Leapmotor founder and CEO Zhu Jiangming used his company's 2026 tech day on September 16 to push back against the industry's humanoid-robot hype, saying that a robot that can walk is not impressive—one that can make money is. The event, held in Huzhou, Zhejiang, was reported by PCauto.
Walking is 'the dumbest way to move'
Zhu said Leapmotor has the technical depth to enter embodied AI, but has deliberately kept its efforts small-scale and in a 'follow' mode. The company has built a fully self-developed robot prototype, yet Zhu refused to bring it on stage. 'A robot that can walk is no big deal; a robot that can make money is what we will show,' he said, adding that Leapmotor will not chase concepts and will only launch when a viable business case emerges.
He compared current humanoid robots to a one- or two-year-old child, far from the '20-year-old' that can create value, and noted the lack of a clear business model. Zhu was blunt: 'Walking is the dumbest way to move,' citing the high energy consumption of humanoid structures compared with wheeled or deformable designs. Leapmotor, he said, will not 'circle around the human form of primitive society,' but will use its AI and automation know-how to explore wheeled and other practical forms as a potential second growth curve.
Why Leapmotor can afford to wait
Leapmotor's confidence comes from 11 years of in-house development. The company controls motors, controllers, batteries, perception algorithms, and computing platforms, and has built 18 component plants. Self-developed parts cover 65% of vehicle cost, with a 90% platform commonality rate. Since 2026, monthly deliveries have exceeded 100,000 units, and the company has been profitable for three consecutive half-years.
Zhu estimates that intelligent vehicles and humanoid robots share more than 70% of their technology in perception, decision-making, and execution, allowing Leapmotor to reuse its AI algorithms, computing platforms, and motion control. Its intelligent technology research institute has five departments, including one for embodied AI, and a robot prototype is complete. Senior vice president Zhou Hongtao said the robot can dance and run, but 'cannot create real value yet, so now is not the time to release it.'
Three hurdles separate a demo from a profitable product: reliability, cycle time, and deployment cost. Leapmotor is targeting factory automation and logistics first. Industry research cited by PCauto puts the average price of a mid-tier humanoid robot at about 100,000 yuan (about $14,100), with annual costs of 56,000 yuan (about $7,900) and a payback period of 1.5 to 3 years. However, hidden costs such as environmental adaptation and system integration account for 40% to 60% of total deployment, and real-world availability is only 50% to 70%, far below the 99% of traditional industrial robots. A Grant Thornton survey found that businesses generally accept an 18-month payback, and most will not buy if it exceeds three years. Zhu's hard line for factory automation: recover investment through labor savings within three years, or postpone.
Leapmotor's approach contrasts with rivals. Tesla's Optimus is being tested for assembly and handling at its Fremont factory, Xpeng's IRON is slated for mass production in late 2026 and delivery in 2027, and Xiaomi's 'Iron Man' robot achieves a 98% success rate in tightening nuts at its car plant, close to the 99% for humans. Leapmotor, with its prototype and a team of dozens, is choosing to wait. Zhu did not say 'no'—he said 'not yet.' As the embodied-AI sector shifts from dream-based pricing to performance-based pricing, that stance may prove more valuable than a robot taking a few steps.
