Morgan Stanley: Humanoid Robot Market Could Reach $7.5 Trillion by 2050
nashnova research
Morgan Stanley projects the global humanoid-robot market at $7.5 trillion by 2050 with nearly 1 billion units deployed, yet its latest China survey shows commercialization timelines slipping and satisfaction falling — scaling still hinges on clearing the twin hurdles of price and reliability.
A $7.5 trillion vision — so why is adoption slipping?
Morgan Stanley maps a gradual ramp: ~28.1 million units by 2036 → ~138.5 million by 2040 → ~430 million by 2044 → ~1 billion by 2050. Early volumes stay small.
Its latest China survey: 64% of respondents expect to adopt humanoid robots within three years, but the average expected adoption date has shifted back roughly one year. Satisfaction dropped from 23% to just 18%.
This means → The long-term thesis is intact, but near-term experience is pushing potential buyers away — the product isn't good enough yet, so companies won't commit.
72% of respondents say the unit price must fall below RMB 100,000 (~$15,000) to make economic sense, up from 53% last year. In plain terms = the price bar keeps dropping, which signals growing skepticism about current value for money.
Why did China's robot stocks suddenly rally?
Since October 15, Unitree (宇树科技) surged 62% and Shuanghuan Transmission rose 31%, while the MSCI China Index was roughly flat.
Morgan Stanley cites three drivers: ① light positioning beforehand left room to add; ② advances in frontier large language models may accelerate embodied intelligence — the push to give AI a physical body that can manipulate the real world; ③ Unitree founder Wang Xingxing now sees an intelligence inflection point in 2027–2028, more optimistic than before.
Unitree's Q3 2026 revenue hit RMB 665 million, up 70% year-on-year, but net profit fell 17% to RMB 113 million as R&D spending rose to 19% of revenue. This means → The company is trading margin for R&D, betting it can build a technology moat before the inflection arrives.
Who gets paid first — component makers or full-machine builders?
Morgan Stanley's call: component suppliers benefit ahead of full-machine OEMs. The logic is simple — OEMs need extensive pilots to gather data, and more pilots mean component orders materialize first.
Leader Harmonics posted Q3 revenue up 56% year-on-year to RMB 242 million; net profit rose 29% to RMB 52 million.
Shuanghuan Transmission disclosed that robot-reducer revenue — reducers convert a motor's high-speed spin into the slow, high-torque movement a joint needs — topped RMB 100 million in the first nine months of 2026. Full-year target: RMB 150–200 million; 2027 target: RMB 400–500 million.
Sanhua Intelligent Controls is expanding from actuators into dexterous-hand micro lead screws, thermal management, and sensors, aiming to cover 50%–60% of a humanoid robot's hardware value. This means → Component makers are seizing the "picks-and-shovels" position — whichever OEM wins, the parts still flow through them.
What are the global tech giants doing?
Meta and Fourier Intelligence have built a humanoid prototype code-named "Max" and are developing a more capable second generation.
Apple's new "HALO" research generates synthetic training data from 1,000 simulated home scenes; in 360 single-task tests across six previously unmapped environments, it achieved a 91% success rate.
Xiaomi demonstrated a humanoid completing a full eight-hour factory shift, sorting center-console side panels on its EV production line. Tesla confirmed its Fremont plant has started an Optimus production line and produced its first production-quality robots, though external delivery timelines and commercial revenue remain unconfirmed.
Schaeffler opened a ~7,500 m² humanoid-robot lighthouse factory in Taicang; global new orders rose to roughly €400 million. This reflects a shift from "bullish talk" to real capital commitments on production lines and data.
What policy cards are China and the US playing?
China issued 19 measures explicitly promoting embodied intelligence. Shanghai plans to deploy 8,000 humanoid robots in factories by 2028. MIIT and the Ministry of Finance are seeking comments on government-procurement standards — the draft requires that domestically produced components account for at least 70% of cost, with a three-year transition period.
China's foreign-investment security review body blocked SoftBank's acquisition of ABB's robotics business in China, part of a $5.375 billion global deal. The Financial Times reported regulators are concerned about robot data being transferred overseas for AI training.
In the US, President Trump issued a proclamation on October 28 calling for negotiations on robot and industrial-machinery imports. No new tariffs were imposed immediately; officials must report back within 180 days.
This means → Both countries are treating robotics as a strategic asset — China demands "domestic content + data stays onshore," while the US is keeping a tariff slot open.
What is the real bottleneck for scaling?
Morgan Stanley's summary: the sector remains catalyst-driven in the near term — new product launches, policy tailwinds, tech breakthroughs — rather than pulled by fundamentals.
The market needs three things before real volume arrives: quantifiable productivity gains, economic-viability proof, and repeat orders.
Before the capability inflection point, data collection is becoming the strategic priority — robot companies and foundation-model developers in both China and the US are ramping up data procurement.
In plain terms = the virtuous cycle is "product works → companies buy → data grows → product improves." That flywheel hasn't truly started spinning yet — it's stuck at step one: the product isn't good enough and the price isn't low enough.
市场有风险,内容仅供研究参考,不构成投资建议。
