Humanoid robots are moving from science fiction into factories, warehouses and potentially our homes, as tech giants and ambitious startups race to bring them into the real world.
Adobe Stock
Humanoid robots are having their ChatGPT moment.
For decades, they belonged in science fiction, glossy lab demos and viral videos. Now they are starting to appear in factories, warehouses and test homes, with some of the world’s biggest technology companies racing against ambitious startups to build robots that can work safely alongside people.
The appeal is easy to understand. Most robots are designed for one specific task. A humanoid robot, in theory, can operate in spaces built for humans, use tools designed for humans and move between different jobs without the need to rebuild the environment around the machine.
And with billions of humanoid robots predicted to walk among us in the coming decades, investors, manufacturers and technology giants are paying attention. If humanoid robots can be produced at scale, they could help tackle labor shortages, take on dangerous work and eventually support everyday tasks in homes, hospitals and care settings.
The field is still young, and many bold claims remain unproven. Even so, momentum is building fast. In China, the humanoid robot ecosystem is expanding rapidly. The Ministry of Industry and Information Technology said the country had more than 140 humanoid robot manufacturers by 2025, with over 330 product models, while major players in the U.S., Europe and Asia are pushing prototypes toward commercial deployment.
So, who are the companies leading this race, and which names should business leaders be watching?
1X Technologies has attracted attention because of its backing from OpenAI and its focus on humanoid robots for the home.
The company has shifted from wheeled industrial robots toward domestic humanoids designed to live and work around people. Its Neo robot is being marketed as a home assistant, which, in demonstrations, has been seen to perform everyday chores such as folding laundry, unloading dishwashers and tidying up clutter.
Neo can reportedly be bought for $20,000 or rented for $499 per month, and deliveries are currently expected to begin in 2026. The company’s big bet is that the first mass-market humanoid robots may arrive through the front door rather than the factory gate.
That is a bold strategy. Home environments are messy, unpredictable and full of edge cases, which makes them far harder than controlled industrial settings. If 1X can make domestic humanoids reliable and useful, it could help define what a robot assistant in the home will look like.
AgiBot is one of the most important Chinese humanoid robotics companies, even though it remains less well-known in the West.
Backed by BYD, the company is believed to be among the first to deliver humanoids at serious scale, with reports suggesting it has already delivered 10,000 units. Its Yuanzheng A2 series includes highly dexterous hands capable of delicate tasks such as threading a needle.
AgiBot has also generated headlines for endurance and mobility. One of its robots reportedly completed a 106km trek from Suzhou to Shanghai without falling over, setting a world record in the process.
The company reflects China’s fast-moving approach to humanoid robotics. Rather than focusing purely on spectacle, AgiBot appears to be moving quickly toward volume, deployment and industrial relevance.
Agility Robotics is one of the clearest examples of humanoids moving from lab demos into practical work.
Its Digit robot has been developed with warehouse and logistics tasks in mind, and the company has worked with Amazon on exploring how robots could support operations across large fulfillment networks. Digit’s backward-bending legs give it a distinctive look, while also improving stability and lifting ability.
Agility has also invested in production capacity. Its RoboFab facility in Oregon is reported to be capable of manufacturing 10,000 humanoid robots a year.
Agility is one of the companies trying to prove that humanoids can be manufactured in meaningful numbers and deployed in real operational environments.
Apptronik is another serious contender in the industrial humanoid race.
The U.S. company’s Apollo robot builds on earlier work carried out for NASA’s Valkyrie robot and has benefited from backing from Google DeepMind. Apollo is designed for factory and logistics work, with swappable hands that can be replaced by tools and hot-swappable batteries that allow it to keep working without long recharge breaks.
The robot is already being used on factory floors by companies including Mercedes-Benz.
That places Apptronik in a strong position. Rather than presenting humanoid robots as a distant consumer fantasy, it is targeting environments where the business case is more immediate, especially manufacturing, warehousing and repetitive physical work.
Boston Dynamics has done more than almost any other company to shape public expectations of what robots can do.
Its videos of walking, jumping and highly mobile robots have gone viral for years, sometimes inspiring awe and sometimes giving people the feeling that the future has arrived slightly too early. The company began as an MIT spinout and was later acquired by Hyundai.
Its famous Atlas robot has long been a benchmark for humanoid mobility. More recently, Boston Dynamics has focused on moving Atlas toward industrial labor roles and pilot deployments.
The company’s challenge is turning world-class robotics engineering into scalable commercial products. Its advantage is credibility. When Boston Dynamics shows a robot doing something difficult, the world pays attention.
BYD is best known as one of the world’s largest electric vehicle makers, but it is also becoming an important player in humanoid robotics.
Like Tesla, BYD sees humanoid robots as a potential way to increase efficiency in large-scale manufacturing. The company is developing its own robotics capabilities while also partnering with other robotics companies, including AgiBot and UBTech.
This makes strategic sense. EV factories are complex, high-volume environments where automation can deliver measurable value. If humanoids can support assembly, logistics and inspection tasks, carmakers have a strong incentive to put them to work.
Engine AI is another Chinese manufacturer gaining attention, particularly for the natural movement of its humanoid robots.
Many humanoids still move with the stiff, cautious gait that people associate with early robotics. Engine AI has focused on creating robots that move in a smoother and more fluid way. In 2025, it reportedly achieved the first front flip by a humanoid robot.
A robot that can move dynamically, recover balance and handle complex physical motion has more potential in real-world environments.
Figure AI has quickly become one of the most closely watched startups in humanoid robotics.
Its Figure 02 robot has already been used to support vehicle production at BMW’s Spartanburg plant in the U.S., including work connected to the production of tens of thousands of vehicles. The company has also attracted high-profile backing from OpenAI, Nvidia and Microsoft.
Fourier began by building medical rehabilitation robotics before expanding into general-purpose humanoid robots.
Its GR series, including the GR-1 and newer GR-3, is designed for a wide range of use cases, from warehouse support to clinical research. One of its more distinctive features is a cushioned, soft exterior intended to make the robot feel safer and more approachable around people.
That design choice points to an important issue. If humanoid robots are going to work around humans, they need to be trusted. Industrial robots have traditionally been separated from people by cages and safety zones. Humanoids will have to operate much closer to us.
Fourier’s healthcare and rehabilitation background gives it useful experience in building machines that interact physically with people.
Neura Robotics
Germany-based Neura is a frontrunner in Europe’s physical AI race. Its aesthetic appearance is down to the fact that it was designed in collaboration with the studio responsible for the looks of the Porsche 911. The 4NE1 features synthetic “skin” that can sense contact in a way that makes interactions with people safer and more natural. It’s no softy though, being capable of lifting up to 100 kg.
Germany-based Neura Robotics is one of Europe’s most interesting companies in what is often called physical AI.
Its 4NE1 humanoid has been designed with attention to both function and appearance, including input from the design studio responsible for the Porsche 911. It also features synthetic skin that can sense contact, making interactions with people safer and more natural.
Despite its friendly look, 4NE1 is built for serious work and is reportedly capable of lifting up to 100kg.
Noble Machines is a California startup founded by former Apple, SpaceX and NASA employees.
The company has moved quickly, reportedly putting its first humanoid robots to work at a Fortune 500 company within 18 months. Its focus is on difficult, dirty and dangerous jobs that expose human workers to risk.
That is likely to be one of the strongest early use cases for humanoid robots. The first wave of valuable deployments may come in jobs that are physically demanding, repetitive or unsafe.
Samsung is approaching humanoid robotics from the perspective of a global manufacturing powerhouse.
The company has acquired stakes in a number of robotics firms, including Rainbow Robotics and has announced plans to transform its global production facilities into AI-driven factories by 2030. Humanoid robots are expected to play a central role in that strategy, with possible responsibilities including assembly, logistics and facility management.
Samsung has a major advantage that many startups lack: it can manufacture at scale. It also has enormous experience in electronics, sensors, semiconductors and consumer technology.
If Samsung decides humanoid robots are a strategic priority, it has the resources and industrial footprint to become a serious force in the market.
Sanctuary AI has focused heavily on intelligence and dexterity.
Its Phoenix humanoid is designed to excel at fine motor control, which is one of the hardest problems in robotics. Moving around is difficult enough, but using hands with human-like precision is even harder.
The company’s stated mission is to create and deploy millions of humanoid robots into industry to help address global labor shortages. Its partnership with Magna International, a major automotive supplier, has already seen its technology deployed in automotive sub-assembly roles.
Tesla is one of the most talked-about companies in humanoid robotics, largely because of Elon Musk’s huge ambitions for Optimus.
Musk has said that humanoid robots could become more important to Tesla than cars. That sounds extraordinary, but it reflects a clear strategic logic. Tesla has expertise in AI, batteries, electric motors, manufacturing and real-world automation, all of which are relevant to building humanoid robots at scale.
Optimus prototypes are regularly demonstrated and are reportedly being piloted internally within Tesla facilities.
The big question is whether Tesla can do for humanoid robots what it did for electric vehicles: take an ambitious technology, manufacture it at scale and make it feel inevitable. That remains far from guaranteed, but Tesla’s manufacturing depth makes it impossible to ignore.
Toyota has taken a broad and thoughtful approach to robotics.
The company is exploring humanoid robots for vehicle assembly while also investing in soft robotics for care and companionship. Through the Toyota Research Institute, it has developed platforms such as Punyo, a soft, squishy humanoid designed to interact safely with people and objects.
Toyota’s interest in care robotics is significant as Japan has one of the world’s oldest populations, and many countries face similar demographic pressures. Robots that can safely support older adults, assist caregivers and handle physical tasks in homes could become increasingly valuable.
UBTech is another Chinese company that may not be widely known in the West, but it has become an important name in industrial humanoids.
The company has focused on factory deployments rather than generating constant media attention. That approach has reportedly helped it build one of the world’s largest datasets of humanoid industrial activity and one of the largest numbers of active humanoid deployments in automotive factories.
Data could be a huge advantage in humanoid robotics. Robots improve by learning from real-world tasks, failures and edge cases. The more deployed robots a company has, the more practical experience it can feed back into its systems.
Unitree Robotics has become one of the most visible Chinese robotics companies, helped by its focus on affordability and commercial delivery.
Its humanoid models, including the G1 and H1, are priced at around $16,000, making them far more accessible than many competing systems. The company reportedly delivered 5,500 units to customers in 2025, with many believed to be used in research, education and small-scale industrial pilots.
Lower-cost robots can accelerate experimentation across universities, labs and smaller companies. More users mean more feedback, more applications and potentially faster development.
Xiaomi is best known for consumer electronics, but it has also moved into humanoid robotics with its CyberOne model.
The company has highlighted features including a 4.7-billion-parameter vision, language and action model, as well as an active liquid-cooling system described as a form of robotic "sweat." CyberOne is reportedly being piloted in Chinese EV manufacturing.
Humanoid robots bring together several of the most important technology trends of the moment: artificial intelligence, advanced sensors, computer vision, batteries, automation and high-precision manufacturing.
The big opportunity is generalization. A robot built for one task can be useful, but a robot that can move through human environments and perform many different tasks could be transformational.
The hard part is everything else.
Humanoid robots must be safe, reliable, affordable and genuinely useful. They need better hands, better balance, better reasoning and better common sense. They also need business models that make sense outside carefully controlled demos.
For now, factories and warehouses are likely to lead adoption because they offer structured environments and clearer returns on investment. Homes, hospitals and care settings may follow later, once robots become safer, cheaper and better at handling the unpredictable nature of everyday life.
The excitement is justified, but so is caution. Humanoid robots are still in their early stages. Some companies will overpromise, some prototypes will never become products, and some business models will fail.
Yet the direction of travel is clear. AI is moving from screens into the physical world, and humanoid robots are one of the most visible signs of that shift.
The race is no longer about whether robots can walk across a stage or perform a carefully rehearsed trick. The real question is which companies can build robots that people and businesses actually want to use.
That is where the next chapter in the evolution of technology will be written.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。