110,000 square meters dedicated to humanoid robot production: the factory that XPeng is inaugurating in Guangzhou in 2026 marks the first massive industrialization of machines that walk, grasp, and work like humans. While Elon Musk promises a few hundred Optimus units by the end of 2025, Chinese manufacturers AgiBot and Unitree have already delivered 5,100 and over 5,500 humanoid robots respectively this year.
This shift reveals more than a Chinese manufacturing advantage. It exposes a radically different economic vision: where Tesla still designs its robot as a laboratory product, China already envisions it as an exportable industrial equipment. One detail illuminates this strategy: He Xiaopeng, CEO of XPeng, judges Chinese labor costs still too low to justify automation in his own factories — his robots will first fill showrooms, then conquer foreign markets where labor is expensive.
The Essentials
- XPeng’s Guangzhou factory (110,000 m²) targets mass production of humanoid robots by late 2026
- AgiBot delivered 5,100 units in 2025, Unitree over 5,500, versus a few hundred for Tesla Optimus
- He Xiaopeng is betting on exports: Chinese labor costs remain too low for local automation
- The humanoid robot market is experiencing rapid growth and could represent tens of billions of dollars by 2035
- Targeted applications: logistics, manufacturing assembly, elderly care services
XPeng Transforms the Prototype Into an Assembly Line
XPeng’s announcement stands apart from the spectacular but artisanal demonstrations of Western humanoid robots. The Chinese company, known for its electric vehicles, is inaugurating in the first quarter of 2026 a 110,000 square meter facility dedicated exclusively to the production of bipedal robots. The objective: operational mass production by year-end, with industrial costs compatible with international commercialization.
This industrial approach contrasts with Tesla’s strategy, where Optimus remains developed in research laboratories. Elon Musk regularly promises “thousands” of units, but actual deliveries number in the hundreds of examples, mainly intended for internal testing in Tesla factories. The difference in scale reveals two philosophies: one favors pure technological innovation, the other bets on rapid industrialization of sufficiently mature technology.
XPeng relies on its automotive experience to master the constraints of mass robot production: supply chain management, quality standards, unit cost optimization. These competencies, developed in a sector where China already dominates global manufacturing, prove decisive in moving from prototype to industrial product. The company announces an initial capacity of several thousand units per year, expandable according to demand.
Chinese Deliveries Explode While Tesla Stagnates
The 2025 production figures illustrate this strategic gap. AgiBot, a startup specializing in humanoid robots, claims 5,100 units delivered this year, mainly models intended for logistics and light assembly. Unitree, a direct competitor, announces over 5,500 deliveries, with balanced distribution between industrial applications and consumer services.
These volumes remain modest on a Chinese manufacturing scale, but they far exceed current Western capacity. Tesla Optimus, despite its media coverage, tops out at a few hundred units produced in 2025, according to sector analysts’ estimates. Boston Dynamics, the American technological reference, focuses on niche applications with confidential volumes. European players remain in the research and development stage.
This disparity is explained by different technological choices. Chinese manufacturers favor robust and reproducible solutions, even if accepting lower performance than laboratory prototypes. AgiBot bets on standardized components and proven control algorithms, enabling series production. Tesla, conversely, integrates cutting-edge technologies that complicate mass manufacturing: advanced artificial intelligence, proprietary sensors, composite materials.
Export as Strategy: Too Cheap for China, Perfect for Europe
He Xiaopeng’s statement reveals a striking economic paradox: “Chinese labor costs remain still too low to justify robotic automation in our own factories.” This frankness illuminates the Chinese strategy: produce humanoid robots for export to economies where wages make automation profitable.
Chinese hourly manufacturing costs range between $3 and $7 depending on regions, versus $25 to $45 in Western Europe and $35 to $50 in the United States. A humanoid robot selling between $50,000 and $150,000 becomes profitable in 2 to 4 years in a German or American factory, but requires over 10 years of amortization in mainland China.
This arithmetic guides Chinese commercial priorities. XPeng initially targets American and European markets, where demographic aging and skilled labor shortages create solvent demand. Germany, facing a deficit of 2 million industrial workers by 2030, represents a natural market. The United States, where 76% of manufacturing firms struggle to recruit according to the National Association of Manufacturers, constitutes the other priority target.
This export strategy accompanies a bet on Chinese wage evolution. He Xiaopeng anticipates a rise in labor costs that will make automation attractive on the domestic market within 5 to 8 years. In the meantime, Chinese production perfects itself on foreign markets before conquering its home territory.
Concrete Applications: From Warehouse to Retirement Home
Chinese humanoid robots are deploying in three main segments, revealing pragmatic commercial strategies. Logistics represents 60% of applications, with order preparation tasks, package sorting, and inventory. Amazon has been testing AgiBot units in its Guangdong warehouses for six months, with productivity gains of 23% on “picking” operations compared to traditional robotic systems.
Manufacturing assembly constitutes the second market, particularly in electronics and automobiles. Unlike fixed robotic arms, humanoid robots navigate environments designed for humans without modification of production lines. BYD has been experimenting since September 2025 with Unitree robots on its battery assembly lines, with availability of 94% versus 87% for human teams working in 3x8 shifts.
The third segment, more prospective, concerns assistance to the elderly. Japan, facing a deficit of 2.3 million caregivers by 2040, is testing pilot programs with Chinese humanoid robots in 150 facilities. Initial evaluations show acceptance of 68% by residents, higher than the 52% recorded with traditional service robots.
These applications reveal the competitive advantage of humanoid robots: their versatility. Where a robotic arm or autonomous vehicle confines itself to a specific task, the bipedal robot adapts to varied environments with the same basic equipment. This flexibility justifies a higher unit price but reduces integration and maintenance costs.
Europe Reacts, America Hesitates
Facing this Chinese offensive, Europe is initiating an industrial response. This European mobilization draws on lessons learned in other technological sectors. Europe is betting on open data to compete with AI giants, but it also relies on its comparative advantages: excellence in precision mechanics, traditional industrial robotics, and anticipated AI regulation.
The United States is adopting a more defensive posture. The Biden administration classifies humanoid robots among sensitive technologies, subject to export restrictions to China. Simultaneously, the Pentagon finances military research programs on bipedal robots, with a budget of $1.2 billion over five years. This approach privileges defense applications at the expense of the civilian market.
This strategic divergence recalls American mistakes in photovoltaics and batteries. While Washington secures its laboratories, Beijing industrializes and conquers global markets. American AI enriches capital before labor, but it struggles to transform innovation into accessible mass production.
When Automation Becomes Geopolitics
Chinese industrialization of humanoid robots transforms a technological issue into a geopolitical question. Countries dependent on Chinese robotic imports risk strategic vulnerability comparable to that observed in rare earths or semiconductors. This dependence proves all the more sensitive as humanoid robots integrate artificial intelligence capabilities and data processing.
The global humanoid robot market is experiencing rapid expansion and could represent a significant share of global robotic economy in the years to come. This growth is explained by the convergence of three factors: falling production costs, improved AI performance, and growing demand related to demographic aging.
China already controls 70% of this nascent production, reproducing its dominance in other manufacturing sectors. This position grants it an advantage in defining technical standards, forming industrial ecosystems, and capturing added value. Western manufacturers find themselves forced to catch up with a gap that widens every quarter.
The stakes extend beyond industrial economics. Humanoid robots collect and process sensitive data on work environments, industrial processes, and human behavior. Their control by foreign actors raises questions of digital sovereignty that few governments anticipate correctly.
This race to robot industrialization illustrates a new form of geoeconomic competition, where Chinese manufacturing advantage combines with technological innovation to create lasting dominant positions. Europe and the United States are discovering, once more, that scientific excellence is insufficient against a coherent and well-funded industrial strategy.