Zhiyuan Robot, led by Peng Zhihui, has achieved what even Elon Musk has not yet accomplished. While Tesla recently confirmed that its humanoid robot, Optimus Gen3, will begin small-scale trial production this summer, with mass production not expected until 2027, Zhiyuan has dramatically accelerated this timeline. The production ramp-up, originally planned in years, has been compressed into months.
On March 30, Zhiyuan Robot announced that its production base had rolled out over ten thousand humanoid robots. This milestone comes less than three months after the company celebrated the production of its five-thousandth unit. Zhiyuan's "ten-thousandth unit" signifies that China's embodied intelligence technology has moved from the laboratory into the stage of scaled commercialization.
For this Chinese company, founded less than three years ago, surpassing an industry titan in mass production is a statement in itself.
**Breaking the Mass Production Barrier**
If we chart the development curve of China's humanoid robots over the past two years, 2023-2024 represents the "Demo Curve," while 2025 marks the beginning of the "Engineering Reality Curve." The distinction lies in the fact that the former relies on algorithms, motion libraries, scene choreography, and rendering to create "effects," whereas the latter adheres to a single standard: continuous, fault-free operation.
Many perceive mass production as a series of standardized processes like assembly lines, mold opening, injection molding, and assembly. However, for the humanoid robot product category, scaling itself is one of the most challenging technical problems. In a post-event interview on March 30, Peng Zhihui, Co-founder, President, and CTO of Zhiyuan Robot, pointed out that scaling in the robotics industry is far more difficult than imagined. He contrasted this with consumer electronics, where a malfunction might simply require a reboot. But for robots, any minor quality flaw can be magnified infinitely during actual operation, potentially causing personal injury or environmental damage with severe consequences.
This difficulty was fully evident during Zhiyuan's mass production phase in 2024. Wang Chuang, Senior Vice President and President of the General Business Division at Zhiyuan, recalled that year as the company's "darkest hour," struggling to scale from the first unit to the 200th unit. After a new product launch in August of that year, where Zhiyuan released five robot models—including the popular Expedition A2 priced over 500,000 yuan per unit—orders flooded in. However, the company was initially unable to fulfill them due to an inability to mass-produce, implement the technology effectively, or establish a viable business model.
Wang Chuang stated that the production line lacked standardization initially. The more units they produced, the less time the team had for productive work because significant time was consumed by repairs. Each robot that came off the line was even different, requiring engineers to calibrate them individually. It was a difficult period where manpower bridged the industrialization gap.
By the end of 2024, to solve the massive problems arising from the first batch of mass production, Zhiyuan had stationed over 150 R&D personnel at the factory frontlines. Each order-of-magnitude increase in scale presented entirely new dimensional challenges.
Subsequently, scaling from 200 to 1,000 units shifted the biggest bottleneck from the production line to the supply chain. Peng Zhihui revealed that when Zhiyuan first decided to develop humanoid robots, they surveyed the industry and found no mature, reliable core component suppliers capable of batch delivery that met their requirements. An internal employee noted that early suppliers might manage to deliver components for one or two hundred units, but when order volumes reached thousands or even ten thousand units, the existing system and quality standards collapsed instantly, making the product unsustainable for large-scale rework.
Traditional joints, reducers, dexterous hands, and batteries—no single supplier could meet the stability requirements for thousand-unit-level deliveries, let alone consistency.
With no ready-made path, Zhiyuan had to build its own. Peng Zhihui said, "We grew together with our supply chain, bringing them in for joint R&D." They redefined everything from material processes and tooling fixtures to test benches and aging procedures. For core components like joints and dexterous hands, Zhiyuan adopted new processes developed jointly with suppliers, making the parts lighter, more precise, longer-lasting, and lower-cost. They also established a "half-hour supply circle," requiring core suppliers to be able to respond within 30 minutes. This deeply integrated supply chain strategy, while resource-intensive in the short term, constitutes Zhiyuan's moat in the long run.
**The Beginning of Engineering Compound Interest**
If supply chain capability determines whether Zhiyuan can produce ten thousand robots, then the compound interest of the "data flywheel" determines the value these ten thousand robots can generate after production.
"Previously, the approach to building robots was to first create a 'body'—the hardware—and then develop and insert some 'brain,' models, and algorithms into it. But now, with ten thousand units produced, the body and brain are evolving simultaneously," Peng Zhihui stated.
Now, every robot produced is deployed in automotive manufacturing lines, 3C electronics workshops, and commercial service spaces, continuously collecting data. This real-world data, in turn, trains Zhiyuan's base models at an unprecedented speed, making the models more generalized and practical, thereby enabling the robots to tackle more complex scenarios. "The robots will become smarter with use," Peng said. "Ten thousand units is the critical node that gets our flywheel spinning completely."
This logic is highly consistent with the evolution path of the autonomous driving industry. The qualitative leap of Tesla's FSD12 around 2025 was primarily due to the continuous contribution of real driving data from millions of vehicles on the road. Currently, Zhiyuan is replicating this path in the humanoid robot field.
In Wang Chuang's view, any disruptive technology in its early stages seems to progress painstakingly slow. But once it crosses a certain singularity and arrives like a tsunami, everyone is astonished by its speed. Just like electric vehicles and smart driving before, humanoid robots are approaching this explosive tipping point.
According to IDC data, global humanoid robot shipments in 2025 are expected to reach nearly 18,000 units, a year-on-year increase of approximately 508%. Chinese companies dominate global shipments, surpassing their American counterparts. In 2026, the industry enters a critical period for scaling mass production. TrendForce predicts global shipments will exceed 50,000 units, a further increase of over 700% year-on-year.
Having achieved the ten-thousandth unit milestone in less than three months into the year, Wang Chuang predicted at the launch event, based on this pace, that "one hundred thousand units might be possible by the end of 2027." This prediction is based on two premises: first, achieving a fully autonomous deployment state where robots operate without human control, understand their environment, charge autonomously, and continuously adapt to more complex tasks; second, globalization. Wang Chuang said, "The demand for robot categories is universal globally. Aging populations, declining birthrates, labor shortages, and the increasing difficulty in filling dull, repetitive positions are global issues."
From the Expedition A1 launched in August 2023 to the ten-thousandth Expedition A3 rolling off the production line in March 2026, Zhiyuan has achieved a nearly impossible leap in less than three years. The company's story is essentially another concentrated release of the Chinese manufacturing system's capabilities in a brand-new product category—a script that has been enacted many times before, yet remains震撼ing each time.
Ten thousand units is not the finish line. But from this day forward, humanoid robots are no longer just a futuristic concept; they are a product of the present.
Comments