UBTECH ROBOTICS is making a striking impression among humanoid robot companies in 2026. At its annual global launch event in June, the company unveiled its ultra-bionic humanoid robot, the U1 series, available in three configurations: the half-body U1 Lite, the full-featured U1 Pro, and the high-dynamic U1 Ultra. The new product quickly went viral, with pre-orders across all channels surpassing 10,000 units by the end of the month.
In a conversation with the tech outlet Zhike Planet, UBTECH's Chief Brand Officer Tan Min repeatedly emphasized the word "deployment." According to Tan, the company's business strategy has long been anchored on three main pillars: industrial, commercial, and home applications. These scenarios can form a closed loop, where, for example, service experience gained in industrial settings can eventually be applied to home companionship. He noted, "ROI is not the primary concern; what matters is whether we can secure a favorable position in the competition for the next generation of productivity models. We don't necessarily have to lead, but we absolutely cannot fall behind."
The following is an edited excerpt of the dialogue between Zhike Planet and Tan Min, covering UBTECH's development, commercial pathways, ROI assessments for humanoid robots in industrial settings, and the future of the technology.
Zhike Planet: Why did you choose to develop this type of bionic humanoid robot?
Tan Min: We see humanoid robots as a parallel track with two distinct forms: one is mechanical, and the other is a lifelike humanoid appearance. Similar characters have appeared in numerous Hollywood films, and the future we envision aligns with that—both mechanical and realistic humanoid forms will eventually become viable options. When robots become general-purpose humanoids, customers will have a choice: if they prefer a mechanical look for their home, they can buy that; if they prefer a lifelike appearance, they can choose that. However, from the perspective of human companionship, we believe an ultra-bionic humanoid is an essential path. This is the "dual-line coexistence" UBTECH has always pursued. The ultra-bionic robot didn't emerge out of thin air; we've been preparing for it for over two to three years.
Zhike Planet: What are the biggest differences between industrial humanoid robots and these bionic humanoid robots for home use?
Tan Min: The two types of robots are very different. Industrial humanoid robots prioritize practicality and functionality, starting from customer needs to solve real-world problems. For instance, if a robot is meant for material handling, it must address that task efficiently, considering the optimal form and methods. This is why we have both bipedal and wheeled industrial robots—they emphasize functional attributes to solve actual production issues and boost efficiency. In contrast, bionic humanoid robots are more emotion-oriented, with less rational functionality at this stage. We chose to enter the home market by focusing on companionship before service because service capabilities need to be built through vertical scenario applications—specifically, by first integrating into industrial settings. This enables the bionic humanoid to develop the ability to serve in the real world. Therefore, we must first run through closed vertical scenarios before opening them up. Industrial environments are closed or semi-closed, while homes are open, and commercial settings fall somewhere in between. That's why we launched on three fronts simultaneously: starting with industrial, then commercial, and finally home applications. This is a carefully calculated business strategy.
Zhike Planet: For consumer-side (C-end) robots, what are the key challenges in mass production compared to B-end?
Tan Min: Mass production for B-end and C-end is fundamentally the same. The core challenge is that this is an entirely new category with no comparable reference globally. We've designed a new production line specifically for the ultra-bionic robot, with a new operational workflow distinct from industrial humanoids. It's a more complex process due to the bionic nature, and the standard operating procedures are still being refined. On one hand, manufacturing is far more complicated than for industrial robots, involving new materials like bionic skin and core components that have never existed in the market. This includes micro servos, which differ from those used in dexterous hands, and our binocular vision system, which is different from that of industrial robots because our eyes have eyeballs—industrial robots' eyes are more like car cameras. In comparison, this is significantly more challenging.
Zhike Planet: After selling humanoid robots in the C-end market, how will after-sales service be handled?
Tan Min: When we launched our ultra-bionic humanoid robot, we announced a partnership with JD.com. Currently, JD is upgrading and training its delivery personnel for robot maintenance and delivery services, making it the first e-commerce company to offer robot transport and repair. This is a key reason we chose JD as a major C-end sales partner. Additionally, our strategic partner network includes import car 4S dealerships and luxury car dealers, who offer one-on-one service and have over a decade of experience. At the World Robot Conference, we showcased our collaboration with Mercedes-Benz's import agents, who brought dozens of top sales performers for training on-site. We'll soon leverage JD's network and the luxury car import sales channels to build out our C-end market presence.
Zhike Planet: Will you open your own stores in the future?
Tan Min: Yes, we will. C-end consumers still need opportunities to experience our products, like at exhibitions. Our store plans will center around JD and our 4S dealership network. These dealers already have access to a large base of high-spending consumers, so our first experience stores may launch in these locations. We're also exploring flagship stores in prime commercial districts of major cities.
Where the Real ROI Lies: Securing a Favorable Position in Intelligent Manufacturing
Zhike Planet: What's the current status of UBTECH's deployments in industrial settings?
Tan Min: Our Walker S series industrial humanoid robots have undergone three iterations over more than two years, with the S3 expected to launch in the second half of this year. Last year, the S2 achieved mass production and delivery of over 1,000 units—a global first for full-size industrial humanoid robots. This marks a critical milestone in bringing humanoid robots into factories and transitioning to large-scale commercial production. In August, the super-factory in Liuzhou, built in collaboration with Siemens, began production on a 10,000-unit line. I believe the significance of scaled production lies in delivering consistent, stable robots to application scenarios, collecting real-world data, and feeding it back into simulation environments to form a data loop. This gradually builds a world model, creating an ecosystem closed loop. We now have an increasing number of Fortune 500 clients, with applications progressing from material handling, depalletizing, sorting, to machine tending. We've successfully closed the data and training loops. We're now in the fourth year of our five-year plan, focusing on improving efficiency in handling, stacking, loading, and sorting, and scaling deliveries from thousands to tens of thousands of units. I believe that in 3-5 years, humanoid robots will truly enter various industries. Additionally, we've expanded into intelligent manufacturing through our Fenglong brand, developing fifth-generation dexterous hands and robot joints to strengthen supply chain integration. We've also established Xixuan Chuangzhi with Moxin to develop edge chips for robots, with plans for tape-out next year and mass production the year after.
Zhike Planet: You mentioned the "five-year plan"—this is year four. What's the goal for next year, the final year?
Tan Min: The primary focus of these five years is to get robots "employed" in factories—scaled deployment in real scenarios. We didn't jump straight to a general-purpose large model because current computing power, algorithms, and electricity aren't sufficient for a general humanoid robot. So our path is clear: start with vertical applications in factories, focusing on simple, repetitive, labor-intensive tasks like sorting and handling, then move to commercial and home companionship. These five years are about testing and applying in closed or semi-closed environments, building core capabilities and foundational models, and validating AI through real-world applications. In years three and four, we began constructing the framework for three major models, which together form the evolution of our "robot brain"—a key transition from year three to four. The second focus is mass production. Once the 10,000-unit factory is fully operational, we can collect real-world scenario data and deploy at scale, creating a data loop. The ultra-bionic robot, which attracted much attention this year, isn't just about the hardware—it's supported by vision, voice, motion control, and interaction systems, all built on this foundation. We're making steady progress, always toward the goal of deploying humanoid robots, bringing them into homes and empowering various industries.
Zhike Planet: However, the ROI of humanoid robots in industrial settings currently appears low.
Tan Min: The combination of humanoid robots and AI is a disruptive innovation still in its very early stages. Currently, large models are mostly confined to the 2D world—operating on screens like phones and monitors. But entering the real world requires entirely new models and modalities, and this is a two-way interaction. As for ROI, any disruptive innovation is difficult to calculate in its earliest phase. When Tesla first launched, its ROI certainly wasn't better than gasoline cars, yet within just 10-20 years, new energy vehicles have surpassed traditional ones in China's car purchase rates. All disruptive innovations require a long cycle to materialize. So, I believe the key is scenario-based deployment. For intelligent manufacturing companies, ROI isn't the top priority; what matters more is whether they can secure a favorable position in the competition for next-generation productivity models. If they can't lead, they absolutely cannot fall behind. The rise of humanoid robots, on one hand, helps AI connect with the real world and move toward general AI; on the other, it creates a new form of "mental productivity," shaping the paradigm for next-generation intelligent manufacturing. The entire manufacturing industry will undergo comprehensive upgrades, and we can feel this urgency when engaging with Fortune 500 companies. The competition isn't just about "replacing a worker or calculating the cost of moving boxes"—though that is a consideration—but, like the early days of new energy vehicles and smartphones, there are many uncertainties. The true ROI calculation lies in whether your company can occupy a favorable position in the future intelligent manufacturing model and whether your R&D investment keeps you globally competitive over the next 3-5 or 5-10 years. That's what global leaders care most about.
Zhike Planet: How do you view the surge in robot orders? Are industrial applications a false demand?
Tan Min: Demand in industrial settings has been growing since last year, and it's definitely not false demand. First, there's a thirst for real-world data and applying new quality productivity to traditional scenarios. Take our robots moving boxes: these repetitive, physically demanding, and potentially dangerous jobs are ones many people no longer want to do. This labor gap must be filled through technological innovation to find a sustainable path for next-generation intelligent manufacturing. We aim to offer the market a diverse range of products and solutions. You can see robots with dexterous hands, grippers, clamps, wheeled designs, and bipedal forms. The diversity this year indicates accelerated industry deployment, increasingly aligned with cutting-edge needs—this is a very positive development.
Zhike Planet: What kind of company do you hope UBTECH becomes in the next 3-5 years?
Tan Min: Starting this year, you'll see our three-pronged business model in action. With industry as our starting point, we've built the foundational capabilities for humanoid robots, enabling UBTECH to expand from industrial to commercial services, then home companionship, and eventually home services. We hope that within 3-5 years, we'll grow from 3-5 workstations to 10, 20, 30 or more different job types, and from 2-3 industries to over 10 different sectors. Simultaneously, UBTECH's robots will receive excellent training for their "robot brains." Looking further ahead, the global industry is expected to reach an artificial superintelligence milestone around 2030. By then, we'll have prepared our foundational models and scenario applications: after accumulating extensive scenario data and training the brain, the commercial side can surge, and home companionship can too, truly allowing industrial, commercial, and home scenarios to enter scaled production and commercialization. That's the direction we're working toward over the next 3-5 years.