The Factory Is Ready. The Buyers Aren't.
The $200 million bet that more robots mean more jobs rests on a single, unproven premise: that American manufacturing's competitiveness problem is a cost problem, not a demand problem — and that cheaper production alone can summon back the buyers who left decades ago.
TL;DR
- A Beijing smartphone factory has already produced 300 humanoid robots and is racing toward 10,000 by year-end — and 500,000 annually by 2030.
- The majority of orders so far are one or two units. Bain's Beijing partner is watching for repeat purchases. He hasn't seen them yet.
- China shipped 85% of the world's humanoid robots last year. The US leads on the "brains." Neither country has solved the demand problem.
- Chinese regulators are warning of a bubble even as the state funds the flag-planting. The tension is not a bug. It is the strategy.
- The question is not whether the robots work. It is whether anyone needs them to.
What Happened
Evelyn Cheng of CNBC reported from Beijing this week on a story that crystallises the central tension in humanoid robotics. A 20-year-old smartphone and electronics manufacturer called Lingyi iTech has converted a factory to produce humanoid robots. Weeks after opening in late April, it had already produced 300 units. The target is 10,000 this year and 500,000 annually by 2030.
At that scale, the price of a humanoid — currently around $30,000 — could halve, according to Lingyi's Vice President Philip Yang.
The factory is real. The production numbers are directionally credible. EngineAI, another Shenzhen-based startup, launched its own intelligent manufacturing base in late May, claiming a new humanoid robot rolls off the line every 15 minutes. Figure, the US-based competitor, announced a 24-fold production increase — from one unit per day to one per hour — in under 120 days. The supply side is accelerating.
The demand side is not.
"The majority of humanoid orders so far are just one or two robots," Bain's Beijing-based partner Xin Cheng told CNBC. He is watching for repeat orders. The implication is clear: he has not seen them.
What It Actually Means
This is not a story about whether the technology works. It is a story about whether anyone has figured out what the technology is for.
The numbers are worth sitting with. China had more than 140 humanoid robot manufacturers and more than 330 models in 2025, according to the Ministry of Industry and Information Technology. Of the more than 13,000 humanoid robots shipped globally last year, AGIBOT and Unitree — two Chinese leaders — each shipped over 5,000. US rivals Figure AI and Tesla each shipped a few hundred or fewer, according to Omdia. Chinese humanoid robots accounted for roughly 85% of the global total, per Barclays.
And yet: the majority of orders are one or two units. The Chinese government itself publicly warned last year about the risk of a bubble, citing "the lagging state of commercialization and applications."
The tension is not a contradiction. It is the strategy.
Omdia's chief analyst Lian Jye Su told CNBC: "The Chinese humanoid companies carry the mandate from the government to craft this image of China as a strong industrial economy that has deep tech capability and deep tech expertise. They receive government support just for carrying that flag or carrying that mandate."
This is the analytical key. The humanoid industry in China is not being built primarily to meet market demand. It is being built to plant a flag. The demand, the argument goes, will follow. Whether it does is the question that determines whether this is an industrial revolution or the world's most expensive nation-branding exercise.
The evidence for demand is thin but not non-existent. Morgan Stanley reports that more than 2 billion yuan ($295 million) worth of orders in 2025 came from state-owned enterprises — power plants, data centres, entertainment. Matrix Robotics, a Shanghai-based startup, claims roughly 1,000 orders from coffee chains and hotels. Beijing opened a government showroom in August filled with robots for sale, including a soccer-playing humanoid for 199,000 yuan ($29,400) and a package-sorting model for 349,999 yuan. Cumulative orders surpassed 30 million yuan as of late May.
But these are pilot programmes, government procurement, and novelty purchases. They are not the kind of repeat commercial orders that build an industry. As Chibo Tang of Gobi Partners put it to the AP: "Without the demand and without that scale from the market, these companies are not able to really go into mass production."
Hype Deconstruction
"500,000 units annually by 2030" is a target, not a forecast. Lingyi iTech is a contract manufacturer. It builds what customers order. The 500,000 figure is capacity, not demand. Conflating the two is the single most common error in humanoid robotics coverage.
The "85% global market share" statistic is misleading. It measures shipments, not revenue, not deployed-and-working units, and certainly not profitable commercial operations. Many of those shipments went to research labs, state-owned enterprises fulfilling government procurement targets, and companies buying one or two units for evaluation. That is not a market. It is a subsidised exploration phase.
"Prices will halve at scale" assumes the scale arrives. The economics of mass production are well understood. Unit costs fall as volume rises. But the causal arrow points the other way: volume rises when there is demand. The price reduction is a consequence of demand, not a substitute for it.
The bubble warning is real and sourced. China's Ministry of Industry and Information Technology — not a foreign sceptic, not a short-seller — warned about a bubble in 2025. When the state that is funding the flag-planting also warns about the bubble, the tension is worth taking seriously.
Stakeholder Landscape
Who benefits directly: Chinese humanoid manufacturers with government backing — AGIBOT, Unitree, EngineAI, Matrix Robotics. They receive state support, access to government procurement contracts, and a manufacturing ecosystem that can produce hardware at costs US and European competitors cannot match. Unitree reported 1.7 billion yuan ($250 million) in revenue last year with a profit of over 278 million yuan ($41 million). It is already planning a Shanghai IPO.
Who faces the demand problem most acutely: The same companies. Government procurement and one-off research purchases do not build sustainable businesses. The Bain observation — that most orders are one or two units — applies to the entire sector. The companies that survive will be those that find commercial repeat buyers, not those that win the most government contracts.
Who should pay attention: Industrial automation integrators and enterprise buyers. The technology is not ready for widespread deployment, but the direction of travel is clear. Humanoid robots will become cheaper, more capable, and more available. The firms that begin evaluating them now — in structured environments like warehouses and factories — will have an advantage when the technology matures.
Who is largely unaffected (for now): Workers in roles requiring high dexterity and unstructured problem-solving. Most humanoid robots are still "performative rather than functional," as Samm Sacks of the New America think tank told the AP. The gap between a robot that can dance and a robot that can replace a warehouse worker is measured in years, not months.
Who benefits from the noise: The Chinese state. The flag-planting strategy is working. The world is talking about China's humanoid dominance. Whether the commercial reality catches up to the narrative is a separate question — and one that will be answered over the next three to five years.
Cross-Layer Implications
Geopolitics: The humanoid robotics race is not primarily commercial. It is strategic. China is building manufacturing capacity and hardware supply chains. The US is building AI "brains" — Nvidia's Isaac GR00T platform, Figure AI's embodied intelligence, Tesla's Optimus. The country that integrates both layers first wins the physical AI era. Neither has done it yet.
Software ecosystem: Nvidia is launching a robotics system for developers later this year in collaboration with Unitree. This is Nvidia's attempt to extend its AI dominance into the physical world — the same playbook that worked for CUDA in the data centre. Bain's Cheng noted that for Chinese companies, "software is just a layer of humanoid development that adds to systems for manufacturing, low-priced parts and gathering training data." The software layer is where value will accrue over time.
Data: Government-backed centres in Beijing and other parts of China are training robots with people guiding them through daily tasks. One state-backed data collection centre told CNBC it is working to provide robot training data to partners in South Korea and Germany. The country that accumulates the largest corpus of real-world robot training data will have an advantage that is difficult to replicate. This is a quieter race than the manufacturing one, and it may matter more.
Talent: The automation engineer shortage is global and acute. China's rapid scaling of humanoid production will intensify demand for engineers who can integrate, maintain, and programme these systems. The talent bottleneck may constrain deployment even if the demand materialises.
Regulatory: China has introduced a mandatory digital registration system for humanoid robots — unique identification codes enforced by the Ministry of Industry and Information Technology. This is a regulatory infrastructure play. It positions the state to track, manage, and control humanoid deployment at scale. No equivalent framework exists in the US or Europe.
What This Means for You
If you're an investor in robotics: The demand question is the only question that matters. Production capacity, unit economics, and technical milestones are secondary. Watch for repeat commercial orders — not government procurement, not research purchases, not one-off pilots. The first company that demonstrates a customer buying its third, fourth, and fifth humanoid robot for commercial deployment will have crossed the chasm. No one has done it yet.
If you're a manufacturer or warehouse operator: Do not budget for humanoid deployment in 2026 or 2027. But start building the organisational capability to evaluate these systems. Visit a showroom. Talk to integrators. Understand the difference between a humanoid robot that can perform a task in a demo and one that can perform it reliably for eight hours in your facility. The gap is large, but it is closing.
If you're a policymaker: The humanoid robotics race is not a market. It is an industrial policy competition with market characteristics. The US is currently losing on manufacturing scale and winning on AI software. Neither advantage is permanent. The policy question is whether to compete on manufacturing, on software, on data infrastructure, or on all three — and whether the resources being committed match the strategic stakes.
If you're a worker: Nothing changes today. But the skills that will be valuable in a humanoid-rich workplace are design, maintenance, technical operation, and — counterintuitively — the interpersonal and creative skills that robots cannot replicate. The robots are coming. They are not here yet. Use the gap.
Uncertainty Ledger
- Repeat orders: The single most important metric for the industry's commercial viability. No company has publicly demonstrated a pattern of repeat commercial orders. Until one does, the industry is in an exploration phase, not a growth phase.
- Unitree IPO: Unitree's planned Shanghai IPO in the coming weeks will be a market test of how public investors value humanoid robotics companies. The pricing and performance will signal whether the financial market believes the demand will materialise.
- Government procurement dependency: The proportion of humanoid revenue that comes from state-owned enterprises versus commercial buyers is not publicly disclosed. If the majority is government procurement, the industry is vulnerable to policy shifts.
- US competitive response: The US currently has no equivalent to China's state-backed humanoid manufacturing ecosystem. Whether the US responds with industrial policy, defence procurement, or relies on private-sector competition will shape the global competitive landscape.
- Bubble risk: The Chinese government's own bubble warning is the most credible signal that the supply of humanoid robots has outpaced demand. Whether the bubble deflates gradually or pops depends on whether commercial demand materialises before government patience runs out.
Bottom Line
China can build humanoid robots faster than anyone else. A smartphone factory in Beijing is already producing them, and the supply chain is scaling toward hundreds of thousands of units per year. The problem is that almost nobody is buying more than one or two. The industry is being propelled by government mandate and strategic ambition, not by market demand. That does not mean the demand will never arrive — the technology is improving, costs are falling, and use cases in logistics and manufacturing are becoming legible. But until someone demonstrates a pattern of repeat commercial orders, the humanoid robotics industry is a supply-side story in search of a demand-side reality. The factory is ready. The buyers aren't.
Sources: CNBC (Evelyn Cheng, "The China Connection," 8 Jun 2026) — Tier 1; AP News (Chan Ho-him, 6 Jun 2026) — Tier 1; Manufacturing.net / AP (8 Jun 2026) — Tier 2; Robotics & Automation News (Figure ramp, 27 May 2026) — Tier 2; EngineAI press release / PR Newswire (29 May 2026) — Tier 3; Morgan Stanley research (cited in AP) — Tier 2; Omdia (Lian Jye Su, cited in CNBC) — Tier 2; Bain & Company (Xin Cheng, cited in CNBC) — Tier 2; Barclays research (cited in AP) — Tier 2; Mercator Institute for China Studies (cited in AP) — Tier 2
