Dialogue with Zhang Xingzhi of WABO Technology: The Post-90s Entrepreneur Backed by Unitree Investors Aims to Build the Foundational Large Model in the Brain-Computer Interface Field

The ultimate romance of science fiction lies in prefiguring the trajectory of human technological evolution.

Sixteen years ago, the scene in Avatar where humans controlled alien bodies through neural link devices deeply impressed Zhang Xingzhi, an undergraduate computer science student at Tianjin University. The underlying logic of bypassing all intermediaries and allowing thoughts to directly interface with reality planted a seed in his mind.

At that time, brain-computer interfaces were still a science fiction concept, and neuroscience was far from mature. To nurture this nascent idea, Zhang Xingzhi adopted a pragmatic, long-term strategy: first immersing himself in the industry to build capabilities, and waiting for the right moment to seize the opportunity.

From his undergraduate research on Kinect-based human-computer interaction, to his tenure as an intelligent manufacturing architect at SAP and Tencent, and later to his role as Vice President and Dean of the Research Institute at a company listed on China’s A-share market, he has spent over a decade thoroughly mastering the entire value chain of smart hardware—from R&D and production to supply and sales—as well as the iterative logic of AI algorithms and cutting-edge human-computer interaction technologies. Throughout this period, he has remained firmly convinced that brain-computer interfaces are poised to reach an inflection point for industrial-scale commercialization.

In 2025, the inflection point arrived. On the policy front, brain-computer interfaces (BCIs) were listed among the six major future industries. Technologically, dry electrodes, large language models, and domestically produced computing power matured in tandem, placing China’s non-invasive BCI sector on par with that of the United States. Zhang Xingzhi resigned from his executive position to found WABO Technology full-time, dedicated to transforming the science-fiction vision of Pandora into technological achievements that benefit every ordinary person.

Since its establishment, WABO Technology has advanced at a pace far exceeding expectations: it completed three rounds of financing within three months, with investors including Vertex Ventures, HONGTAI APLUS, the Future Industry Fund under CDIG, and Orinno Capital.

More importantly, WABO Technology has avoided crowding into the medical sector, instead anchoring itself to a differentiated path centered on the "Human Neural Intent Model." Adhering to an integrated hardware-software strategy, it leverages self-developed hardware to collect neural signal data from the real physical world, thereby building a generalizable EEG foundation model that is cross-subject and cross-task.

Recently, VCBeat interviewed Zhang Xingzhi, Founder of WABO Technology，Deconstructing the Startup’s Underlying Logic and the Post-90s Founder’s Unconventional Take on the Brain-Computer Interface Race.

Zhang Xingzhi, Founder of WABO Technology

Before embarking on his entrepreneurial journey, Zhang Xingzhi’s career trajectory was already regarded as the epitome of success by peers. His research experiences at Tianjin University and Tsinghua University laid a solid technical foundation in human-computer interaction. After graduation, he advanced through roles at multinational corporations, major internet companies, and publicly listed firms, ultimately becoming Vice President and Dean of the Research Institute of an A-share listed company at the age of 33.

But the idea in his mind of creating a “domestically produced Avatar” never faded.

What triggered the final decision was Zhang Xingzhi’s judgment on the industry’s inflection point: “Over the past two to three years, following the explosive growth of large language models, people have witnessed AI’s transformative impact across various industries. We believe “Brain-computer interfaces will also have their own GPT moment.”

In Zhang Xingzhi’s logic, the qualitative leap of large language models relies on massive amounts of internet text data, whereas the qualitative leap in brain-computer interfaces (BCIs) depends on vast quantities of real-world neural signal data. When data accumulation reaches a critical threshold, emergent capabilities arise in the models, transitioning them from laboratory-level demonstrations to truly practical, deployable applications.

By the end of 2025, Zhang Xingzhi embarked on full-time entrepreneurship, assembling a core team in Chengdu. Diverging from the traditional organizational model of hard-tech enterprises, which is heavily reliant on manpower and high costs, WABO Technology adopted an ultra-lean, AI-native startup model. The core team focused on technological breakthroughs, product iteration, and strategic decision-making, achieving rapid launch and validation through exceptional operational efficiency.

Exceptional efficiency and a clear technological roadmap have earned WABO Technology sustained recognition from top-tier capital. The pace of securing three rounds of financing within three months is extremely rare in the hard-tech sector. Vertex Ventures, the lead investor and an early backer of Unitree Robotics, is an institution renowned for its acumen in hard tech; it identified the team’s full-stack technical capabilities and differentiated strategic path at an early stage.

More compelling is the case of an investor’s “reverse entry.” After engaging with the company’s project, the seed-round investor of WABO Tech not only invested in the startup but also chose to join WABO Tech full-time upon completion of the investment, serving as the company’s head of financing.

Zhang Xingzhi has always believed that “in the hard-tech sector, ultimate technological and product prowess is the only passport. All connections and background can only help you qualify for a seat at the table. Whether you can ultimately succeed depends not on storytelling, but on tangible execution and iterative capabilities.”

The brain-computer interface (BCI) industry is broadly divided into two technological pathways: one follows a hardware-intensive approach, represented by invasive and semi-invasive methods, which acquire high-precision neural signals via implanted electrodes. This pathway is primarily used for treating severe neurological disorders, characterized by high unit costs, complex operations, and the need for surgical implantation. The other adopts a hardware-light approach, predominantly non-invasive, which lowers the barrier to use and expands the user base, albeit at the expense of signal quality.

In the past, the industry was trapped in a binary dilemma: “invasive methods are difficult to popularize, while non-invasive ones are impractical.”

WABO Technology has carved out a third path: entering the market with a non-invasive technical approach, achieving extreme lightweight hardware while developing highly sophisticated deep-learning models, and systematically compensating for signal limitations through its foundational EEG large model.

In Zhang Xingzhi’s view, this is the inevitable choice for achieving human-machine integration—non-invasive approaches are the only ones capable of scaling to the required data volume.

Over the eight months since commencing formal operations, WABO Technology has successfully validated the Scaling Law in the field of brain-computer interfaces (BCI) through its neural intent model, which was trained on a massive volume of self-collected neural signal data. This signifies that the key to future progress lies in expanding the scale of high-quality data, thereby preparing for the "GPT moment" of large EEG models.

Based on this assessment, WABO Technology regards data as a core critical asset and is determined to solidify its data acquisition capabilities.

How to Solidify Data Acquisition? WABO Technology Decides to Start with Data Acquisition Hardware, as Hardware Serves as the Data Entry Point for the Entire System. It has established a complete engineering platform spanning from flexible electrodes and domestically produced AFE acquisition modules to data acquisition terminals: the core acquisition chip was jointly developed with local enterprises, achieving full domestic substitution of EEG analog front-ends; its self-developed flexible dry electrodes completely eliminate the need for conductive gel, requiring no complex preparation for wear and significantly improving signal acquisition efficiency.

In this process, WABO Technology places particular emphasis on collecting data in real-world settings. “We believe that if a model is trained exclusively on data collected in laboratory environments, it will likely only be applicable within such controlled settings,” Zhang Xingzhi candidly stated. He acknowledged that although self-collecting data is more labor-intensive in the early stages, it is well worth the effort. For datasets of comparable size, self-collected data holds far greater significance for our model training than externally sourced data.

Furthermore, unlike other model companies that emphasize data duration, WABO Technology places greater emphasis on the number of people behind the data. Zhang Xingzhi told VCBeat, "The cross-subject problem is the first hurdle preventing brain-computer interfaces from achieving large-scale application. To solve this problem, it is necessary to expand the population base of subjects."

The benefits of this approach are also reflected in the low calibration requirements for different subjects—eliminating the need for repeated user-specific training and calibration, thereby delivering a near plug-and-play experience and further enhancing WABO’s efficiency in data acquisition.

To this end, WABO Technology even established its own “Data Acquisition Center,” leveraging business ingenuity to attract more participants to the ambitious blueprint of building a “Chinese-made Avatar.”

Zhang Xingzhi emphasized: “Hardware is merely our tentacle for data acquisition; the neural intent model is the true moat.” In his strategic vision, full-stack capabilities in both hardware and software serve as the basic entry requirement; however, the ultimate competition in this sector will inevitably center on model capabilities and data scale.

The Brain-Computer Interface (BCI) sector has never lacked grand narratives, with a constant stream of sci-fi concepts ranging from consciousness uploading to memory storage.

However, as a frontline entrepreneur, Zhang Xingzhi is remarkably pragmatic.

In his view, while phased breakthroughs will be achieved across various niche scenarios over the next three to five years, a truly epoch-making qualitative leap worthy of historical record will require cumulative efforts from the entire industry. “Current brain-computer interface models are clearly still in the pre-GPT-2 era.” Zhang Xingzhi is well aware of the industry’s current position. Everyone is awaiting the inflection point where quantitative changes lead to qualitative transformation, and the core variable driving this inflection point is data.

This is also why WABO Technology would rather slow down its commercialization pace to prioritize data accumulation—when the model’s capabilities are sufficiently robust, commercial success will follow naturally; if the model is not user-friendly enough, even the most flashy scenarios merely represent pseudo-demand.

Regarding the ultimate trajectory of the industry, Zhang Xingzhi has a clear judgment: brain science, artificial intelligence, and embodied intelligence are likely to converge at the same endpoint. As embodied intelligence advances, it becomes increasingly evident that enabling robots to truly understand humans and achieve higher levels of intelligence ultimately requires learning from the human brain. The human brain is the most perfect intelligent entity known in the universe. Only by understanding its operational logic and decoding human neural intentions can we create intelligent machines that genuinely comprehend humans.

“Unintentionally, we are replicating humans step by step; this is the most fundamental path of technological evolution.” In Zhang Xingzhi’s vision, the ultimate form is not about humans adapting to machines, but rather machines proactively understanding human intentions. People will not need to learn complex operational commands, speak, or perform manual actions; they will only need to generate a thought, and the device will execute it accurately. At that point, brain-computer interfaces will no longer be standalone products, but the underlying interaction gateway for all intelligent devices.

At the end of the interview, when asked which three words he hoped the public would use to remember WABO Technology, Zhang Xingzhi’s answer was somewhat unexpected: vision, dream, and taste.

Vision and dreams are easy to understand, “Taste” is a term that Zhang Xingzhi repeatedly emphasizes—academic taste determines one’s judgment in selecting research directions, while product taste determines the degree of excellence in execution. He strongly concurs with Yang Zhenning’s views on academic taste: choosing the right direction yields twice the result with half the effort, while choosing the wrong one means that no matter how hard you try, you will only stray further down the wrong path.

From sci-fi scenes on the big screen to startups collaborating between Chengdu and Shanghai, Zhang Xingzhi has transformed a teenager’s obsession with science fiction into tangible products over the course of more than a decade. He often says that his personal dictionary contains no word for “failure,” only “success” and “growth”; as long as he remains at the table, he has not lost.

This long-termist mindset is also reflected in WABO Technology’s assessment of the brain-computer interface sector.

The Brain-Computer Interface (BCI) race is destined to be a long marathon. While some monetize through medical qualifications and others secure financing by hyping concepts, WABO Technology has chosen the most challenging yet imaginative path: solidifying data acquisition, strengthening model development, and ultimately breaking down the barriers between human intent and the physical world.

At the end of this road lies the “Chinese Avatar” that the young man, once seated in an IMAX theater, had long yearned for.