"They're coming": how humanoid robots are storming factories, warehouses and our hearts

Just a few years ago, humanoid robots evoked two emotions in us: admiration for Boston Dynamics' videos with salts and nostalgia for the Terminator. But in 2025, they are already assembling cars at BMW factories, carrying boxes to Amazon, and learning to make coffee in hospital cafeterias. This is not science fiction - this is the economy of the future, which is coming faster than we can update the firmware on our own smartphones.

Let's find out who is behind this wave of "iron evolution", why the humanoid market risks becoming a trillion-dollar market by 2050, and whether we should be afraid of competing for jobs with machines that never take sick leave.

Why now: the era of humanoids has begun

In 2025, the humanoid robotics industry has finally stopped being a club for enthusiastic engineers and turned into a serious business. Here are the reasons for this growth:

• Next-generation artificial intelligence: large language models (LLMs) such as OpenAI GPT now teach robots not only to speak but also to act in the physical world.
• Billions of investments: the market annually receives more than $1 billion in venture capital and strategic investments from Google, Microsoft, and NVIDIA.
• Labour shortage: In the US and the EU, warehouses and factories are experiencing an acute shortage of workers, especially in "hard" and monotonous positions.
• Currently, humanoid robots account for less than 1% of the industrial robot market, but MarketsandMarkets forecasts that this will grow from $2.92 billion in 2025 to $15.26 billion in 2030. And Morgan Stanley talks about a potential market of $5 trillion by 2050. This forecast assumes that almost 1 billion humanoid robots will be put into operation by the middle of the century, of which the vast majority (90%) will be used for industrial and commercial purposes.

Almost 1 billion humanoid robots will be put into operation by the middle of the century

The main players: who is building the humanoid army

So who are these players? What can they do? And why are investors, from OpenAI to Jeff Bezos, pouring money into them like it's Black Friday?

Figure AI (Figure 02)

California-based Figure AI is developing Figure 02, a humanoid robot with a focus on industrial and logistics applications. The robot is 1.7 m tall, weighs 60 kg, and can lift loads of up to 25 kg. Figure 02 features powerful integration with OpenAI's extensive language models, which allows it to perform tasks based on complex speech commands and understand the context of situations. Figure AI is working with BMW, where Figure 02 is being tested at the Spartanburg, South Carolina, plant, performing tasks such as carrying and sorting components. The company has strong investors: NVIDIA, Microsoft, OpenAI, and Jeff Bezos. The total amount of funding has already reached $675 million. Figure AI's strategic advantage lies in the combination of hardware with advanced cognitive functions. The challenge facing Figure AI is to turn successful test programmes into full-fledged commercial solutions and further reduce the cost of robots.

The robot is already being tested at the BMW plant in Spartanburg, South Carolina. Here, Figure 02 moves metal sheets and other components, demonstrating its ability to work in an environment where traditional automation is not. The company's key bet is not on hardware but on deep AI integration, which allows the robot to learn new tasks and adapt to unforeseen situations

Apptronik (Apollo)

Apptronik, an American startup based in Texas, has created Apollo, a humanoid robot specifically designed for industrial use. The robot is 1.73 m tall and has a lifting capacity of up to 55 kg. The main feature of Apollo is its modular design, which allows it to be easily adapted to various tasks, including automotive production and logistics. Among Apptronik's key partners are Mercedes-Benz and GXO Logistics, which are testing Apollo in their production and warehouses. The robot uses artificial intelligence created in collaboration with Google DeepMind, which allows it to perform complex tasks with high accuracy. Apptronik also established a separate subsidiary, Elevate Robotics, to develop industrial robotics. In 2025, the company received significant funding ($403 million), including Google and Mercedes-Benz. Apptronik's main challenge is to scale up production and prove Apollo's cost-effectiveness in an industrial environment.

Agility Robotics (Digit)

The American company Agility Robotics became known for the humanoid Digit, developed on the basis of research in the field of dynamic motion at Oregon State University. Digit is a bipedal robot with a height of approximately 1.2 m and a weight of about 30 kg. It features reverse-curved legs that enable stable and efficient walking. Digit is designed to solve specific tasks in the field of logistics and warehouse automation, such as carrying boxes, containers and other goods. Key customers already conducting pilot tests of Digit include Amazon and GXO Logistics. Agility Robotics has opened a dedicated RoboFab factory in Oregon to produce several hundred robots a year. The main challenge for the company is to expand commercial applications, ensure high reliability of robots for work in complex and diverse environments, and reduce costs. Digit is sold for more than $250,000, but is also available as a robot-as-a-service (RaaS) model at an estimated price of $10 per hour.

Tesla (Optimus)

Tesla, headed by Elon Musk, aims to revolutionise robotics in the same way it did electric cars. Its Optimus project is a universal humanoid robot about 1.73 m tall and weighing 57 kg. The project's features include an extremely low price (less than $20,000) and integration with artificial intelligence technologies developed for Tesla's autopilot. Optimus uses only cameras for environmental perception (no LiDAR). In 2025, it is planned to produce 5,000 units, and by 2026, to increase production to 50,000 units per year. Despite the ambitious plans, the robot is still in the research stage and has no commercial applications. The main challenge remains the real economic feasibility of such a universal solution, as well as addressing ethical and social issues regarding its integration into society.

Boston Dynamics (Atlas)

Boston Dynamics remains a symbol of cutting-edge achievements in robot mobility. Atlas, their most famous robot, has been demonstrating amazing agility for over a decade, from parkour to complex gymnastic moves. In 2024, the company switched from hydraulics to electric drives, a step towards commercialising Atlas. This transition made the robot quieter, more powerful, and more energy efficient. Boston Dynamics is now focusing on practical applications in industry and logistics where Atlas can work alongside humans. Nevertheless, Atlas is still primarily a research platform that sets standards in the field of humanoid robots. The main challenge for Boston Dynamics is to reduce the cost and adapt Atlas' complex technologies to real-world industrial applications.

Sanctuary AI (Phoenix)

The Canadian company Sanctuary AI has created Phoenix, a humanoid aimed at achieving human levels of intelligence and motor skills. Phoenix is 1.65 m tall, weighs about 55 kg, and has extremely developed five-fingered hands that allow performing complex manipulations: folding clothes, sorting goods, etc. Sanctuary AI relies on the integration of powerful artificial intelligence with high-precision sensors, which allows Phoenix to operate effectively in service industries and light manufacturing. In 2025, pilot programmes will continue, but the robot is not yet in commercial mass use. The main challenge for Sanctuary AI is to demonstrate the robot's clear commercial efficiency and adapt it to a wide range of tasks outside the laboratory.

Unitree Robotics (G1)

China's Unitree Robotics, known for its quadrupedal robots, has created a humanoid model G1 with a low cost of $16,000. The main idea of G1 is to make humanoid robots accessible to a wide range of researchers, educational institutions and developers. The robot is focused primarily on research and development. It does not have extremely sophisticated manipulative abilities, but is capable of stable walking and mobility. Unitree's strategic goal is to form a large community around G1 that will independently develop its software. This can significantly speed up the adaptation of humanoids to real-world tasks. The challenge for Unitree Robotics remains to maintain a balance between low price and sufficient functionality to make the robot truly useful outside of laboratories.

Fourier Intelligence (GR-1/GR-2)

Fourier Intelligence specialises in medical humanoids(GR-1 and GR-2) focused on rehabilitation and medical care. The Chinese company is actively using its previous experience in the development of medical exoskeletons to create robotic assistants. In 2025, Fourier attracted more than $109 million in investments to boost development. The GR-1 and GR-2 robots can provide physical support to patients during rehabilitation and perform health monitoring, which meets China's growing need for medical staff and elderly care. However, the company is still at the stage of commercial and research implementations, and the main challenge remains to ensure that the robots are accurate, reliable and safe enough to work alongside people in medical facilities.

Wandercraft (Calvin)

The French company Wandercraft, which has gained popularity in the field of medical exoskeletons, is currently actively developing the industrial humanoid Calvin. This is a robot being developed in partnership with the Renault Group specifically for use in automotive factories. Calvin has an advantage due to the company's experience in creating high-precision mechanical systems. The robot is still in the development stage, but the prospects for its commercial application look promising thanks to the significant support from Renault. The challenge for Wandercraft is to transform specialised medical technologies into more versatile industrial solutions that can be effectively integrated into mass production.

NEURA Robotics (4NE-1)

German company NEURA Robotics is a prominent European player in the field of humanoid robots, which is actively developing the 4NE-1 project, a "cognitive robot" for industrial and service applications. The main goal is to create autonomous machines capable of working directly alongside humans through the deep integration of advanced AI and modern sensor technologies. A special feature of 4NE-1 is its own Neuraverse ecosystem, which provides robots with environmental recognition, decision-making and complex manipulation tasks. In 2025, NEURA Robotics attracted €120 million in investments, including Volvo Cars Tech Fund and Lingotto Investment Management as strategic partners. The main challenges for NEURA Robotics remain scaling up production, ensuring the safety of people in shared work areas, and further developing the Neuraverse software platform to ensure maximum adaptability and efficiency of the 4NE-1 robot in a wide range of real-world commercial applications.

1X Technologies (NEO)

1X Technologies is a Norwegian-American company that is actively developing the NEO universal humanoid robot. The main idea of the robot is to enable it to perform a wide range of tasks autonomously, both in industry and in the home environment. NEO is based on the company's proprietary Redwood artificial intelligence integrated with NVIDIA's advanced GR00T model. This enables the robot to efficiently recognise and perform household tasks such as cleaning, sorting and other household chores, and is also able to adapt to industrial environments. The company's key investors and partners include OpenAI and NVIDIA, which confirms 1X Technologies' serious ambitions in developing NEO's cognitive functions. In 2025, the robot is still at the prototype stage, but it is already showing promising results. The company's main challenge is to strike a balance between functionality, affordable price, and ensuring reliability and safety of work alongside humans.

Why investors are excited (and a little scared)

Investors are euphoric (but also a little nervous) because humanoid robots are starting to look like an increasingly realistic business case. The current cost of one such robot ranges from $30,000 to $150,000. Tesla promises to create its Optimus for a fantastically low price of $20,000, but even experienced Boston Dynamics engineers meet such statements with a sceptical smile. However, despite doubts about the cost, the actual profitability figures are impressive: Toyota is already reporting savings of $4.7 million annually per production line, and Samsung has demonstrated a 52% increase in production cycle speed and a 36% reduction in rejects. On average, the payback period for a humanoid robot has been reduced to 18-24 months, while in 2019 it took 36-48 months. It is no surprise that the humanoid robot market is gaining real traction among investors.

Where are humanoid robots already working?

Humanoid robots are confidently moving beyond laboratories, gradually becoming part of the real economy. In the manufacturing sector, automotive giants such as BMW, Mercedes-Benz, and Toyota are already actively employing humanoid robots in their factories, where robots take on the most difficult and monotonous tasks on conveyors. Logistics is also keeping up with the pace: Agility Robotics' bipedal Digit is undergoing pilot tests at Amazon and GXO Logistics warehouses, where it sorts packages and transports containers. The medical sector has not been left out either - China is launching a large-scale programme to deploy 200 humanoid robots in nursing homes. Their task is not only to help with household chores and health monitoring, but also to communicate socially. Finally, in particularly dangerous areas, such as man-made disaster sites or nuclear facilities, the four-legged robot Spot from Boston Dynamics is already being actively used. Although Spot is not yet a humanoid, it is a direct predecessor to humanoid robots, which will be the next logical step in such conditions.

• Factories - BMW, Mercedes-Benz, and Toyota use humanoids on conveyors for heavy and monotonous tasks.
• Warehouses - Digit Agility Robotics is being tested at Amazon and GXO Logistics - it moves containers and sorts packages.
• Healthcare - China launches 200 humanoid robots in nursing homes to help with household chores, communication, and health monitoring.
• Dangerous areas - Boston Dynamics Spot (not yet a humanoid, but a close relative) is already checking ruins and nuclear facilities. Humanoids are the next step.

Problems: batteries, ethics, and "awkward dances"

Despite all the euphoria surrounding humanoid robots, the industry faces a number of serious problems. The first of these is batteries: even the most advanced models today can last no more than 1-4 hours on a single charge, which significantly limits their autonomy and requires constant charging or battery replacement. The second and no less important challenge is social integration. People may not yet be ready to see a humanoid as a full-fledged colleague, rather than a competitor who takes away their jobs. And the third and most delicate area is ethical responsibility: the question arises as to who will be held accountable if the robot makes a mistake, for example, breaks a coffee machine in the office or makes a serious error in a hospital.

What's next?

Despite the obstacles, analysts are optimistic that humanoid robotics will develop in stages. The first phase (2025-2030) will cover structured environments such as factories and warehouses, where robots will work according to clear rules and clear tasks. In the second phase (2030-2035), robots will gradually move to semi-structured spaces, such as hotels, shopping malls, and hospitals, where more flexible behaviour will be required. It is only after 2035 that we can expect to see the massive integration of robots into homes, where they will finally become everyday human assistants.

For those who want to know more

• What slows down self-driving cars
• G-SHOCK and XG - how Casio changed its course from "survival watches" to neon style for the TikTok generation
• Editor's column: we already need a Turing test for humans
• From failed rice cooker to PlayStation triumph: the story of Akio Morita
• How conspiracy theories led to the hacking of NASA's servers and ruined a sysadmin's life: the story of Gary McKinnon