20 Years of Entrepreneurship, One Year as CEO, He Increased Company Profits by 446%

Ask AI · How does Wu Fengli balance strategic resolve with rapid iteration?

“Look inward, I change first”

Text | Reporters Kong Yuexin from China Entrepreneur

Editor | Ma Jiying

Header image source | Interviewee

Every day, as long as conditions permit, Wu Fengli, Chairman and CEO of Topstar, puts on running shoes and runs 10 kilometers. He has maintained this habit for over a decade.

“Initially, it was for weight loss, but later it became a way to ‘detach’.” Wu Fengli told China Entrepreneur that during exercise, he can merge with the outside world—feeling spring in spring, summer in summer. More importantly, in a state of emptiness, he can step back from concrete work and engage in abstract thinking: “Finding a balance between fast and slow, opportunity and crisis, heat and calm.”

This sense of balance may be Wu Fengli’s stance in facing the current AI wave. Since founding in 2007 and listing on the Shenzhen Stock Exchange’s ChiNext in 2017 as “Guangdong’s First Robot Stock,” Topstar has developed four core intelligent equipment categories: embodied intelligence, industrial robots, CNC machine tools, and injection molding machines. It has become a full-stack industrial robot company mastering control, servo, and vision technologies, with total shipments exceeding 150,000 units. As of March 23, its market value was 11.643 billion yuan.

In 2023, amid the surge of AI large models, Wu Fengli led Topstar through a transformation, divesting its early core business—smart energy and environmental management systems (referred to as “green energy business”), and developing a new generation of motion controllers deeply integrated with AI, launching industry-specific models and the X5 robot control platform.

In August 2024, Wu Fengli, as Chairman of Topstar, resumed his role as CEO. “Previously, competition was less fierce, and decision-making had some gray areas. But now, with industry intensifying, the arrival of the AI era, and traditional businesses at a crossroads, stagnation will only lead to narrower dead ends.” For Wu Fengli, the key value lies in how to turn around.

The results of the transformation accelerated in 2025. In September 2025, Topstar launched its first humanoid robot, “Xiao Tuo,” the world’s first embodied intelligent robot applied in the injection molding industry.

Now, standing at the cusp of explosive embodied intelligence, Wu Fengli takes another crucial step. In January 2026, Topstar officially submitted its listing application to the Hong Kong Stock Exchange, aiming for a “dual A+H” platform. Wu Fengli hopes that Hong Kong listing will broaden international financing channels and integrate into the global industrial ecosystem.

In the first three quarters of 2025, Topstar’s net profit attributable to shareholders was 49.34 million yuan, a year-on-year increase of 446.75%, turning losses into profits. Additionally, overseas sales revenue accounted for 25% of total revenue, with sales networks covering over 50 countries, making it a supplier to well-known companies such as Bourne Optical, Luxshare Precision, Foxconn, Samsung, Corning, and CATL.

Source: Visual China

“Listing in Hong Kong is not just about financing; it’s a key step to connect with international capital and rules,” Wu Fengli said. He aims to leverage the international capital platform to accelerate technological iteration and global penetration in this wave of intelligence.

Below is an excerpt from Wu Fengli’s interview (with edits):

Cutting the Hand

China Entrepreneur: In 2025, you launched the humanoid robot “Xiao Tuo.” When did the idea of entering the humanoid robot space emerge? What motivated your decision to go all-in?

Wu Fengli: Using “all-in” might not be entirely accurate; internally, we prefer to call it “finding the right scenario, laying eggs along the way, forming small closed loops.”

Since founding, our business expansion has not been driven by chasing trends but by actual customer needs. Entering embodied intelligence is the same—within our core fields like injection molding and 3C, there are many flexible assembly and complex sorting scenarios that are excellent entry points for humanoid robots.

Our strategy is “scenario-driven, small steps, quick iterations.” We don’t aim to produce a universal humanoid robot in one step; instead, we identify one or two scenarios that can truly create value, let the robot operate in specific environments, solve real problems for customers, and continuously iterate through this process.

New businesses are not sustained by headquarters “blood transfusions,” but by forming their own commercial closed loops. Meanwhile, we also have stable revenue streams from injection equipment and CNC machine tools, allowing us to earn while developing, making our growth more stable, long-term, and sustainable.

China Entrepreneur: Recently, Topstar implemented “focusing on products and shrinking projects,” actively reducing its smart energy systems business. Was this “slimming down” driven by the need to focus on embodied intelligence?

Wu Fengli: “Shrinking projects” indeed brought some short-term performance pressure, but this was not a strategic wavering. It was a necessary choice based on our return to the essence of manufacturing and long-term judgment of future competitive landscapes.

Smart energy and environmental management systems tend to have high-value orders and can scale easily, but they are also characterized by low standardization, limited gross margins, and long delivery cycles, which can adversely affect net profit and cash flow. Starting in 2023, we gradually reduced this business, and now we have largely divested, leading to a healthier business structure and improved operational quality. According to our 2025 earnings forecast, we expect a net profit of 60-80 million yuan, turning losses into profits.

While shrinking these projects, we focus resources on three main product lines: industrial robots, injection molding machines, and CNC machine tools. The biggest difference between “products” and “projects” is that products are general and standardized, capable of serving numerous industries through core technology spillover effects.

“Slimming down” is essentially strategic focus. During the painful transition, we sacrifice short-term scale figures to build long-term core competitiveness. When external conditions improve and industry upgrades accelerate, a healthier business structure, stronger technological barriers, and more stable profitability will position Topstar better for future growth.

China Entrepreneur: Did your move from industrial robots to humanoids come after the explosion of large models? Or was it an existing idea?

Wu Fengli: Strictly speaking, we had considered it earlier but didn’t take the step until AI arrived. Because once you decide to do it, you need to invest, but listed companies don’t support unlimited investment, nor do they have endless firepower for “saturation attacks.”

So, we balance our investments, and that’s also why I personally got involved—only by personally engaging can I maintain a sense of balance.

China Entrepreneur: What challenges did you face during R&D? Did your previous experience with industrial robots give you an advantage?

Wu Fengli: Our past experience is definitely valuable, especially in supply chain management, standardized large-scale operations, cost control, and manufacturing capabilities, giving us significant advantages.

But currently, we are still in prototyping and scenario validation stages, so these advantages haven’t fully manifested yet. Like Huawei, which has a manufacturing gene, they’re often not the fastest at the start but tend to stay longer and go further. We are similar—never a rapid breakout company, historically not investing much in advertising or promotion, but able to develop steadily because of long-term accumulation.

China Entrepreneur: “Xiao Tuo” uses wheel-based design instead of bipedal. What scenario considerations led to this choice? In your view, what should be the core of humanoid robots in industrial scenarios?

Wu Fengli: All our considerations are based on real industrial scenarios. Over the past 20 years, we’ve served over 15,000 companies and reached 200,000 clients. Through extensive scenario research and practical feedback, we’ve realized that industrial demands for robots require balancing efficiency, stability, and cost—this led us to choose a wheeled design.

Compared to bipedal forms, wheels are more aligned with industrial logic: higher mobility efficiency, longer endurance; inherently statically stable, avoiding falls and balancing issues; and better cost control. Most industrial floors are flat, further enhancing the suitability and practicality of wheeled solutions.

China Entrepreneur: Data is one of the biggest challenges in robotics today. With many manufacturing clients, do you have an advantage in data collection? What are your current data collection challenges?

Wu Fengli: Serving many manufacturing clients across diverse scenarios, I have a large sample size, which helps in data collection and model iteration. The more realistic the scenarios and the richer the data, the smarter the robots become, enabling better service to my manufacturing clients and continuous iteration.

But we also face significant challenges: first, manufacturing scenarios are fragmented and non-standardized, making data standardization difficult; second, high-quality labeled data is scarce, and efficiently and cost-effectively transforming on-site data into training data remains a common industry challenge.

China Entrepreneur: Do you collect data by placing robots directly in clients’ factories? Many companies face privacy restrictions, making factory access difficult. How do you handle this?

Wu Fengli: Yes, that’s what I call value exchange—clients provide data, and you deliver benefits in return. This has been a path we’ve wanted to explore, though it’s challenging. Data is extremely valuable for model and product iteration.

China Entrepreneur: What are the challenges between laboratory R&D and large-scale production?

Wu Fengli: Topstar’s products don’t go directly from lab to market; they grow from market needs. For example, wheeled robots and quadruped robots are first defined by scenarios and customer demands, then we develop the product and refine the technology. The desired product form and performance are guided by customer feedback. Our product development logic always starts from real customer needs, pain points, and scenarios.

Source: Visual China

China Entrepreneur: Regarding embodied large models, there’s a debate: one side advocates starting with vertical scenarios and gradually expanding; the other favors building a comprehensive general large model. How do you see these routes?

Wu Fengli: If the goal is to build a platform company, the latter is the core objective; but if it’s a startup, I believe it’s essential to assess your “ammunition”—funds and resources—to support results.

Startups should focus on “small entry points,” emphasizing “one meter wide, one thousand meters deep,” solving specific customer pain points. We began by addressing a very specific problem: clarifying the issue, defining the product, then the technology, and then testing with customers, iterating based on feedback—this process is very precise.

Many entrepreneurs today don’t even know who their customers are. The fundamental questions are: first, who are your customers? Second, what problems can you solve for them? Without clear answers, the entrepreneurial journey will be very difficult.

Diving In

China Entrepreneur: Building models and humanoid robots requires significant investment. As a listed company, did you face doubts or opposition from shareholders? How did you respond?

Wu Fengli: That was a tough decision. Listed companies face quarterly financial pressures, which is why we divested the “green energy” business: firstly, it was not well aligned; secondly, resources, especially talent, are limited, so we must focus on core areas.

When making that decision, it felt like “cutting off the hand,” but afterward, deciding to invest in core vertical models and AI became easier. Still, challenges remain—robotics is not highly profitable, requiring heavy talent and capital, with low returns and long cycles. But I believe that technological innovation is the direction, deepening core businesses is essential, and long-termism is the foundation. Holding onto these three principles gives us resolve, avoiding being swayed by short-term gains or quarterly results.

Of course, we also need to align with market expectations and plan our annual shipment targets.

Source: Visual China

China Entrepreneur: What was the biggest skepticism from shareholders or senior management at the time?

Wu Fengli: The main skepticism was about my divestment of the “green energy” business. Some saw it as a stable revenue source, but I chose to reallocate resources into vertical models and AI, which are more uncertain, and paid less attention to the assured profitability of green energy.

My reasoning was that traditional green energy businesses would distract us and drain resources. Only by focusing on core tracks can we build long-term competitiveness. This strategic difference caused some internal and shareholder doubts.

This is the difficulty of long-termism, but also its correctness—thinking long-term, not just short-term gains.

China Entrepreneur: When did this philosophy take shape?

Wu Fengli: Over my 20+ years of entrepreneurial twists and turns, I gained deep insights. Except for 2024, when we divested, we have been profitable every year since inception.

But I realized that some solutions we see as optimal now are just short-term stopgaps; if we only pursue immediate benefits, we may neglect long-term development. That’s why I started adopting a longer-term perspective.

In August 2024, I resumed the CEO role, having been Chairman before. That year, a question haunted me: Should we aim to build a long-lasting enterprise or just a profitable business?

We used to focus on monthly and yearly profits, and while we continued technological innovation, I wondered: can we avoid “big but shallow” products and instead develop “small but deep” ones? This is a strategic trade-off I pondered for a long time. Ultimately, I decided we must be a product-driven company, not project-driven. To do that, we need to make strategic choices about what to keep and what to abandon.

China Entrepreneur: Did you discuss your decision to become CEO again with others, or was it a personal conclusion?

Wu Fengli: I believe such strategic choices are fundamentally based on deep industry and self-reflection, as well as accumulated experience.

When market competition is less intense, decision-making can have some gray areas, allowing trial and error, “letting the bullets fly a bit longer.” But as competition intensifies and at critical junctures like the AI era, our traditional business faces a crossroads: one path is stagnation—seems sustainable but leads to a narrowing dead end; the other is decisive pivot—initially narrow but with a long horizon, and persistence will open broader prospects.

Deciding “how to turn the car” is precisely the core value of the first position—making the right directional choice at key moments. But more important than choosing the direction is how to implement it once decided. Direction is a one-time decision—whether to go straight, left, or right—once made, the real challenge is how to act.

Deep, long-term thinking is needed to understand the challenges, costs, and resources required. Only by personally engaging can one withstand internal and external pressures, mobilize resources precisely, and execute step-by-step according to the strategic plan, ensuring the chosen path is walked through, stabilized, and made into the future.

China Entrepreneur: After taking the top role again, what is your biggest personal change?

Wu Fengli: I pay more attention to details and prefer to connect dots rather than focus on isolated issues. Often, things don’t happen in isolation; when you truly get involved and handle small matters, you find the “feel” and understand the nuances.

More importantly, in managing these specifics, I’ve learned to step back at the right moments, integrating seemingly unrelated elements and analyzing holistically. When stepping out of details and viewing from an abstract, macro perspective, the “feel” accumulated from practice helps me sense the temperature and details of matters, restoring their true form; meanwhile, the macro view allows me to see the full picture, piecing together fragmented information and experiences. It’s like “blind men touching an elephant,” but we aim for “blind men piecing the elephant”—seeing beyond local parts, understanding the whole, and looking further. Only decisions made this way are more objective and balanced.

China Entrepreneur: Can you give an example of a problem you’ve connected holistically in the past year?

Wu Fengli: For example, humanoid robots. It took me over ten years to make industrial robots profitable, turning core technologies into real market competitiveness. Continuing with traditional methods is fine—deeper technical accumulation over time.

But the advent of AI has shattered this inertia. Its technical route is entirely different: no longer just programming, but autonomous decision-making, large models, big data—fundamentally changing everything. At this point, the key question is the direction: do we turn around and fully embrace AI? Or run both in parallel—maintain traditional and innovate? Or deeply connect existing businesses with AI to enable mutual empowerment?

Different choices lead to different futures. If we stick to traditional programming robots, solid in product, core tech, and market, we can keep making profits, but risk becoming a traditional tech company missing AI opportunities; if we abandon core businesses and go all-in on AI, it seems trend-following, but many startups without legacy burdens start with humanoids or quadrupeds, rising with the trend, and can switch tracks easily. As a listed company with a core business, losing that foundation—even if embracing AI—may undermine our footing. Conversely, clinging to the old and resisting change risks missing the AI era altogether.

So, perhaps the third way: maintain core business and fundamentals, while connecting AI through investments or partnerships. But AI’s technical path and our existing tech are completely different, requiring different talents. How to connect them well, how to fuse talent and manage scientifically—these are urgent issues. It’s not just a technological paradigm shift but a comprehensive change in management and business models.

Facing such choices, if I don’t personally get involved and understand the new tech and directions, I can’t find the integration points between old and new, nor convince internal staff. And staff may find it hard to abandon old views and actively embrace new tech and ideas.

Therefore, as CEO, there’s no one else I trust more—I have to lead myself.

CEO Management Requires a “Knife”

China Entrepreneur: Regarding talent organization and management, what specific difficulties or challenges do you face? As a traditional robotics company transforming, what talent issues have you encountered?

Wu Fengli: Our controllers, servo drives, and vision leaders are carefully selected experts in their fields. But everyone understands that vision has its own domain, control has its core boundaries, servos have their technical edges, hardware like mechanical structures has its own track, and application personnel must be familiar with scenarios and processes—these are separate domains.

I want to say that talent must be viewed dialectically: we need deep expertise and solid experience in each field, and truly understand the industry. But the problem is, the deeper someone dives into a domain, the more they risk being confined within their own cognitive world—“drilling too deep, unable to come out”—becoming overly confident or even stubborn. They believe they are the best in their area, which makes talent integration a key challenge. How to connect them, how to help them see beyond their own limits? We need to guide them to detach, realize they see only part of the picture.

China Entrepreneur: Do you have new methods for organizational management?

Wu Fengli: Still exploring. It involves constant discussion about future directions, and inevitably some disagreements and clashes. Ultimately, shared values and goals are what unify us. This process challenges their existing cognition and may require sacrifices; management models must also change, which is difficult for experts across fields. How to help everyone understand and be willing to change? Sometimes, individuals can change, but their teams may not.

I once said that the CEO must lead with a “knife,” which is about making choices. My approach is to look inward first—I change myself.

China Entrepreneur: What’s the biggest change in your behavior?

Wu Fengli: I now focus more on the battlefield map—market share of products. I track daily changes in market share, industry trends, product development progress, down to individuals and days.

We spent over half a year creating this battlefield map, with separate charts for overseas and domestic markets. For China, it includes prefecture-level cities, their GDP, total industry market size, our current market share, customer distribution, target industries, daily progress, and growth rates.

It’s not just about viewing the map—I also make decisions based on the data. We hold weekly meetings to analyze entry points, market layout, resource allocation, etc.

As Chairman, I also serve as CEO, product manager, marketing lead, and sales service head. Since taking on the CEO role, I pay more attention to talent—interviewing all senior managers myself, sometimes for over a year without finding the right fit.

China Entrepreneur: Can you keep up with the workload?

Wu Fengli: We used to operate with a divisional system, each responsible for a line, and I only coordinated among leaders.

Now, we’ve adopted matrix management, integrating HR, IT, and all functions into resource-sharing platforms. In this setup, only I can connect all departments and resources, so I must personally get involved.

Source: Interviewee

China Entrepreneur: After the advent of AI, how do you feel? Are you anxious? How do you cope with anxiety?

Wu Fengli: As a long-time entrepreneur, I was naturally anxious when AI first appeared—it brought huge uncertainty. But as I learned more, I saw it as a tremendous opportunity. We’ve experienced the industrial era, the internet era, and now the AI era. Being at the forefront of robotics, we already have an advantage.

But I realize the time scale has changed completely: manufacturing usually plans by year, but in the AI era, progress is weekly, with rapid tech changes and iterations.

It’s impossible not to feel anxious, but that anxiety is also our greatest motivation—it keeps us thinking, alert, and iterating at high frequency. The world changes so fast that a moment’s oversight can be disastrous.

Years of experience also tell me: maintain strategic resolve on big directions, but be swift in small adjustments and resource deployment. Balancing this strong resolve with rapid iteration is a challenge and a joy for entrepreneurs.

Gradually Connecting Business Scenarios

China Entrepreneur: Are your humanoid robots still mainly B2B?

Wu Fengli: As a profitable listed company, we can’t rely on storytelling or PPTs. I must strengthen our core business (B2B), but at the same time, we are exploring B2C products.

China Entrepreneur: B2C refers to household robots?

Wu Fengli: Of course. We first leverage our manufacturing expertise to find application scenarios in manufacturing and deepen them—our main focus is manufacturing. I see manufacturing as a training ground for robots, but we all know it’s not the end goal.

Our business model is based on two industrial “mother machines”: five-axis CNC machines and injection molding equipment. We chose these because processing metals and plastics in the physical world depends on these two types of equipment; as long as you serve manufacturing, you inevitably use them, and thus, you can’t do without Topstar.

Using these two industrial mother machines, we can cover all manufacturing sectors. All manufacturing clients need labor, and where there’s labor, there’s demand for robots—creating many application scenarios. More importantly, this also helps us collect massive data, aligning with our “scenario defines product” logic.

Six-axis articulated robots and humanoids are different forms: for home use, humanoids are needed because homes and infrastructure are designed around human form, and scenarios dictate the form and function of products.

Our robots come equipped with vision, force sensors, and other sensors, inherently capable of big data collection. With big data, we can iterate models. As models evolve, we can generalize scenarios, enabling robots to extend beyond industrial use into commercial, service, and household settings.

This forms our closed loop: relying on two basic industrial mother machines, covering over 200,000 manufacturing clients; these clients provide vast application scenarios, enabling broad coverage and data collection; then, models are iterated and upgraded, promoting scenario generalization, and moving robots from specialized to general-purpose.

China Entrepreneur: How is the mass production and commercialization of “Xiao Tuo” humanoid robot progressing?

Wu Fengli: We are testing it in several scenarios, such as sorting and handling. Some clients are already trying it out, but I’m not rushing to large-scale deployment because current demand is limited.

We need to do one-on-one training, and I believe slow is fast. We’re not an entertainment company.

China Entrepreneur: Market adoption depends on cost. If the cost can’t be lowered to match or be lower than manual labor, clients won’t use it. How do you address this?

Wu Fengli: We have manufacturing genes. Only when the market size grows will our standardization and fine management capabilities fully emerge.

China Entrepreneur: Everyone talks about going overseas. How is your international expansion?

Wu Fengli: We have nine regional teams covering over 50 countries and regions, each a huge blue ocean.

We follow a “racehorse” strategy—currently Southeast Asia is moving fastest, but that may change. Last year, overseas growth exceeded 50%, much faster than domestically.

China Entrepreneur: How do you educate the market for humanoids overseas?

Wu Fengli: We call it the “211 strategy”—two sales teams (technical + business), one after-sales team, and one demo site. Each region has a team with pre-sales application engineers, post-sales debugging engineers, plus sales and agents. The smallest “battle unit” is three people, with good coordination; they cover customer needs and speak through the product.

China Entrepreneur: What role do you see Topstar playing in the next five years?

Wu Fengli: This year, Topstar officially updated its vision—to become a globally leading embodied intelligent technology company. We will continue to strengthen R&D in embodied intelligence, build full-stack self-developed technology barriers; expand service scenarios, fully leverage process and data assets; and based on real scenario needs, continuously launch and iterate embodied intelligence and other robot products, creating a comprehensive, scenario-rich intelligent ecosystem.