40+ Humanoid Robot Companies Raising Billions

Over 40 companies are racing to build humanoid robots. Dozens more provide the technology stack underneath them, from foundation models to actuators to batteries.

The space is massively fragmented. China has more than half the robot makers. The US has Tesla, Figure AI, Agility Robotics, and a handful of others. Europe has a few players. Most won't survive.

Here's who matters and why.

Figure AI just closed over $1 billion at a $39 billion valuation in September 2025. That makes it one of the most valuable startups in any category. Backers include Microsoft, OpenAI, NVIDIA, Amazon, and Intel Capital.

Figure terminated its collaboration agreement with OpenAI in February 2025. CEO Brett Adcock said OpenAI's intelligence models weren't the right fit for humanoid robotics. Figure is now building fully end-to-end robot AI in-house. The company aims to produce 100,000 robots and plans to show something no one has ever seen on a humanoid within 30 days of the announcement.

Figure's 01 and 02 robots work in BMW factories in South Carolina. The company focuses on general-purpose humanoids for manufacturing, logistics, and retail. Earlier funding rounds valued Figure at $2.6 billion in February 2024 when it raised $675 million.

Tesla Optimus isn't raising outside capital, but Elon Musk claims it will become Tesla's bestselling product. The company plans to produce 5,000 Optimus robots in 2025 for internal use at Tesla facilities. Target price: $20,000 to $30,000 once production scales. Commercial sales planned for 2026.

Optimus Gen 2 showed improvements in walking speed (30% faster), weight (10 kg lighter), and hand dexterity (11 degrees of freedom with tactile sensing). The third iteration is expected to feature hands with 22 degrees of freedom. Tesla integrates its Full Self-Driving software for environmental perception and decision-making.

Critics point out that most Tesla demos happen in curated environments with short runs. Continuous operation, robust error recovery, and adaptation to dynamic spaces haven't been demonstrated publicly. Milan Kovac, head of the Optimus program since 2022, resigned in June 2025.

Agility Robotics (Oregon) builds Digit, a bipedal humanoid specifically designed for warehouse automation. Amazon is testing Digit in its facilities. The company also has multi-year deployment agreements with GXO Logistics. Agility raised $150 million led by DCVC with participation from Amazon's Industrial Innovation Fund.

Digit has bird-like legs and human-like arms. It walks, climbs stairs, and handles packages. Unlike general-purpose humanoids, Digit focuses on the narrow use case of logistics and last-mile delivery. Industry estimates suggest it costs less than $250,000 per unit, though Agility hasn't disclosed official pricing.

Apptronik (Texas) builds Apollo, a modular humanoid designed for versatility across industries including construction, production, and manufacturing. The company positions Apollo as adaptable to various environments with a modular design that allows different configurations.

Sanctuary AI (Canada) develops Phoenix, a humanoid with advanced cognitive capabilities. The company partnered with Microsoft to develop "large behavior models" that ground AI in real-world experiences. Phoenix features improved uptime, miniaturized hydraulics, better tactile sensing, and enhanced visual acuity in its sixth generation.

Sanctuary AI recently began pilots with automotive manufacturer Magna. Phoenix is Sanctuary's first humanoid to walk on two legs.

Boston Dynamics (owned by Hyundai) built Atlas, the benchmark for dynamic humanoid robotics. Atlas performs backflips, parkour, and navigates industrial sites with near-human precision. The latest version is designed for real-world applications, not just research.

Boston Dynamics is working with Hyundai to build next-generation automotive manufacturing capabilities. Atlas serves as a testing ground for new applications. Unlike other companies on this list, Boston Dynamics hasn't announced commercial availability or pricing for Atlas.

China has more humanoid robot makers than the rest of the world combined, according to the Goldman Sachs map. Key players include:

UBTECH builds Walker S1, which has received more than 500 orders from major automakers including BYD. Walker S1 delivers 360-degree safety monitoring with special sensors to grip items with the right amount of force. UBTECH positions Walker S1 for generalized industrial tasks.

Fourier introduced GR-2, its second-generation humanoid, in September 2024. The company followed up with footage of the robot weight training.

Xiaomi is developing humanoid technology, though details on specific models and deployment timelines are limited.

BYD appears as both a maker and a customer, ordering Walker S1 units from UBTECH for its manufacturing facilities.

Other Chinese companies on the list: Addverb (Reliance), AGIBOT, Beijing Embodied Intelligent Robot Innovation Center, CASBOT, CATL, CloudMinds (Dataa), Codroid (Estun), DEEP Robotics, ENGINEAI, GAC, GALBOT, HUAWEI, HumaNoid, JAKA, KEPLER, Leju Robot, LimX Dynamics, Mentee, NEURA, PaxIni Technology, Robot Era, Sancturay AI, Shanghai Embodied Intelligent Innovation Center, TiS Robot, Unitree, UniX AI, XPENG, Zhejiang Humanoid Robot Innovation Center.

Most are privately owned. Many won't survive. The ones that do will likely consolidate or get acquired.

1X Technologies (Norway, backed by OpenAI) takes a different approach by targeting households rather than industry. The company unveiled NEO in August 2024. NEO Beta stands 1.6 meters (5'4"), walks at 4 km/h (2.5 mph), and runs up to 7.5 mph. It weighs 30 kg (66 lb), carries up to 20 kg (44 lb), and runs 2-4 hours on a single charge.

1X started testing NEO Beta in a small number of homes. The startup says robots could be ready for wider deployment in a few years. Once production scales, 1X plans to price NEO about the same as a modest car. The company raised $100 million in Series B and hired veterans from BMW and Tesla.

NEURA Robotics (Germany) secured $123 million to advance its cognitive humanoids. The company unveiled the third generation of its humanoid, 4NE1, designed for domestic and professional environments. 4NE1 is powered by NVIDIA GR00T N1 and was trained in Isaac Sim and Isaac Lab before real-world deployment.

Other European companies on the list are minimal, with most activity concentrated in the US and China.

The Goldman Sachs graphic breaks down the full stack:

Foundation Models: Tesla, NVIDIA, OpenAI, Alphabet, Meta, iFlytek, Huawei

GPU/CPU: NVIDIA, Intel

Actuator Assembly: Sanhua, Tuopu

Rotary Actuators: Harmonic Drive, LeaderDrive, Sling, Reach Machinery, FORE Intelligent, Zhongda Leader, Shuanghuan, Laifual, Picea Motion

Linear Actuators: Schaeffler, Sanhua, Beite, XCC, Seenpin, LeaderDrive, Best, Hengli Hydraulic, Zhenyu, CSB, Shuanglin

Dexterous Hands: Maxon, Moons' Electric, LeaderDrive, Zhaowei, Fortior, Inovance, RoboSense, Leadshine, Leili, Veichi, FORE Intelligent, Paxini

Vision (camera, LiDAR, etc): DJI (Livox), Intel, Orbbec, RoboSense, Hesai, Sunny Optical, Velodyne, Innoviz, Luminar

Sensors: Novanta (ATI Industrial Automation), Vishay Precision, Kunwei, Murata, TDK

Batteries: CATL, Samsung SDI, Panasonic

This reveals the supply chain dependencies. NVIDIA provides both foundation models and chips. Chinese component makers dominate actuators and sensors. Japanese companies lead in precision components. The battery technology comes from EV supply chains.

Morgan Stanley forecasts a $3 trillion US market for humanoid robots by 2050. Bank of America projects costs dropping from $35,000 per unit in 2025 to $17,000 by 2030, potentially enabling one million sales in that timeframe.

The artificial intelligence in robotics market will grow from $23.12 billion in 2025 to $64.74 billion in 2029 at a 29.4% CAGR, according to The Business Research Company.

Gartner projects that at least 15% of work decisions will be made autonomously by agentic AI by 2028, compared to 0% in 2024. That creates the addressable market for humanoid robots in enterprise settings.

The investment landscape shows over 1,700 funding rounds supporting 390+ companies. Average funding per round: $11.6 million. Top investors contributed $1.5 billion, with Boston Dynamics investing $400 million.

The Goldman Sachs map shows 40+ humanoid robot makers. History suggests maybe five will survive as independent companies. The rest will fail, get acquired, or pivot to components.

The winners will be the ones that pick a narrow use case and execute relentlessly. Agility Robotics focuses on warehouse logistics. UBTECH targets automotive manufacturing. Figure AI goes after general-purpose industrial work. Tesla leverages its manufacturing and AI capabilities.

The losers will be the ones that try to build general-purpose household robots without a clear path to revenue. Consumer robotics is littered with failures. The market isn't ready. The technology isn't ready. The price points don't work.

The component makers and technology enablers will do fine. NVIDIA sells chips regardless of which robot maker wins. Actuator suppliers, sensor manufacturers, and battery companies benefit from the entire ecosystem scaling.

Watch the deployment numbers, not the funding announcements. Figure has factory pilots with BMW. Agility has multi-year contracts with Amazon and GXO. UBTECH has 500 orders from automakers. Those are real. Everything else is potential.

That's where humanoid robotics stands in 2025.

Billions in funding. Dozens of companies. A handful with real deployments. The rest racing to prove their robots can do something useful before the money runs out.

Forward this to someone tracking robotics or enterprise automation.