Humanoid robot factories scale up as industrial robot stock hits 5 million

The International Federation of Robotics said on 27 September that the global operational stock of industrial robots rose 9 percent to a record 5 million units in 2025. Factories installed more than 600,000 new units during the year, an 11 percent jump, according to the Frankfurt-based organisation.

"Industrial automation is progressing at high speed," said Jane Heffner, president of the IFR, in the statement. "The new mark of five million robots operational in factories worldwide is more than double the number seven years ago. The strongest growth is taking place in Asia, followed by Americas. Europe is moving ahead more slowly."

China alone accounted for 59 percent of global installations, with 354,000 units in 2025, up 20 percent year on year. The United States overtook Japan to become the second-largest market, installing almost 38,500 units, a 12 percent rise. The IFR forecasts installations will grow 9 percent to 655,000 units in 2026 and reach 806,000 by 2029.

Those figures cover industrial robots broadly, not humanoids specifically. They frame the context for a wave of humanoid robot factories now being built. They also explain why the IFR's Heffner stressed regional divergence rather than a single global boom.

The humanoid factory race

Tesla's Fremont factory in California stopped making the Model S sedan and Model X SUV as of May 2026. The company switched both line workers and engineers over to working on Optimus, according to reporting by The Information cited by Ars Technica on 26 September. Tesla has reportedly scaled production to hundreds of robots per week, but is targeting more than 1,000 per week by the end of 2026.

The same Ars Technica account describes several production problems. Tesla is struggling to get production line equipment to line up components precisely, and running the line too fast has created limitations. Human workers must manually assemble Optimus hands and forearms that together have more than 100 small components such as screws. Touch sensors for the hands have been unreliable enough that Tesla developed a glove-like layer of sensors that can be replaced without replacing the entire robot hand. The robot's AI capabilities are also reportedly insufficient for general-purpose operations. One source described Optimus robots as currently requiring programming to do specific tasks in carefully controlled environments. Tesla did not respond to those characterisations in the Ars Technica piece.

There is a labour dimension too. The Information described some workers complaining because they "knew the robots were designed to eventually replace them," according to Ars Technica. Tesla has apparently shifted data collection responsibility to dedicated teams and set up "training hubs" for such teams.

"We're exploring a range of robot designs and form factors, including wheels, to address different operating environments and customer needs."

That is Jonathan Hurst, CEO and chief robot officer of Agility Robotics, speaking to The Robot Report in an article published on 25 September. Agility, whose Digit humanoid is one of the furthest-along commercial deployments, is now exploring wheeled robots. Hurst said it is "too early to say" whether a wheeled version of Digit or an entirely new robot is planned, and there is no timeline for market entry.

The form factor question matters commercially. Agility is going public through a SPAC with Churchill Capital Corp. XI, valuing the company at $2.5 billion and expected to generate more than $620 million in gross proceeds. The S-4 filing showed $1.8 million in 2025 revenue against a $140 million operating loss, and more than $300 million in multi-year Digit v5 orders from one customer.

Agility's move away from a pure legged strategy is notable for a company whose identity is rooted in bipedal locomotion. With Digit 5, Agility dropped the backwards-bending knees used on earlier versions. The Robot Report framed the wheeled exploration as a sign the company is willing to adapt to market realities rather than stay wedded to one design philosophy.

Toyota's $6.42 billion a year

Toyota is planning a larger deployment. Starting in 2028, the Toyota Motor group aims to invest $6.42 billion every year in upgrading factories with a new robotic workforce, according to Nikkei Asia, as reported by Ars Technica on 23 September. The company wants 150,000 robots in its own automotive plants and another 250,000 at group company facilities producing components and materials.

Toyota has already begun deploying ELEY humanoid robots across assembly lines to learn from human workers. Nikkei Asia reported that workers are teaching the robots, which roll on wheels, to perform tasks requiring precise hand movements by wearing "jigs" based on the humanoids' fingers. Hiroki Nakajima, a Toyota executive vice president, told Nikkei Asia the company aims for "a world where robots coexist with humans, rather than replacing them."

It is unclear how many robots Toyota has already deployed overall or how many of the new robots will be humanoids. The company has 18,000 veteran workers across 60 factories worldwide, including 11 US manufacturing plants.

Hyundai is pursuing a parallel path. The South Korean automaker plans to deploy up to 25,000 Atlas humanoid robots developed by its US subsidiary Boston Dynamics over the next several years, with deployment starting in 2028. On 21 September, Boston Dynamics announced a new robotics training center at Hyundai Motor Group Metaplant America in Georgia, designed to train Atlas robots on tasks such as sequencing automotive parts.

Asked about the broader safety debate, The Robot Report published a piece on 27 September arguing that Isaac Asimov's Three Laws are no longer sufficient. The article points to Dario Amodei, CEO of Anthropic, who on 12 September called for slowing frontier AI development, and to a statement signed by more than 2,500 economists and technologists under the heading "We Must Act Now." Jensen Huang, CEO of Nvidia, said concerns about AI dangers are overblown, according to the same article.

For all the production news, the underlying economics remain unsettled. Waymo has logged more than 220 million autonomous miles and gives around half a million paid rides a week, but of 86 million Waymo miles reported through 2025, only 54 percent carried a passenger, according to The Robot Report on 26 September. Each robotaxi reset costs roughly two hours. The article's author, Dan Keene of Aseon Labs, disclosed that his company is building robotic servicing stations for autonomous fleets.

Elsewhere, the physical AI push is drawing fresh capital. O-ID, a Tokyo-based company founded by two Europeans, raised $1.2 million in pre-seed funding led by Chicago-based TAWANI Ventures, with participation from Hustle Fund, Techstars and angel investors, Tech.eu reported on 24 September. The company is building modular humanoid robots for manufacturing and logistics, with factory pilots planned for 2027. Japan has targeted 10.5 trillion yen, about $65.1 billion, in public and private investment in physical AI, aiming to deploy 10 million AI-equipped robots by 2040, according to Tech.eu.

On the research side, Stanford and Caltech researchers built HomeBody, a system that lets a Unitree G1 robot navigate an unfamiliar kitchen and tidy up using GPT-6 Astra as a swappable vision-language model, The Decoder reported on 27 September. Limitations include latency, overheating finger servos and high compute costs.

The common thread is that the factory, not the demo, is now the bottleneck. Tesla is assembling hands by hand. Toyota is training workers to train machines. Agility is rethinking legs. The IFR's 5 million figure shows the industrial base is deep. Whether humanoids join it at scale depends on lines that can run faster than a person can teach them.