Unveiling Li Zexiang: An Engineering Professor and His Sweet Spot

By Lili Yu

Edited by Jing Liu

One day in the late 1990s, at the Hong Kong University of Science and Technology on the Clear Water Bay Peninsula, Liu Yingli — then Deputy Secretary-General of the Shenzhen municipal government — knocked on door after door in the university's laboratories. Shenzhen was desperate to transform its industrial base, but hamstrung by its lack of top-tier universities and research institutions. So it devised a workaround: court professors from elite schools, lure them to Shenzhen to collaborate with local companies, and hope a few startups might emerge.

When he reached Lab 3126, the door opened to a professor who, unusually, didn't speak Cantonese — he had a thick Hunan-accented Mandarin. Though born on the mainland, he had made his way to Hong Kong via Carnegie Mellon and UC Berkeley, the two premier American institutions for robotics. That pedigree meant access to a vast industrial network in the robotics field.

His name would later become widely known: Li Zexiang, professor of electronic and computer engineering at HKUST and founder of its Automation Technology Center.

That knock on the door marked the beginning of the story for Googol Technology — China's earliest motion control company. Soon after, in 1999, the Shenzhen municipal government, HKUST, and Peking University jointly established the Shenzhen-Hong Kong Institute of Industry, Education and Research, with Googol becoming HKUST's first tenant. Twenty-four years later, on August 15, 2023, Googol listed on the ChiNext board.

For Professor Li, founding Googol represented his pivotal leap from academia to industry. For decades, Chinese universities had been severely isolated from the industrial world. Li and his partners — HKUST professor and former dean of engineering P.C. Pao, and his longtime HKUST colleague Jie Gan, then associate dean at CKGSB — broke through that wall together.

Yet in interviews, Li repeatedly emphasized to An Yong Waves that many people mischaracterize him as a representative of "scientist entrepreneurship." Though Googol was the first IPO in his ecosystem, he more often played the role of a "rebel" against that very path.

Scientist entrepreneurship is fraught with obstacles, and Li had long since pivoted to another route: supporting student founders.

In fact, on August 15, the day of Googol's listing, his WeChat Moments remained completely quiet — his reposts had nothing to do with it. The IPO might as well have been someone else's affair.

Indeed, Li's greater renown stems from a long list of companies: not just the classroom-incubated DJI, now worth hundreds of billions, but also QKM Technology and ePropulsion from Lab 3126, plus Narwal and HAI Robotics from the XbotPark robotics base in Songshan Lake. Of the roughly one hundred master's students, PhDs, and postdocs who passed through Lab 3126, some fifty went on to build respected companies. Data from 2021 showed that XbotPark had incubated over sixty companies, with 15% reaching unicorn or near-unicorn status, and an overall survival rate of 80%.

The shift from hands-on founder to student enabler reflected Li's insight into how industry really works.

"So-called technology commercialization is actually a false proposition — human transformation is what matters," Li told An Yong Waves in an interview months earlier. History had already proven that "using accumulated technology to find applications is extremely inefficient; the more efficient approach is letting problems drive technology." Israel's example had also shown that technology transfer ultimately depends on "the flow of people."

So he set out to build a larger system that would let people move freely. Over the past decade, beyond the XbotPark Songshan Lake headquarters, Li established bases in Changzhou, Ningbo, Chongqing, Shenzhen, and Hong Kong — systematically constructing comprehensive entrepreneurial support systems for engineering students still in their ivory towers, covering everything from startup methodology to supply chains to funding. In 2022, he launched the XbotPark Guangzhou base (the Greater Bay Area National Innovation Center Intelligent Systems Innovation Base), pushing further into integrated industry-academia-research development.

This path is equally long, but that's precisely why Li's story matters: at a moment when China is reshaping its industrial landscape, his narrative reveals the true submerged forces driving industrial transformation, and the sieges and obstacles they face.

Will Li achieve success on a broader scale?

It seems too early to tell. But his student, QKM Technology founder Jinbo Shi, says, "He tossed a pebble into still water — at least the water changed." In any case, she hopes Li will have enough resources to keep experimenting, "let his bullet fly a while longer," because "he runs faster than anyone."

The Secret of a Single Course

In 2019, Zhe Chen, managing director at 5Y Capital, first met HAI Robotics founder Yuqi Chen. What struck him was that despite HAI's unremarkable state — little revenue, business just getting off the ground, the company hanging by a thread — Yuqi was "extraordinarily proud." When asked questions, he cut straight to the point: "I spent four years building just one product."

In Chen's view, HAI's product "was completely different from the wave of copycat Kiva solutions in China. Both in principle and structure, the design was highly innovative — but because of the engineering difficulties, no one had ever successfully commercialized it before."

Also in 2019, Source Code Capital partner Kaishi Chang came across a Narwal product on Douyin. After researching it, he found "this was entirely original innovation from first principles, not the micro-innovation of imitating foreign products."

Both HAI and Narwal were early residents of the XbotPark robotics base. To some extent, these founders' obsession with product — seen by some as unrealistic — fit the archetype Li favors.

This also recalls Li's breakthrough discovery: DJI's Frank Wang, found in the classroom.

P.C. Pao, the HKUST professor and former dean of engineering who co-invested in DJI, sees Wang's discovery as largely "a kind of accident" — but his success gave Li the "unique feel" for finding more Wangs.

And that unique feel largely came from a distinctive competition.

An earlier anecdote about this competition: in 2014, when Sequoia Capital founding partner Michael Moritz flew to Shenzhen and asked Wang how he could help, Wang unexpectedly requested assistance in organizing a global robotics competition — because he himself had competed twice and benefited enormously.

This competition, Robocon, held the secret to Li's talent selection.

Typically, Robocon rules require teams to develop several robots within ten months. Students from different majors — mechanical, electrical, computer — form a team, coordinating progress, project management, logistics, and publicity.

These rules amount to a miniature entrepreneurship trial: "clear goals, limited resources, limited time, non-unique methods" — filtering for those with the innate talent and constitution for founding.

Indeed, Robocon Asia-Pacific contestants went on to build much of China's robotics industry: Geek+ (logistics robots), Ninebot (self-balancing vehicles), Shenzhen Norch (inspection robots), Pudu Robotics (food delivery robots), AgileX Robotics (mobile platforms), INNFOS (linear actuators), and ForwardX Robotics (machine vision).

The competition also led Li to develop a matching course: a robotics competition and design class. Wang took it twice, winning the Hong Kong regional championship in 2004 and international season third place in 2005. During this process, he led development of a helicopter flight control system — later the foundation of DJI.

This course, mirroring the competition's structure, exemplifies the "project-based" teaching model, perfectly aligned with three key features of new engineering education: interdisciplinary integration, hands-on capability, and supply chain management.

In Zhe Liang's view, founder of Qinuo Dynamics, Robocon and this course's greatest value was transforming how students are evaluated — making hands-on ability and problem-solving skills surpass the familiar metrics of credits or paper publications.

According to Wan Xiaokang, Li's student and co-founder of ePropulsion, for a long time Li had recruited students by conventional criteria like elite schools and high scores. But after Googol and Robocon, his standards shifted.

Starting in 2006, Li began recruiting many students from university Robocon teams.

Having grasped Robocon's magic, Li later tried to productize the course. From 2014 to 2018, combining his robotics cultivation and incubation experience, he gradually explored a project-based learning curriculum model. Through research exchanges with MIT, Stanford, and Olin College — American schools successful in engineering education — he refined this methodology.

After integrating design thinking, engineering thinking, and systems thinking — core innovation competencies — he localized it through repeated iterations teaching at XbotPark bases, eventually forming various hard-tech entrepreneurship programs of different durations.

These programs, rotating through different stages, help newly admitted incubatees learn to define products and integrate supply chains, while also serving as an "extended interview" to help Li identify the most suitable people.

In 2015, Junbin Zhang, a 25-year-old Shanghai Jiao Tong University graduate student, came to the Songshan Lake XbotPark base as its first introduced entrepreneur, having written Li a self-recommendation letter.

In the view of a longtime base operations staffer, Zhang perfectly fit Li's founder archetype: young, just out of school; with team management perspective and all-in spirit; preferably battle-hardened by robotics competitions.

The Optimal Position

Beyond DJI, Li was also involved in another student startup around the same time: Bioray Precision. It ended in failure.

A semiconductor packaging equipment startup, it was described by many as a company Li invested tremendous passion in. But the outcome led him to reflect and conclude: projects must be led by students out in front, and sometimes letting go entirely leads to better development.

From buying a Harvard textbook, How to Start a Business, for Googol's early days, to varying degrees of involvement with DJI and Bioray, Li kept exploring how an engineering professor should participate in entrepreneurship and innovation. This is also the crux of many critiques of so-called "professor entrepreneurship."

When An Yong Waves asked Li's student founders, "Does Professor Li interfere in company operations?"

They all answered quickly: "No."

Even back at HKUST, Lab 3126 differed from other labs in that Li rarely assigned "you do this." He asked: "What do you want to do?"

In his students' view, when facing difficulties, Li was never the person to seek empathy from — his answer was always: This is your problem, you decide, you have to decide yourself.

An investor who has incubated many scientist entrepreneurs told An Yong Waves that many scientist founders fail because they cannot overcome excessive ego and control tendencies. Li clearly solved this problem earlier.

He defined Li's role more as a "guardian of the wheat field" — a role that often provides the most effective shelter for innovation. Li's incubation is free-range, because he's convinced: "There's no such thing as a trained entrepreneur; I nurture you as a seedling, you must grow yourself."

Having found a professor's optimal position, what gives Li higher odds of success — to use a popular phrase — is: persisting in what is difficult but correct.

Jie Gan, associate dean at CKGSB who has co-invested with Li in many deals, told An Yong Waves that after visiting dozens of Shenzhen incubators, she found Li's secret also includes that from building the Songshan Lake base in 2014, he emphasized original technology — "this is his biggest difference from others."

In hardware entrepreneurship — a heavy-model domain with complex chains and long cycles — and in China, a society prone to homogeneous competition and relationship-driven experience, this amounts to a bold, even naive experiment.

Li's confidence in originality initially came from DJI.

Once, Li, P.C. Pao, and Wang were eating at a Shenzhen restaurant. During the meal, Pao remarked with feeling: "Decades from now, looking back at Chinese tech history, DJI will be seen as a watershed" — because it was the earliest proof that "Chinese young people can make world-class products and technologically advanced companies."

Behind this lay Li's discovery of another harsh truth: though the Greater Bay Area had built an unparalleled manufacturing supply chain through decades of contract manufacturing and knockoff production, without its own brands, it could only capture tail-end profits. In mobile phones, for example, Apple captured 60% of profits, while Chinese contract manufacturers got less than 2%. So as early as 2014, he tried to lead young people to change this situation through "original innovation" — to earn the money from defining products at the chain's apex.

But young people's disadvantages were obvious. At the time, the venture capital world was still in the final frenzy of mobile internet dividends, with fast-scaling internet projects and seasoned entrepreneurs from big companies center stage. Occasionally an investor would visit, see a group of student founders buried in small appliances and hardware products, and the assessment would often be: really reliable?

Late 2018 was Narwal's most critical funding round. With the product not fully ready, Zhang flew to Beijing frequently, meeting roughly twenty to thirty investment institutions, but few engaged seriously. One investor who met Zhang at the time described him as carrying a backpack, "very introverted" — which made him highly hesitant.

To address funding difficulties, Li, Pao, and Gan established Clear Water Bay Venture Capital. If a project was approved, it could receive funding to enter exploration; if it reached angel round, it could get roughly 2–5 million yuan.

In that late 2018 Narwal round, one institution jumped ship halfway. Li stepped in to fill the gap.

The hesitant investor, later recalling his misjudgment of Yiming Zhang as "a programmer who didn't look like he could achieve something big," went back to chase Narwal — but couldn't get in anymore.

Afterward, Narwal completed five rounds from Series A to E between 2019 and 2021 in just three years.

Meanwhile, completing three rounds in one year was HAI Robotics, whose valuation rose from 300 million yuan to $2 billion.

A Flashlight, a Dog-beating Stick

Leading a group of fresh graduates to conquer the world is undeniably risky, so Li constructed an elaborate system to counter this fragility.

In the view of a former base operations staffer, Li's system "is a closed-loop, top-down design." It effectively captures three core elements of hardware entrepreneurship: "product definition, supply chain, and people" — with corresponding deliverable products for each.

Six kilometers from the old Songshan Lake base in Dongguan, Li built a new XbotPark robotics base headquarters covering over 60,000 square meters with 110,000 square meters of floor space, accommodating more than 100 entrepreneurial teams. This was acquired with early earnings — an ideal sanctuary for young people doing hardware entrepreneurship.

It nearly assembles all elements needed for hard-tech entrepreneurship, including pilot workshops, electronics labs, 3D printing labs, and facilities like kindergartens and libraries.

For product definition, Li believes the most critical need is certain mindset shifts. Design thinking discovers opportunities and problems; engineering thinking uses technology to build things and iterate quickly; business thinking judges how to enter, how to close the loop systematically, and generate cash flow and profit; finally, entrepreneurial thinking with courage and boldness integrates everything.

A former base operations staffer told An Yong Waves that companies entering XbotPark initially, beyond basic hard-tech entrepreneurship camp training, have teachers and project managers frequently interact to correct habits from exam-oriented education — like only solving problems without defining them, or doing user research through random interviews or self-indulgence rather than proper methodology, ensuring the overall product definition direction is sound.

Robotics is fundamentally manufacturing. Hence supply chain integration capability becomes indispensable.

In Jun Huang's view, founder of FA Shinen Capital, supply chain is fundamentally "a kind of trust." Large companies gain trust through volume; small ones cannot. The operations staffer told us that supply chain difficulties for base startups could be called a "mismatch of status."

The Greater Bay Area has a mature supply chain, but this applies mainly to mass production. For early-stage teams, a more common reality is: few companies accept small-batch orders, and even if they do, quality issues are severe — entire batches may be scrapped. And different stages require matching different suppliers, which consumes enormous mental energy for startups.

The base's approach to solving supply chain problems is establishing a pre-graded supplier database, while also "packaging" base startups as a collective B-level customer to interface with major suppliers.

Given that each company's needs aren't standardized, Li specifically planned 10,000 square meters of underground factory and 12,000 square meters of shared factory for small-batch, multi-variety problems in the headquarters design.

How does this system keep running? Not just from continuously evolving setup, but also from relentless spiritual force.

Like patience. In Narwal's third product development year, some investors grew restless and pressed for release. When Li visited the factory to see the product, he asked how it was. Zhang said almost there. Li directly said: "If it's almost, don't release."

Like embracing failure. Enzuo, an agricultural technology company in the base, previously had to disband due to policy changes. Li asked the founder if he still wanted to do it, gave him more money when he said yes, and the team reassembled.

QKM's Jinbo Shi says Li's patience for dormant periods comes from seeing founder growth beyond financial statements — something many investors without industrial training panic about.

And in ePropulsion co-founder Wan Xiaokang's view, beyond practical collaboration in human resources and supply chain, Li's system's greater significance is that it's a spiritual community.

XbotPark's early founder meetings were, in many people's view, essentially misery-sharing sessions. Often someone would finish fifty slides saying "four founders, now it's just me," followed by "the company has two people left."

In one interview, Li said tech entrepreneurship is like walking at night — "one person is very afraid, but two or three or five people together, plus giving you a flashlight and a dog-beating stick, and you're not afraid anymore."

The Innovator's Dilemma

Once, Li chatted with Li Xiang, founder of Li Auto. Talking late into the night, their final sigh was: why might 4,000 Chinese engineers not match 300 Tesla engineers?

The answers exist — systematic, engineering capabilities. But fundamentally it's a people problem, an education problem.

After replicating the Songshan Lake headquarters model across multiple cities, Li found that good entrepreneurs and projects weren't that plentiful, and nearly all companies in the system faced talent shortages.

In fact, as early as Googol, Li encountered a talent drought. Students graduated and went to America, so in 2004 he founded the automation department at the Harbin Institute of Technology Shenzhen Graduate School, cultivating students in HKUST's model, hoping they'd be ready for critical moments. When DJI was founded, these students indeed became core members.

DJI's exemplary significance led Li to continue these educational reform experiments. With greater influence, he tried using unicorn incubation's industrial demands to reverse-engineer new engineering education reforms. In recent years, with six bases including Songshan Lake as centers, Li collaborated with surrounding universities on new engineering and innovation education — from new courses to pilot classes to joint colleges. Shenzhen InnoX Academy is Li's latest experiment. It partners with universities, attracting promising but "three-no" young people — no team, no idea, no resources — helping them find problems they care about and found hard-tech companies.

Many of these projects progress slowly. In a sense, this is the pioneer's predicament. An operations staffer involved in early promotion told An Yong Waves that implementation difficulty lies in needing support from top level (local government and schools) to middle level (deans, academic affairs directors) to bottom level (teachers). But reality is everyone juggles multiple roles, and innovation isn't the main one.

For universities, the biggest driver is often winning innovation competitions or direct benefits from partnering with mature companies, not new teaching model experiments. For teachers, their evaluation systems still revolve around papers and funding. Many university teachers crossing into commerce are even seen by peers as going to make money, "getting dirty."

"Industry-academia-research integration" has been shouted for years, but channels remain heavily obstructed.

Much missing commercial value lies in schools being closed systems. An entrepreneur who studied and worked in America for many years told us that in the U.S. — especially near California — flow between schools and industry is very frequent. Many teachers who spot good entrepreneurial directions can start companies, "and even if they fail, they can come back to teach."

Googol Technology was, in a sense, a victory of this "mobility." At the time, HKUST, building on UC Berkeley and Stanford, had established technology transfer policies allowing professors one day per week for external consulting and professional activities.

New engineering education promotion is equally difficult because it challenges collective inertia. Especially as the involved population expands, the system needs more resources, manpower, even an entirely new evaluation system.

Even in America, though Olin College has operated for 20 years with graduates often outearning MIT and Stanford alumni, it remains a small, beautiful college: only 300+ undergraduates, no graduate students, limited to three degrees — engineering, electrical and computer engineering, and mechanical engineering — difficult to replicate at scale.

Therefore, one of his students told An Yong Waves that rather than building a comprehensive university, Li might be better suited to focusing on an Olin College-model school, using his robotics insights and accumulation to cultivate newer forces.

"Patriot Outside the System"

In 1978, Alcoa visited China and, departing, offered two college scholarships. Li Zexiang, a freshman at Central South University of Mining and Metallurgy, was fortunate to be selected.

This was a dizzying beginning for a country boy from Hunan: first time to Beijing, first time in a car, first time on a plane. Arriving in Los Angeles, seeing the bright lights, it was like another planet.

In America, Li was shocked by the vast gap. As a representative of his country, unable to understand lectures or complete assignments, he nearly broke down under the pressure.

Then the host family's husband saved him, teaching him to run. He ran every day, through snow. "Running and running, I came through."

At Carnegie Mellon in Pittsburgh, a course once required students to spend three hours at an intersection and write a report. Li sat there foolishly, wondering whether to count pedestrians or vehicles — this type of course was his weakness. Only later did he learn this was design thinking: training to discover problems through observation and reflection.

At UC Berkeley, he learned critical thinking and mathematical tools. As a postdoc at MIT, he saw the full engineering process. So for Li, educational reform wasn't a sudden inspiration after DJI.

As early as 1986, after completing his master's at Berkeley and seeing the China-U.S. technology gap, Li joined many overseas students in drafting university reform recommendations sent to the Ministry of Education.

In the 2010 Southern University of Science and Technology reform controversy, shortly after President Qing-Shi Zhu took office, Li volunteered to join the founding team, later recruiting two colleagues. At the time, he told media that the three of them, having gone from the mainland to study abroad, shared "a generation's higher education reform dream."

But in 2011, due to various disagreements, the three teachers decided to withdraw.

When everything seemed interrupted, DJI's success reconnected it all. DJI convinced Li that Chinese young people could absolutely make world-class products and companies.

Due to early accumulation, Li had already achieved financial freedom. But this changed nothing about him.

Wherever he goes, he still dresses like a veteran craftsman, carrying an old, heavily worn dark red backpack everywhere, habitually driving himself.

Hiking is Li's most passionate hobby. He takes students hiking, gives interviews while hiking, even uses hiking to interview people. Lab 3126 was thus once called the Clear Water Bay Sports School; the Songshan Lake base became the Songshan Lake Sports School.

Hiking mirrors his disposition. "Come, I'll take you on a shortcut." When he says this, his students know they're likely embarking on "a terrifyingly wild path."

An early startup founder at the Songshan Lake base told An Yong Waves that one year, he and Li went to America and discovered a high mountain in Seattle they decided to climb. Completely unprepared, Li was even wearing a suit and leather shoes. When they drove up, it was snowing; others hesitated, but Li vanished in a blink. "Most professors doing research are very cautious, afraid of risk — not like him."

In Jie Gan's view, most professors' work is actually very distant from reality, but because it's what they stake their livelihood on, most continue along established paths. Li is the one who actively breaks this situation, taking the unconventional route.

At his 60th birthday party in 2021, Li summarized his life after 20: before 30, figured out academia; before 40, figured out teaching; before 50, figured out entrepreneurship; before 60, figured out innovation.

Regarding where this patriotism comes from, his students tend to attribute it to the special product of that era, calling him "the biggest patriot outside the system." In Li's own view, for someone with hunger in their growing memories, it's like instinct. "Like refugees who've trekked long, suddenly seeing fertile land ahead" — "what do you do? You don't lie down, don't rest" — "you go develop it, plant it, prepare for the next famine."

His student, ePropulsion's Wan Xiaokang, feels this feedback may also come from Li having received kindness.

In 2011, Wan competed in a robotics competition in Thailand. They lost, China's five-year championship ended, and he cried terribly in Thailand. Li comforted him then. Ten years later, at a national competition, seeing the HIT team captain crying from losing, Wan went to comfort him, repeating what Li had told him.

In his memory, Li also mentions much such kindness. Once on the flight to America, an overseas Chinese seeing Li without even a watch insisted on giving him his own, later treating him to meals, helping him integrate into American life.

"Often, it's these kinds of cycles," Wan says.

Thanks to 36Kr author Su Jianxun for contributions to this article. Image source | Visual China

Layout | Guo Yunxiao