzed from the waist down, lies in a sealed pod and controls an artificially created hybrid Avatar with his mind through the device worn on his head.

This is the initial fantasy of brain-computer interface (BCI) for most people - allowing human intentions to cross physical boundaries and directly connect to another carrier. Now with the maturity of industrial chains such as AI and embodied intelligence, this vision has become possible to realize. A landmark financing was thus born:

Investment community learned that Wabo Technology has announced the completion of a new round of financing, co-invested by Vertex Ventures under Singapore's Temasek, Hongtai Fund, Chengdu Future Industry Fund under Chengdu Science and Technology Innovation Investment Group, and Huifengda Capital. Among them, a well-known bet made by Vertex Ventures was its early investment in Unitree Robotics.

Taking this opportunity, the investment community had an exchange with Zhang Xingzhi, founder of Wabo Technology. He is an entrepreneur with extremely geeky romantic temperament: he was struck by the scenes in *Avatar* when he was a teenager, and then embarked on a journey that lasted for more than ten years. It was not until last year that he sensed the timing was ripe, and officially founded Wabo Technology.

Speaking calmly, a sci-fi cyber world is faintly coming.

Led by a Tsinghua PhD, enabling AI to understand human intentions

It was the sci-fi scene in the movie *Avatar* that opened the door to a new world for Zhang Xingzhi.

How can humans map goals, choices and movement intentions to another body to extend their own capabilities in the physical world? The seed of exploration was thus planted - since his undergraduate years, he has been conducting research around "human-computer interaction".

But at that time, brain-computer interface was still only in the realm of imagination. So he chose to proceed from the first principle, first deeply engaged in key industrial links such as manufacturing, consumer electronics and supply chains in the industry, then went back to the academic world to find answers with the problems from the industry, and further pursued master's and doctoral degrees at Tsinghua University, bringing the doubts accumulated in the industry to the academic hinterland for solutions. It seems that he has been traveling back and forth between the industry and academia, but he has always been focused on the same main line - how to make machines truly understand and execute human intentions in a more natural and low-burden way. Every round trip brings him one step closer to the dream originated from *Avatar*.

It was not until 2025 that the best timing finally arrived.

The 15th Five-Year Plan lists brain-computer interface as a future industry for forward-looking layout. At the same time, the development of technologies such as flexible electrodes, low-power chips, edge-side AI and spatial computing terminals has made non-invasive brain-computer interface begin to have the possibility of industrialization.

Sensing that the timing was ripe, Zhang Xingzhi founded Wabo Technology at the end of last year. Founded almost at the same time, also located in Chengdu, adhering to the non-invasive route Wabo Technology and Gestalt Technology which takes the ultrasonic brain-computer interface route, were soon regarded by the outside world as the "Twin Stars of Chengdu BCI".

However, different from most BCI companies that focus on a single medical scenario or BCI hardware, Wabo Technology chooses to focus on more underlying matters. As Zhang Xingzhi believes, Wabo Technology is more positioned as a model company.

After ten years of development, artificial intelligence has been able to understand language, images, sounds and movements, but machines still lack a key data: what exactly a person wants to do before he actually takes action. For example, a camera can see a person pick up a cup, but does not know whether the person is going to drink water or put the cup back in place - a robot can imitate the movement trajectory, but it is difficult to understand when a person confirms, hesitates, makes corrections, or wants the machine to stop immediately.

This is the problem that Wabo Technology hopes to solve. At present, the team is building a "Human Neural Intention Model" - a set of neural signal representation and decoding system for real tasks: in the clear context of tasks, environments and time, integrating EEG, EMG, behavior, vision, device status and execution results, to judge the user's current goals, movement tendencies, confirmation, correction, stop or takeover intentions, and convert them into standardized "intention vectors" that machines can understand.

According to Zhang Xingzhi, Wabo Technology's neural intention model has three key characteristics: the first is task conditioning, which allows the model to know what the user is doing and what legal options are available; the second is cross-user, cross-time and cross-device generalization, reducing the time for re-training and lengthy calibration each time; the third is closed-loop capability. "A truly reliable brain-computer interaction should not only recognize correctly when the user clearly issues an instruction, but also remain silent when the user is distracted, blinking, speaking, moving, or has no control intention." Only when the machine learns both "when to execute" and "when not to execute" at the same time, can neural signals move from laboratory demonstrations to a broader market.

By then, the application boundary of brain-computer interface will far exceed today's imagination: it can not only empower medical rehabilitation, integrate into smart homes and embodied intelligence... even the fantasy of interstellar colonization and consciousness connection in *Avatar* will gradually approach reality from the continuous iteration of the model.

Investors come to the door, launching the "Chengdu + Shanghai" dual-headquarters layout

When he was determined to start a business, Zhang Xingzhi left almost no retreat for himself.

When judging that the industry window was opening, Zhang Xingzhi did not hesitate too much, but invested all his savings in it. In Zhang Xingzhi's words, he is a person with low material desires, and he just really wants to make this thing happen, so he chose to go all in.

Perhaps like-minded people always attract each other. From starting a business alone at the beginning, Zhang Xingzhi quickly attracted a group of like-minded partners - a core team with an average age of less than 30, combining geek temperament and full-stack capabilities, gradually took shape.

Of course, more senior scientific research forces are indispensable. At the level of underlying R&D, Wabo Technology has gathered BCI, neural engineering and neural intention research forces from universities such as Tianjin University and Fudan University, formed an interdisciplinary scientist team, and systematically promoted the underlying research, data system and closed-loop verification of the Human Neural Intention Model.

Soon, investors came to the door intensively.

With less than 5 months of actual operation, Wabo Technology completed the seed round financing invested by Ceyuan Capital. Almost at the same time as the completion of the seed round, this angel round of financing also started. Some investment institutions even directly brought all the post-investment and risk control teams to Wabo Technology's site in order to catch up with this round of financing window, and completed due diligence and decision-making within a week.

As one of the main investors of this round, Vertex Ventures, an early investor of Unitree Robotics which is about to be listed on the Sci-Tech Innovation Board, is very optimistic about Wabo Technology's amazing systematic capabilities in data collection, signal processing, equipment control, model training and engineering implementation, which makes this capital that has frequently laid out in the Physical AI field in recent years sense the next huge opportunity.

To a certain extent, this represents the common judgment of investors - what is truly scarce is not a certain hardware parameter, but a complete system that stably sends neural signals into real scenarios and continuously converts the results into model capabilities.

More importantly, brain-computer interface is only an entry point. The other end of the interface can be drones, robotic arms, humanoid robots... So we see that investors who used to focus on the tracks of embodied intelligence and commercial aerospace also voted for Wabo Technology - not betting on a certain BCI hardware product, but a cross-device reusable representation of human intention.

Especially in the space weightlessness environment, traditional methods relying on gestures, voice or even physical operation interfaces are facing restrictions, and brain-computer interface will become the most natural and important interaction method. In addition, for long-term stay in manned space stations, extreme environments such as space weightlessness, confined space and radiation can easily cause astronauts' neural fatigue and cognitive decline. Space BCI can realize real-time monitoring of astronauts' brain status and dynamic regulation of mental workload, and it will be the key core technology of the new generation of manned aerospace human-machine integration system.

Looking at the world, China and the United States are currently simultaneously deploying on-orbit BCI experiments, but commercial solutions that meet the strict aerospace standards (miniaturization, low power consumption, radiation resistance, high stability) and have engineering implementation capabilities are still scarce. It is worth mentioning that Wabo Technology has joined hands with Tianjin University to lead the industry in these more cutting-edge directions. The BCI team of Tianjin University is a pioneer in the field of space brain-computer interaction in China. It has independently developed equipment that took off with the Tiangong-2 and Shenzhou-11 missions, and completed the world's first space on-orbit BCI scientific experiment in human history.

Wabo Technology's cutting-edge exploration in the field of "space BCI" also made Hongtai Fund, which has frequently laid out in the commercial aerospace field in recent years, see another path with huge imagination space.

Although it is a rising star, Wabo Technology is evolving rapidly. "Whether it is the control object, or the smoothness and accuracy of control, it is updated and iterated on a weekly basis. Every time investors come, they get a brand new experience." This point has also impressed investors. It is reported that many investors have thrown olive branches to Wabo Technology, hoping to lock in the share of the next round of financing.

What is more worth looking forward to is that Wabo Technology will usher in another milestone - it has completed the landing docking with the government of Minhang District, Shanghai and "Brain Intelligence World", and will officially settle in as a BCI innovation center project incubated by Fudan University. It should be noted that "Brain Intelligence World" is the country's first BCI future industry cluster area built with key support from Shanghai, where leading domestic BCI enterprises gather. After the settlement, Wabo Technology plans to launch the dual-headquarters layout of Chengdu and Shanghai.

In terms of division of labor, the Shanghai headquarters focuses on high-end R&D, neural intention models, closed-loop neuromodulation and industry-university-research-medicine collaboration; the Chengdu headquarters focuses on hardware and product R&D, supply chain, engineering delivery and large-scale scenario application, thus realizing the simultaneous construction of two sets of capabilities of "cutting-edge research and industrial engineering".

Welcome the cyber world

Wabo Technology is just a microcosm: brain-computer interface is beginning to usher in an era of explosion.

According to incomplete statistics, in the first 5 months of 2026, there were more than 30 financing events in China's BCI field, with the total financing amount exceeding 4.5 billion yuan, far exceeding the scale of the whole year of 2025. "Almost all directions can get financing," sighed an investor who focuses on this track.

Standing on the front line of the industry, Zhang Xingzhi feels more obvious changes, "In the past, capital paid more attention to the medical device attributes or single-point scientific research achievements of BCI; now, investors from tracks such as embodied intelligence, artificial intelligence, consumer electronics and commercial aerospace are beginning to re-understand the cross value of BCI."

In Zhang Xingzhi's view, the future competition of BCI is not only around a certain hardware device, but about who can establish the ability to continuously generate data, train models, and enter more physical scenarios.

The answer given by Wabo Technology is to take neural signals such as EEG and EMG as the data entry, align neural signals, user behavior, scenario information and execution results in a clear task environment, form a continuously trainable neural intention model, and then convert human intentions into control signals that can be called by physical terminals such as robots, XR, smart homes and spaceships through WABO Interface.

In this system, Wabo Technology's products such as Lingxi and Dreamweaver are not the end points, but important entry points for obtaining real data. The terminals reach users, users generate feedback in real scenarios, data promotes continuous iteration of the model, and the improvement of model capabilities further improves product experience, finally forming a long-term closed loop of "neural data entry - intention model - physical control - data backflow".

This is also a point that investors attach great importance to. Compared with simply accumulating EEG data, Wabo Technology pays more attention to the task relationship behind the data: what neural signal the user generates in what scenario, how the model understands this intention, whether the machine executes successfully, and whether the user confirms, corrects or takes over. The truly valuable data is not an isolated waveform, but a complete link of "task - intention - action - result".

Talking about the future competition pattern of the industry, Zhang Xingzhi judges that it will not be a pattern where a single technical route wins all. Players in directions such as invasive and non-invasive BCI, consumer neural interaction, core devices and data models can find their own positions in the industry. But before that, a series of challenges such as scenario generalization, cross-user and cross-subject adaptation, safety boundary, data standards, cost, mass production and commercial demand still need to be overcome.

"In the next three years, the BCI industry will usher in a real differentiation." In Zhang Xingzhi's view, the companies that finally survive will definitely be a few companies that can truly run through the closed loop of data, models and scenarios.

And in the distance, the vision originated from *Avatar* is getting closer step by step.

This article is from the WeChat official account "Investment" (ID: pedaily2012), written by Wu Qiong, authorized to release by 36Kr.