A humanoid factory in Beijing opened in late April. By early June it had already produced 300 robots for paying clients. The same factory is targeting 10,000 units by the end of this year and half a million per year by 2030. The supply side of China's humanoid robot industry is moving fast. The demand side is a different question entirely.
The Factory That Went From Zero to 300 Robots in Weeks
Lingyi iTech is not a robotics startup. It is a 20-year-old electronics manufacturer that supplies precision components to Apple, Samsung, and Xiaomi. Metal casings for iPhones. Structural parts for flagship Android devices. In early 2026, the company announced it was going all-in on humanoid robots, and opened a dedicated super factory in Beijing's Yizhuang district in late April.
Weeks after opening, the facility had already delivered 300 robots to clients, according to CNBC reporting from Beijing. Its current annual capacity target is 10,000 units. By 2030, the company's Vice President Philip Yang says the plan is 500,000 units per year. That scale, Yang told CNBC, could cut the price of a humanoid roughly in half from where it sits today.
Today a humanoid in China runs about $30,000. At half a million units annually, Yang expects that to drop by at least half, possibly more. The logic is the same one that applied to smartphones: build enough volume and the per-unit cost collapses. Lingyi iTech is betting its next decade on whether that logic holds for bipedal machines.
Lingyi iTech's clients include several of China's leading humanoid developers, among them the Beijing Innovation Center of Humanoid Robotics, AgiBot, and MagicLab. The company's role is manufacturing and assembly for startups that are strong on AI and hardware design but lack the production infrastructure to scale. There are more than 100 humanoid startups in China right now. Almost none of them have their own mass-production facility.
Who Is Actually Buying Right Now
The honest answer is: not many people, in small quantities.
Xin Cheng, a Beijing-based partner at Bain, told CNBC that the majority of humanoid orders so far are just one or two robots per buyer. Not fleets. Not enterprise contracts. One robot in a coffee shop, one robot at a hotel reception, one robot at a government showcase. He is watching whether any of these buyers come back for repeat orders. So far, that data does not exist in volume.
The Beijing municipal government has tried to actively create a buyer pool. The city opened a showroom in August 2025 stocked with humanoids from multiple companies. One of the units is the Booster T1, a soccer-playing humanoid priced at 199,000 yuan (about $29,400). Another is the Galaxea R1 Pro at 349,999 yuan, capable of sorting packages. By late May 2026, cumulative orders at the showroom had passed 30 million yuan, according to a representative for the store. That is roughly $4.4 million across many months of government-backed sales effort.
For context on where these robots are actually being tested in industrial settings, China Post's humanoid robot trial in Guangzhou is one of the more documented real-world deployments available.
Factory and logistics use cases represent the most active buyer segment right now. Robot makers report orders from warehouse operators, postal hubs, and manufacturing floors for humanoids that can sort parcels, handle repetitive assembly tasks, and work in spaces that are structured and predictable. Unitree, one of China's best-capitalized humanoid companies, reported approximately $250 million in revenue and a profit of around $41 million for 2025, according to Reuters reporting. That is the exception, not the rule, in a field with over 100 competitors.
Morgan Stanley doubled its 2026 forecast for Chinese humanoid sales to 28,000 units. Against a backdrop of 100-plus active startups, each requiring capital to survive, 28,000 units is a thin market.
The Household Dream vs. the Industrial Reality
The pitch for humanoid robots has always had two layers. The near-term layer is factories: replace dangerous, repetitive labor in controlled environments where the robot does not need to navigate a messy kitchen or understand an ambiguous instruction. The long-term layer is the home: a general-purpose assistant that can cook, clean, and handle unpredictable domestic tasks.
China's more than 100 humanoid startups are largely targeting the household layer. They show robots dancing, serving coffee, and moving through demo apartments with scripted precision. The demos are good. The gap between a rehearsed demo environment and a real home is the part that does not make it into the press release.
Samm Sacks, a senior fellow at the New America think tank focused on Chinese technology, told Fortune that most humanoid robots remain performative rather than functional, falling short when placed in messy, unpredictable environments. Chibo Tang of Gobi Partners, a VC firm that invests in robotics startups, stated plainly that use cases are "still very limited." Without sufficient demand, Tang added, companies cannot reach the production volumes they need to become viable.
That is the core tension. The factory thesis is real but narrow: China's aging population and rising labor costs create genuine demand for automation in structured settings. The household thesis is the larger addressable market but requires a capability leap that has not yet happened. "The humanoid is not a single product," Philip Yang told CNBC. "It is an ecosystem." Building an ecosystem takes longer than building a factory.
The hardware layer is developing in parallel. Unitree's role in the humanoid ecosystem includes being one of the first platforms selected by NVIDIA for its Isaac GR00T research stack, which speaks to how seriously the underlying AI infrastructure is being built out. But software capable of navigating a real home environment at commercial scale does not yet exist at a price point that makes household deployment practical.
What Happens When Supply Outpaces Demand
Morgan Stanley's analyst Sheng Zhong called 2026 a critical year for humanoid commercialization and warned explicitly that a shakeout is coming. His reasoning: production is likely to run materially ahead of sales, because major players are building robots for internal training and verification rather than shipping them to paying customers. The unit counts look good in headlines. The revenue often is not behind them.
Over 100 startups competing for 28,000 units of projected demand is an equation that does not sustain most of them. The companies with manufacturing infrastructure, like Lingyi iTech on the production side, or with proprietary AI stacks that work in real environments, are better positioned. Startups that are primarily demo companies without either advantage are at risk.
The pattern is not unfamiliar. China's EV industry went through the same compression, with hundreds of entrants narrowing to a handful of survivors once real consumer demand replaced government-subsidized pilots. The humanoid sector has the same structure: massive state support, low barriers to entry for new startups, and a demand curve that has not yet arrived at the scale the supply side assumes.
NVIDIA and LG's humanoid manufacturing push reflects confidence from major platform players that the infrastructure layer will eventually be there. The question is timing, and which of the 100-plus current contenders are still operating when the demand arrives.
My Take
The Lingyi iTech story is the most clarifying data point in this whole space right now. A company that spent 20 years perfecting metal casings for iPhones is now producing humanoid robots in a Beijing super factory. It already has clients. It is already shipping. That is not a startup pitch or a government projection. That is a real manufacturer making a real bet with real production infrastructure.
The demand side is the problem nobody wants to say out loud. Government showrooms, coffee shop demos, and 1-to-2-unit trial orders from curious buyers are not a market. They are a placeholder while everyone waits for the household use case to become real. It might. China's aging population and shrinking labor pool create structural pressure that the EV sector never had quite as sharply. But the gap between "China can build 500,000 humanoids a year by 2030" and "someone will buy 500,000 humanoids a year by 2030" is the entire question. Nobody has answered it yet.
- Lingyi iTech, an Apple/Samsung component supplier, opened a humanoid super factory in Beijing in late April 2026 and produced 300 robots within weeks of opening.
- Current humanoid price in China is approximately $30,000. The company's VP projects that scaling to 500,000 units/year by 2030 could halve the price. This is an unconfirmed projection.
- Most current orders are 1 to 2 units per buyer, per Bain's Beijing partner. Repeat orders are what the market is actually watching for.
- Beijing's government showroom for humanoids had accumulated roughly $4.4 million in orders as of late May 2026, across many months of promoted sales.
- Morgan Stanley forecasts 28,000 humanoid units sold in China for 2026. Over 100 startups are competing for that demand.
- A Morgan Stanley analyst has publicly flagged that a market shakeout is coming, with production expected to materially exceed actual sales this year.
FAQ
Who is currently buying humanoid robots in China?
The primary buyers are factories, logistics operators, and government-backed showrooms. A Beijing showroom opened in August 2025 has facilitated purchases of humanoids from several companies for tasks like package sorting and interactive demonstrations. Most private orders are small, typically one or two units, from buyers in hospitality, retail, and light manufacturing testing the technology for the first time.
How much does a humanoid robot cost in 2026?
Prices vary by capability and manufacturer. The Beijing government showroom carries models ranging from the Booster T1 at approximately $29,400 to the Galaxea R1 Pro at around $50,600. Some premium full-sized humanoids from companies like Matrix Robotics are priced near $99,000. Lingyi iTech's VP stated that scaling production to 500,000 units per year could bring the general price point of around $30,000 down by half or more, though that projection is tied to 2030 production targets that are not yet confirmed.
Are humanoid robots ready for home use?
Not yet, according to independent experts. Researchers and venture investors tracking the space note that most current humanoids perform reliably only in structured, controlled environments. Messy, unpredictable settings like actual homes remain a challenge the technology has not solved commercially. The household market is where most Chinese startups say they are headed, but no company has demonstrated reliable autonomous operation in real home conditions at commercial scale.
Will China's humanoid robot industry face a shakeout?
Morgan Stanley's analyst Sheng Zhong has said publicly that a shakeout is expected in 2026 and beyond, with more than 100 startups competing for a market projected at 28,000 units in China this year. He noted that production is likely to run ahead of actual sales because manufacturers are building robots for internal training rather than shipping to paying customers. Companies with real manufacturing scale or proven AI performance in commercial settings are better positioned to survive consolidation.
Conclusion
China has solved the production side of the humanoid equation faster than most analysts expected. A smartphone component manufacturer is now running a humanoid super factory that produced 300 units in its first weeks of operation. The price is around $30,000 and falling. The supply chain expertise is real.
What China has not solved is the same thing every hardware market eventually runs into: a buyer base large enough and consistent enough to support the number of companies competing for it. One-to-two unit trial orders and government-backed showrooms are not the same thing as sustained commercial demand. The shakeout Morgan Stanley flagged is a structural prediction, not a pessimistic one. The companies that survive it will likely define the humanoid market for the decade after.
Whether the buyer arrives before the funding runs out is the question 100-plus Chinese humanoid startups are quietly trying to answer right now.
Sources: CNBC "The China Connection" newsletter, Evelyn Cheng, June 8, 2026 (cnbc.com); Fortune, June 6, 2026; Gasgoo, April 2026; Morgan Stanley analyst note via Reuters.
0 Comments