The humanoid robot industry chain — from a market cognition map to upstream components, OEMs, capital and applications.
Star-map lights up once the page settles…
Drag to orbit the star-map · scroll/pinch to zoom · click a node for facts · legend focuses a chapter
Set the coordinates before diving into the chain — market pulse / capital markets / shipment landscape / development stages / structural calls / segment map
2026 monthly financing vs. annual base
Jan–Jul 2026 humanoid financing reached ~¥164.1B (GG basis), far above the ¥39.8B full-year 2025 total; H1 alone was ¥104.1B (Yuanda), nearly double 2025, with June's ¥31.5B a record. August's monthly tally is pending; disclosed deals already include XPENG Robotics' $900M+ debut round (post-money $6.3B+) and LingChu's $100M+ round.
By 2025, China had 868 active humanoid-robot-related companies, including 140+ whole-machine OEMs and 330+ released products (MIIT, Qichacha; compiled by Analysys).
Listed caps (Aug 28, 2026 close) & latest primary valuations
Listed 2026-08-19, the first humanoid IPO; priced at ¥150.80, opened +629%, first-day close cap ~¥341.8B; latest close ¥585.00 (-4.88%).
Listed Dec 2023; latest close HK$83.50 (+0.66%); ~¥1.4B orders in 2025, targeting 10k-unit capacity in 2026.
Pursuing HK listing (Aug 2026); 15,000th Genie G2 rolled off in June.
Debut round $900M+ on 2026-08-24, a record for China's embodied AI.
Prior rounds led by Alibaba and NIO (2026-08-21).
Per GG July report; humanoid maker.
¥1B Series B in Feb 2026; ~¥3B raised in total.
~¥2B across two rounds in Feb 2026; ¥10B+ just 26 months after founding.
Tactile sensors; crossed ¥10B in July 2026.
Home robot / dexterous hand; new unicorns as of July 2026.
Capital-market channels are now open: Unitree became the first listed humanoid maker, AgiBot is pursuing a HK listing, RobotEra is reportedly prepping an IPO; Central Huijin holds stakes in 8 core humanoid-chain firms. Primary-market valuations concentrate on shipping leaders — six Chinese firms entered the ¥10B club in early 2026.
Omdia 2025 actuals + 2026 output forecasts
Leju, EngineAI and Fourier rank 4th–6th; Chinese vendors lead overall, shipping 84.7% of the global total.
From hundreds to 100k: global shipments grew from a few hundred (2023) to 3,000 (2024) to ~13,000 (2025); China's 2026 output is expected to pass 100k. Foshan runs China's first 10k-unit annual line (one robot every 30 minutes); AgiBot went from its 10,000th to 15,000th unit in just over two months.
Five-stage evolution of China's embodied-AI industry (Analysys)
Trend judgments from research reports and latest market moves
Global shipments: a few hundred (2023) → 3,000 (2024) → ~13,000 (2025); China's 2026 output may pass 100k. Foshan's 10k-unit line ships one robot every 30 minutes; Unitree's humanoid price fell from ¥593k (2023) to ¥29.9k — the cost curve is clearing the path to scale.
Unitree's first-day cap hit ~¥341.8B as the first listed humanoid maker; AgiBot is pursuing a HK listing, RobotEra is reportedly prepping one, and Central Huijin holds 8 chain names. Capital now follows firms that actually ship — fundraising heat is coupling with delivery.
Chinese vendors shipped 84.7% of the world's humanoids in 2025; Q1 2026 exports reached ¥11.3B (+210%) across 148 countries; Shanghai ports handled ¥8.6B of robot exports in the first five months — 42.5% of China's total.
MIIT and SASAC launched a 10k-unit live-deployment action targeting 100+ high-value manufacturing and livelihood scenarios, while NDRC warns against a blind rush. The commercial battleground has shifted from shows to 3C inspection, logistics sorting and auto manufacturing.
ZoSi Research directory basis: companies covered per segment, tap to jump
Sources: market caps from Tencent/Eastmoney (Aug 28, 2026 close); shipments from Omdia 'General Embodied Robot Market Radar' (Jan 2026); funding from GG Humanoid monthly reports (Jan–Jul 2026) and Yuanda H1-2026 statistics; output and forecasts from MIIT (WAIC 2026), Morgan Stanley, GGII and Deutsche Bank; company counts and 2025 baselines from Analysys and ZoSi Research '2026 Embodied-AI Robot Industry Chain Panorama'. Public information, for reference only.
10 core segments mapped onto the robot body · tap a number to jump

Tap a numbered hotspot to jump to its segment
16 key segments · highest cost share & tech moats
Head-mounted LiDAR adds mapping and navigation redundancy for mobile scenarios.
Stereo, structured-light and ToF depth cameras are the robot's eyes.
Harmonic/RV reducers set joint precision and life; the biggest rotary-joint cost.
High-precision encoders close the loop on joint-angle control.
High torque-density frameless motors directly drive rotary joints.
Coreless motors are small and fast, the mainstream finger-drive choice.
Wrist/ankle 6-axis F/T sensors give contact-force awareness for force control.
Fingertip tactile arrays and e-skin detect slip for dexterous manipulation.
Multi-DoF dexterous hands are the end-effector of general manipulation.
High C-rate packs set runtime and peak power; humanoids carry 1-2 kWh.
Real-time controllers and servo drives handle balance and tracking.
High-performance MEMS IMUs feed attitude data to balance algorithms.
Cross-roller and thin-section bearings carry joint and screw loads.
Rotary/linear actuators integrate motor, reducer, screw and sensing.
Planetary roller screws convert rotation into linear force.
PEEK, carbon fiber and die-casting cut weight, boosting runtime.
VLA, world models and sim data form the embodied-AI software stack.
115 companies · Beijing / Shanghai / Shenzhen / Hangzhou + 20 cities
Founded by Dr. Wang He, assistant professor at Peking Un
Founded by Jiang Zheyuan, a 1998-born Tsinghua PhD stude
Founded by Professor Chen Jianyu, doctoral supervisor at
Jointly established in November 2023 by UBTECH, Jingchen
Founded by researchers from the German Aerospace Center,
Founded by Liu Fang (SJTU CS master's, among Google's ea
Led by CEO Zhang Yufeng, former Horizon Robotics VP and
Founded by Zhang Rui, who previously worked at Beihang U
Founded by Cheng Hao, Tsinghua bachelor's and master's g
Co-founded by CEO Chen Yan (Southeast University CS grad
Founded by Dr. Zhang Zhengtao, researcher and PhD adviso
Co-founded by Dr. Lian Wenzhao (CAS researcher, former F
A wholly-owned subsidiary of listed company XinXing Equi
Co-founded by Wang Kun, former founding member of JD's u
A humanoid robotics company founded and led by CEO Tang
Founded by Tsinghua University's top motion control team
Founded by Dr. Liu Nianfeng (CAS Institute of Automation
Founded by Gao Jiyang, Tsinghua Electronics Department u
JD.com's smart robotics division, established in March 2
Founded by Li Xiang, founder of Autohome. In May 2026, L
Founded by Yang Hongbing (Tsinghua graduate, former Tenc
Founded by Zhang Haixing, former founding leader of Tesl
Founded by Ren Xiaoyu and co-founder Mao Shuhan (both Ts
Founded by Zheng Suibing, RealMan's core team comprises
Co-founded in December 2024 (operations began March 2025
Founded in Beijing in Sep 2015 by Xu Huayang, Tsinghua m
Beijing Tianxia Xianzhichuang Robot Technology Co., Ltd.
Founded by Dong Xiaojian, with core R&D from leading sem
Operated independently as Beijing Xiaomi Robotics Techno
Co-founded by Academician Ren Luquan (Jilin University)
Incubated by the Chinese Academy of Sciences, with CEO S
Founded by Gu Jie, a Shanghai Jiao Tong University mecha
Founded in Shanghai in Nov 2020 by Wang Dezhao, former t
Founded by Bill Huang, former CTO of UTStarcom and head
Founded by Cambridge PhD Niu Tengfei, formerly an autono
Founded by Professor Li Qingdu, executive dean of the Ma
Founded by Zhang Yi, former founder of Zhangmen Educatio
Founded by CEO Qi Jianwei, the first Chinese employee at
A state-backed R&D institution co-founded by leading ent
China's leading collaborative robot maker, JAKA ranked f
The world's No.1 shipper of commercial service robots (I
Founded by Dr. Liang Conghui, formerly a T12 expert rese
Co-founded by Yang Hua, founder of Puremi Technology, an
Founded by Dr. Wang Lei, a Tsinghua electronic engineeri
Founded by Huang Yi (born 2004), who entered Harbin Inst
Founded by CEO Shi Yunlei, a student of German Academy o
One of China's largest automotive groups, SAIC Motor beg
Founded in May 2014 with dual HQs in Shanghai and Shenya
Originally an advanced materials company, Swancor entere
Founded by Dr. Chen Yilun, former Huawei Autonomous Driv
TIS Robot focuses on high-performance lightweight joint
An AI-robotics education company whose founding team has
Founded by Dr. Guo Yandong, formerly a researcher at Mic
Founded by Dr. Jia Kui, tenured professor at CUHK-Shenzh
Founded by Zhao Tongyang, a humanoid-robotics veteran of
Founded by Zhang Wei, tenured professor at SUSTech and f
Founded by Waseda University PhD Xu Jincheng, with found
Founded by Zhou Jian in 2012, who serves as Chairman and
Led by CEO Wang Qian (Tsinghua, USC robotics PhD), CTO W
Founded by Zhang Xinliang, a Harbin University of Scienc
Founded by Lai Jie, first employee of Tencent Robotics X
Founded by Chen Siwei, a National University of Singapor
The world's largest NEV maker, founded by Wang Chuanfu.
Founded in July 2024, Cyborg focuses on all-terrain, all
Founded by Shen Jian, a veteran of 20+ years in top IT e
Shenzhen bionic-intelligence and humanoid company founde
Co-founded by Liu Peichao (Shandong University mechanica
Co-founded in August 2016 by Wu Qilin and peers, with fo
The global smart-device giant spun off from Huawei in 20
Co-founded by CEO Yu Jie (YU JIE), former general manage
Founded by a Harbin Institute of Technology PhD team — C
Founded by Xiao Kai, who was inspired by Musk's humanoid
Founded by Yu Chao, former head of Dreame's humanoid bus
A leading global precision functional-parts maker founde
Founded in Shenzhen in 2025 by CTO Huang Qingqiu (Tsingh
Founded by Li Zhichen, who graduated from Harbin Institu
Founded by Zhang Tao, an HKUST mechanical engineering gr
Founded in Shenzhen in Nov 2018 by Dr. Geng Kaiping. Com
A special-industry embodied-AI company founded in Shenzh
Co-founded by Zhu Qiuguo, a Zhejiang University PhD and
Led by CEO Han Fengtao, former co-founder and CTO of Rok
Founded by Wang Xingxing, a Zhejiang native with degrees
Founded by ex-Alibaba VP Bai Huiyuan as CEO, with chief
Founded and controlled by chairman and general manager G
Founded by CEO Wang Hongtao, a Tsinghua- and Harvard-edu
Wholly owned by Ant Group with RMB 100M registered capit
Founded by Hu Luhui, former Meta chief engineering lead
Incubated from Zhejiang University's academician team an
Co-founded by Dr. Wang Donglin, deputy director of Westl
Jointly established by the Automation Research Institute
Incubated by GAC Group and founded in February 2026 as t
Incubated by Dreame Technology and now led by CTO Chen C
Guangdong Tianji Intelligent Systems Co., Ltd. (Tianji R
Founded by CEO Song Tao (2026 Capek Award 'Entrepreneur
Uses teleoperated robots for warehouse sorting and packi
One of China's 'Big Four' automakers, Changan is shiftin
Led by founder-chairman Dr. Xiao Xiangjiang, former dire
One of China's 'Big Four' automakers, Chery has explored
Makes the mobile robot Proxie for collaborative cart and
Automates lab assays in drug discovery, with robots and
One of China's 'Big Four' automakers, Dongfeng designate
Efort Intelligent Equipment is a leading Chinese industr
Founded by legal representative Feng Zhenyu, EIR Technol
Incubated by Estun Automation, one of China's 'four litt
China's first state-owned automaker, FAW debuted its sel
One of China's major automakers, GAC began its humanoid
Makes climbing robots and AI software to inspect power p
Controlled by Dou Qiaoqi and co-developing the 'Laborer'
Makes AI-driven surface-processing robots for sanding, g
Jointly established by Dongyanggwang Group, AgiBot, Peki
UK general-purpose humanoid maker targeting industrial a
Founded in Suzhou in Dec 2024 by Guo Renjie (Xi'an Jiaot
Founded by Zhao Wei, a Southeast University PhD who bega
Korean consumer-electronics giant formally entering phys
Led by CEO Li Weichong, Li-Gong is a national 'Little Gi
Develops robotic factories for complex metal-part manufa
Builds autonomous factories to assemble drones and robot
Builds general-purpose humanoids for hazardous and heavy
Founded by CEO Yan Xunge (David Yan) and CTO Cui Haotian
Founded by Tuo Hua (born 1983, Peking University master'
Seres Group is an automaker in deep partnership with Hua
A Harbin-Engineering-University-lineage deep-sea robotic
Makes the wheeled home robot Memo. Backed by Bain Capita
Maker of the Optimus humanoid. In August 2026 Elon Musk
Founded in 2011, Nanjing TetraBOT is an embodied-intelli
Established on December 31, 2024, with iSoftStone holdin
Incubated by founder Hu Tianlian from a 2009 Southwest U
Founded by Yang Fengyu, a Yale CS PhD born in 2000 with
Makes solar-powered miniature robots that deliver fertil
Spun out of the Toyota Research Institute, Walden Roboti
Originally PengXing Intelligence, acquired and integrate
Founded by Fan Yong, former researcher at CAS Institute
Founded by Zhang Zhaohui, a robotics PhD candidate at Xi
Founded by He Liang, former deputy director of the Shang
Founded by post-2000 members of Tsinghua University's AI
Co-founded by a research team led by Zhejiang University
Delivery-drone operator moving medical supplies, food an
Team sizes and R&D mix disclosed by leading humanoid makers, China and abroad · undisclosed firms are not estimated
Unit: people · dark segment = R&D staff · bracket = as-of date
Unit: people · dark segment = R&D staff · bracket = as-of date
R&D shares among disclosing firms span 36% to over 80%: Unitree 36.5% (175), DEEP Robotics ~40%, Fourier above 40% (200+), DexForce above 70%, RobotEra and MagicLab above 80%, Agibot over 70% (~1,100 working on embodied foundation models). Algorithm and model teams are the core asset of embodied AI.
UBTECH's 2024 report shows RMB 941M total compensation, about RMB 430K per person; Liepin data puts the average new robot-industry posting at ¥328K/yr and humanoid roles at ¥406K, with bachelor-level salaries up 240% YoY.
Noetix grew from ~50 to ~170 people in four months, average age under 25; Agibot's parent already employs 1,570+ and expects to pass 2,000 by end-2026; Unitree grew from 480 at its prospectus date to 516 by end-2025. International leaders are scaling too: 1X Technologies added ~251 people in a year and Apptronik grew 21.3% YoY (Revelio Labs estimates).
Liepin's 2026 robotics talent report: humanoid job postings +215.8% YoY, the wider robot field +75.26%; master's-level postings grow fastest (+87.8%); 63.1% of robot-trainer demand sits in Beijing, Shenzhen, Shanghai and Hangzhou.
Sources: Unitree / DEEP Robotics STAR Market prospectuses, UBTECH 2024 annual report, Eastmoney (2026-08-20), iFeng Tech (2025-05-12), Baobian (2025-09-20), Liepin '2026 Robotics Talent Supply & Demand Trends'; international figures from Neura Robotics' website, PitchBook, Revelio Labs workforce estimates (2026-03) and media coverage. Only firms with credible public disclosures are shown; undisclosed firms are left blank rather than estimated.
Sources: MechAwakening '2026 China Humanoid Company List' (2026-08) and public industry research; A-share tickers limited to high-confidence names.
Monthly deal ranking → company milestones, from wide to close-up (macro totals in Panorama · Market Pulse)
Monthly funding & capital moves across embodied & physical AI
Funding / mass production / launches / orders
Post-money valuation exceeds .3B, setting a record for single-round private equity financing in China embodied AI. Led by IDG Capital with Gaorong Ventures; Tencent and Alibaba as strategic investors. Funds will accelerate XPeng IRON mass production and physical AI model R&D.
Source ↗Funding data compiled from public sources, including the MechAwakening monthly ranking, GG Humanoid Robots' monthly financing statistics and the Jiaohu Zhichu physical-AI capital infographic; for research reference only.
5 robot categories · 24 application scenarios · market size and deployment
Market size varies by source and scope: industrial manufacturing ~¥15B in China 2026 (TrendForce); elderly-care robots >¥10B in 2026 (CCID); global endgame US$7.5T by 2050 (Morgan Stanley). Per-scenario cards cite specific figures and sources.
4 scenarios
Factories are structured environments with human-designed tools. Pick-place, sorting, QC and loading don't require high intelligence but demand stability, precision and 24/7 uptime. Industrial manufacturing is the fastest-deploying domain; ROI matters more than human-likeness.
Factories need not human-like but machine-like robots. Wheeled base + dual arms is pragmatic. H2 2026: pilot to routine deployment, 10K-unit target.
Humanoids cover parcel sorting, non-standard package handling, and bin transport. Their flexible adaptation outperforms traditional automation on irregular items. Logistics has the second-highest deployment density after manufacturing.
Non-standard package handling is the key differentiator over traditional AGVs. Robotera's logistics PMF solution praised by MFA; 1000-unit delivery signals scale.
Live-line welding and grid inspection have entered substantive deployment. Humanoids focus on high-voltage live-line work, substation inspection, and high-altitude maintenance — replacing humans in hazardous zones.
Power inspection is a hard-need scenario — live-line work is risky and expensive. As climbing and fine manipulation improve, power may become the second scaled-deployment sector after factories.
Sowing, fertilizing, spraying, harvesting, scouting and livestock — robots now cover farmland, orchards, greenhouses and pastures. Farm labor shortages and aging farmers are even more acute than in industry: global farm labor costs rose 22.8% from 2020 to 2025, while a smart robot's per-unit cost is already 62% of manual labor. Purpose-built ag robots plus drones dominate today; humanoids have not yet scaled into fields.
Agriculture is an outdoor unstructured scenario demanding robust perception and manipulation. Harvesting and spraying scale first; orchards and greenhouses are next. Humanoids' likely farm entry is wheeled-base dual-arm platforms for picking and sorting.
7 scenarios
Commercial service has moderate tech barriers but clear business models and fastest monetization — 2026's dark-horse. Concierge, delivery, retail guidance, hotel service, and government halls are deploying, replacing repetitive low-value labor.
Commercial logic is traffic and revenue-per-sqm. Robot leasing and scenario-managed models are spreading for SMBs. 2026: hype-driven to genuine value creation.
Cultural tourism and performance is fastest to monetize — low tech barrier but highly eye-catching. Robot bands, Spring Festival Gala, mall interaction, museum tours. The 2025 Gala featured 16 Unitree robots syncing 132 yangge moves within 0.1s.
Demand will shift from pure entertainment to real commercial value. Bionic humanoids in tourism, exhibition, and emotional companionship are 2026's hottest track.
Universities and institutes are key early adopters — for embodied AI research, robotics teaching, sim-to-real experiments. Education has high fault tolerance and strong iteration demand. The 2026 MIIT real-scenario training action mandates 100+ high-value scenarios and 10K-unit deployment.
Education is the bridge from lab to real scenarios. University fault tolerance suits tech validation; scaling to education market requires price points schools can afford.
Retail has clear business models — replacing greeters, guides, cashiers, inventory. Supermarkets, convenience stores, unmanned retail are main scenarios. Humanoids' advantages: 24/7 and multimodal interaction; needs cost down to retail-acceptable levels.
Cost and reliability are key variables. JD's RMB 10B channel strategy could be the breakthrough — lowering OEM customer-acquisition costs and accelerating retail penetration.
Hotel scenarios cover reception, room delivery, cleaning, front desk. Advantages: 24/7 service and multilingual interaction; needs deep PMS integration. Robot leasing is spreading for lightweight hotel deployment.
Core challenge: balancing service standardization with robot flexibility. Chain hotels with unified PMS suit scaled deployment; independents rely on managed-service models. Leasing lowers trial barriers.
Sweeping, scrubbing, vacuuming, disinfection and environment monitoring across malls, airports, rail hubs, hospitals and offices — one of the most mature commercial robot categories. Rising labor costs and cleaner shortages drive demand, while improved navigation has solved most stuck/bump/missed-spot issues.
The dual '#1' claims show no absolute leader yet, but concentration will rise and weak players exit. Going global is the key growth engine — China's complete commercial-robot supply chain delivers strong global competitiveness.
Food delivery, bussing and host-guiding in restaurants is one of robotics' most commercially successful scenarios. Chinese vendors dominate globally on cost and supply chain; going overseas is the core growth engine. Leaders are expanding from delivery into cleaning, industrial delivery and embodied AI.
Delivery robots have passed validation into mature volume; competition shifts to per-unit efficiency and full-scenario service. Next: from delivery to whole-store service, with humanoids unifying bussing and cleaning.
3 scenarios
Aging population and care supply-demand gap are core drivers. 2025: 323M aged 60+, 130M empty-nest, 5M+ nursing gap. Products expanded from cleaning/delivery to five categories: daily care, health monitoring, emotional companionship, transfer assistance, and rehabilitation.
Elderly care is a hard need — 5M+ nursing gap growing. Institution-first, home-fastest pattern is clear. As emotion models and bionic expressions mature, companion robots (42% CAGR) will outpace nursing robots.
Once predicted as harder to land, bionic humanoids became 2026's sexiest track. Equipped with emotion models, 33-axis facial tendon networks and active DoF produce human-like micro-expressions for emotional support, eldercare companionship, and psychological care.
Emotional value is the unique differentiator from all other smart devices. When a robot is a companion, not just a tool, the market ceiling exceeds productivity-replacement logic. Emotion models + bionic expressions are core barriers.
Home is unstructured — complex tasks, long-tail scenarios, close contact with elderly and children, high safety/privacy barriers. Companionship, chore assistance, and nursing care are appealing but generalization is the bottleneck. Short-term: no scale. Long-term: largest market globally.
Home is the endgame market but hardest short-term. The bottleneck is generalization, not hardware cost — home task diversity and unpredictability far exceed factories. Real scale likely post-2028 with VLA model breakthroughs.
3 scenarios
Hospitals are among the densest robot venues: drug/specimen/meal delivery, sterile-supply runs and UV disinfection follow fixed routes with high frequency and night shifts. Certification barriers are high, but once in, supply-chain stickiness is strong.
Hospital robotics is moving from point delivery to delivery + disinfection + surgery chains. Remote surgery's 2026 Nature publication marks a clinical inflection; remote and autonomous operation are the next mainline.
Medical rehabilitation has high barriers but rigid demand. Surgical, rehab, and nursing robots are piloted in top hospitals. Domestic products match international standards in fall detection, remote monitoring, and rehab training; multimodal perception and AI continue to advance.
Medical barriers: long certification cycles and high clinical validation costs, but once approved, deep moat. Rehab robots are the best testbed for multimodal perception and motion control; aging will drive rehab first.
Smart city is a long-term scenario for humanoids in public services. Urban inspection, security patrol, government halls, community service, transit guidance are piloting. Integration with urban infrastructure will form a new governance paradigm; scaling needs policy and standards.
Smart-city scale depends on government procurement and standards. Beijing's RMB 100B fund and MIIT action provide strong policy catalysts. Once public-service operational standards form, city-level deployment opens a trillion-yuan market.
7 scenarios
Emergency rescue is still in early exploration. Humanoids have irreplaceable value in disaster relief, nuclear environments, deep-sea exploration, and firefighting. At WRC 2026, the Chinese Institute of Electronics and the Ministry of Emergency Management released 'Embodied AI Emergency Application Requirements.'
Emergency rescue is humanoids' ultimate value scenario — replacing humans in unreachable environments. Technical requirements far exceed commercial, but demand is rigid with ample government procurement budgets. Deep-sea, nuclear, extreme-temp products already emerging.
Space exploration, on-orbit servicing, sampling, station support and experiments — the hardest robotic domain. China now fields a full spectrum from robotic arms to in-cabin robots and deep-space probes, with routine on-orbit maintenance and module relocation.
With the station in its utilization phase, on-orbit servicing (refueling, repair, debris removal) and deep-space sample return are the next big markets. Tianwen-2 asteroid sampling and Chang'e lunar operations continue; China's space robotics ranks in the world's first tier.
Seabed survey, underwater inspection, offshore-engineering O&M, aquaculture patrol and deep-sea science — a typical high-value 'human-unreachable' domain. ROV/AUV dominate, while humanoid and bio-inspired underwater robots enter validation. Pressure resistance, sealing and underwater comms are core barriers.
Ocean economy and deep-sea minerals are national strategies; underwater robotics is one of few special tracks driven by both import substitution and exports. As waterproofing and buoyancy control mature, underwater humanoid operations may move from trials to engineering deployment.
Fire reconnaissance, extinguishing, smoke extraction and hazmat handling — firefighting is the most commercially mature special-robot scenario. Petrochemical parks, hazmat warehouses and large venues have rigid unmanned-firefighting demand with ample procurement budgets.
Firefighting robots are upgrading from reconnaissance to recon + extinguishing + smoke-extraction chains; heat resistance and autonomous navigation are key barriers. After the ministry's requirement list, government procurement may scale further.
Routine inspection, radiation-zone work, maintenance and decommissioning — high-risk scenes where humans cannot linger. Radiation-hardened electronics, teleoperation and autonomous navigation are core barriers; global first-generation reactor decommissioning plus new units coming online make long-term demand certain.
Nuclear certification cycles are long but moats are deep. Radiation-hardened chips, sensors and whole-robot localization are the breakthrough directions; the decommissioning market will release intensively over the next decade.
Underground inspection, auxiliary transport, drainage/ventilation monitoring and emergency response — mining is the strongest policy-driven scenario for 'machines replacing humans'. Unmanned hazardous underground environments are a regulatory hard requirement with rigid budgets.
Policy is mining robotics' first driver, with 'fewer people, more safety' regulation intensifying. Unmanned-underground rates will become a mine KPI; inspection and auxiliary transport scale first, with emergency response and extraction next.
Reconnaissance, logistics and battlefield support — military robotics centers on unmanned ground vehicles. Quadruped platforms were the first to become practical for recon and patrol thanks to all-terrain mobility; procurement is classified but demand is rigid and unit prices high.
Military robots are expanding from single-unit recon to coordinated squads and logistics. Quadruped/tracked hybrids and jam-resistant comms are the focus, with strong bidirectional civilian-military tech spillover (e.g., quadruped control feeding commercial products).
In 2026 the industry crossed the threshold from show-off to utility. 50+ scenarios mapped, 100+ integrators forming, 10K-unit targets set. Policy (MIIT training), capital (RMB 46B+ in H1), and scenario demand are driving humanoids into real-world operation mode.
Industrial lands first — structured environments, ready tooling, urgent cost-reduction. Commercial is closest to monetization but shifting from hype to value. Home has highest barriers, no short-term scale, but largest long-term market. Factories need machine-like, not human-like, robots.
Morgan Stanley predicts 1 billion humanoids by 2050, a US$7.5T market — about 7% of current global GDP. Before 2030 the market stays flat, but from 2035 penetration far exceeds linear expectations. China's role may mirror EVs: winning the world with cost and supply-chain advantages.
H1 2026: China shipped 40K+ humanoids, 97% global share. Agibot led 2025 with 5100 units, Unitree 4200. China's structural advantages in cost control, supply-chain completeness, and scenario richness may replicate its EV success path in humanoids.
Sources: '2026 Humanoid Robot Industry Development Report' (China Humanoid & Embodied AI Council, WRC 2026), TrendForce, Morgan Stanley 'Humanoid Horizons', '50+ Scenario Application Atlas' (New Strategy Consulting), CCID, ChinaIR and other public research reports; for research reference only.