Are Humanoid Robots Really Becoming Mainstream? The Requirements for 'Working Robots' Behind the 7,000 Units Worldwide
Are Humanoid Robots Really Becoming Mainstream? The Requirements for "Working Robots" Behind the 7,000 Units Worldwide
This analysis examines whether humanoid robots are becoming mainstream by comparing global sales data from the International Federation of Robotics (IFR) with actual industrial deployments. It highlights that while approximately 7,000 full-sized humanoid robots were sold in 2025, traditional industrial robots remain dominant in volume. The piece details specific use cases, such as Figure AI's robot handling over 90,000 parts at a BMW Spartanburg plant and Hexagon Robotics' wheeled 'AEON' unit used for battery assembly at a Leipzig facility, arguing that practical adoption depends on task-specific design rather than human-like form factors alone. The text concludes that current deployments represent a phase of exploring specific tasks rather than achieving broad profitability or replacing all existing automation.
Full article
Article text is not available for this item. Read it at the original source.
Read full article at sourcenote.com · Opens in a new tab; source language may differ.
The article analyzes the discrepancy between global sales figures for humanoid robots and their actual industrial deployment, citing specific case studies from BMW and GXO to evaluate operational track records rather than just unit counts.
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
parts handled
>90,000
View original evidence
er year alone does not tell us how many hours they worked on-site or how much cost they reduced. To evaluate their adoption, it is necessary to go one step deeper than just sales figures. 2. At a BMW factory, a humanoid robot handled over 90,000 parts Regarding actual work, the case of the automaker BMW is a useful reference. At the Spartanburg plant in South Carolina, USA, BMW introduced Figure AI's "Figure 02" to t
Open source S9vehicles produced with robot assistance
>30,000
View original evidence
of over 30,000 BMW X3 vehicles .[3] Do not misinterpret this as "a robot manufactured 30,000 cars." What the humanoid was responsible for was a portion of the parts-handling tasks among the many processes required to complete a car. Nevertheless, there is a difference between a video of a robot moving for a few minutes at an exhibition and a track record of repeatedly handling parts on a mass-production line. What i
Open source S10parts handled per hour
≈72
View original evidence
with existing equipment, and role division with workers are all in place. Dividing over 90,000 parts by 1,250 hours, the simple average is approximately 72 parts per hour. However, the processing capacity of a single robot cannot be determined by this calculation alone. This is because unless the time periods counted, the number of units operated, and the consistency of work conditions are aligned, it cannot be compa
Open source S12containers transported
>100,000
View original evidence
frequently swapped, the value of a human-like form may increase. 4. What is happening in logistics is not "generalization" but "task fixation" Let's look at another specific example. At the facilities of logistics company GXO, the bipedal robot "Digit" from Agility Robotics is responsible for moving containers. In October 2026, GXO announced that Digit had transported a total of over 100,000 containers at its base in
Open source S17effective work hours at 80% rate
2,400 hours
View original evidence
rate is 80%, 2,400 hours out of 3,000 hours can be used for work. Dividing by 7 million yen per year, it is approximately 2,917 yen per hour of effective work. If it is 60%, it is 1,800 hours, so it becomes approximately 3,889 yen. Even with the same robot, the cost per hour of work changes significantly just by increasing the time it is stopped. Furthermore, let's assume the cost when a human performs that task is
Open source S22effective work hours at 60% rate
1,800 hours
View original evidence
rate is 80%, 2,400 hours out of 3,000 hours can be used for work. Dividing by 7 million yen per year, it is approximately 2,917 yen per hour of effective work. If it is 60%, it is 1,800 hours, so it becomes approximately 3,889 yen. Even with the same robot, the cost per hour of work changes significantly just by increasing the time it is stopped. Furthermore, let's assume the cost when a human performs that task is
Open source S22
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
er year alone does not tell us how many hours they worked on-site or how much cost they reduced. To evaluate their adoption, it is necessary to go one step deeper than just sales figures. 2. At a BMW factory, a humanoid robot handled over 90,000 parts Regarding actual work, the case of the automaker BMW is a useful reference. At the Spartanburg plant in South Carolina, USA, BMW introduced Figure AI's "Figure 02" to t
Open source S9
of over 30,000 BMW X3 vehicles .[3] Do not misinterpret this as "a robot manufactured 30,000 cars." What the humanoid was responsible for was a portion of the parts-handling tasks among the many processes required to complete a car. Nevertheless, there is a difference between a video of a robot moving for a few minutes at an exhibition and a track record of repeatedly handling parts on a mass-production line. What i
Open source S10
with existing equipment, and role division with workers are all in place. Dividing over 90,000 parts by 1,250 hours, the simple average is approximately 72 parts per hour. However, the processing capacity of a single robot cannot be determined by this calculation alone. This is because unless the time periods counted, the number of units operated, and the consistency of work conditions are aligned, it cannot be compa
Open source S12
frequently swapped, the value of a human-like form may increase. 4. What is happening in logistics is not "generalization" but "task fixation" Let's look at another specific example. At the facilities of logistics company GXO, the bipedal robot "Digit" from Agility Robotics is responsible for moving containers. In October 2026, GXO announced that Digit had transported a total of over 100,000 containers at its base in
Open source S17
of commercial operation progress and evidence of management success should be received separately. A contrast is Amazon. The company announced in June 2025 that the number of robots introduced into its logistics network had reached a cumulative total of 1 million. The core consists of dedicated robots that move shelves and cargo, and there is no need for them to have a human appearance.[6] This "1 millionth unit" is
Open source S19
rate is 80%, 2,400 hours out of 3,000 hours can be used for work. Dividing by 7 million yen per year, it is approximately 2,917 yen per hour of effective work. If it is 60%, it is 1,800 hours, so it becomes approximately 3,889 yen. Even with the same robot, the cost per hour of work changes significantly just by increasing the time it is stopped. Furthermore, let's assume the cost when a human performs that task is
Open source S22
er 2026, the global sales volume in 2025 for full-sized humanoid robots over 140 centimeters tall was approximately 7,000 units. The scope covers commercial and professional use, excluding research and development or entertainment purposes.[1] In the same document, global shipments of professional service robots were approximately 250,000 units. Among them, transportation and logistics applications accounted for 117,
Open source S5
Source：Industrial robotics — News discovery · note.com