Factories around the world now run on roughly 5.08 million industrial robots, and the International Federation of Robotics says that is the highest operational count it has ever recorded. The figure rose 9% over the previous year, according to the IFR's World Robotics report published in late September. Jane Heffner, the federation's president, described it without ceremony: "Industrial automation is progressing at high speed."
Installations hit a record as well. Manufacturers took delivery of 603,000 new industrial robots last year, an 11% increase and the first time the annual total has crossed the 600,000 mark. The IFR, a Frankfurt-based non-profit that has tracked the sector for close to three decades, counts machines that weld, paint, machine metal, move parts and stack pallets. Not one of them looks like a person.
That is the story buried under the humanoid hype. The machines doing paid work in 2026 are arms bolted to pedestals, not androids. Bank of America still forecasts 90,000 humanoid shipments this year and 1.2 million by 2030, but the IFR's first-ever count of the category found only about 7,000 humanoids sold worldwide last year for industrial and professional use. Many buyers bought them to generate training data for AI models rather than to run production lines. The distance between those two numbers is where the industry's capital and its press releases currently sit, and the mismatch is starting to show up in earnings calls, IPO filings and courtrooms.
China is placing the orders
China installed 354,000 robots last year, up 20%, and now accounts for 59% of all deployments worldwide. No other market is close, and the gap is still widening.
The United States overtook Japan for second place, installing just under 38,500 units, up 12%. That is a quiet reversal. Japan held the number two slot for decades, and its loss of it says as much about soft domestic demand as it does about anything happening in America.
Heffner told reporters the growth is concentrated where you would expect. The strongest expansion is in Asia, followed by the Americas. Europe, she said, is moving ahead more slowly. The report's numbers back her up.
Europe is installing fewer, not more
Europe's stock of working robots keeps growing, but the flow of new ones has stalled. The European Union passed 700,000 operational robots for the first time, reaching 712,000, which is 84% of every robot running anywhere in Europe. That stock has more than doubled since 2015, when it stood at 390,000.
New installations tell the opposite story. EU installations fell 11% in 2025 to 60,500 units, even as Asia accelerated. The bloc's five-year compound annual growth rate sits at a modest 3%.
Germany remains the largest European market and takes 41% of EU installations, yet the number of robots installed there fell 8% to under 25,000. Automotive is still the single biggest buyer in Germany, and the car sector's retreat is doing the damage. Italy, France and Spain each added somewhere between 4,300 and 7,800 robots, with installations in all three down by double digits.
A stalled install rate in Europe is not a collapse. It is a pause, and pauses in manufacturing investment tend to end when order books fill. Whether that happens before the humanoid vendors claim the narrative is another question.
The humanoid math still does not close
The IFR tracked humanoids as a separate category for the first time this year, and the findings were sobering for the category's loudest promoters. About 7,000 humanoids were sold worldwide in 2025 for industrial and professional use. Many of those went to companies collecting data, not lines running shifts. Car manufacturers piloting the machines typically run fewer than ten of them.
Elon Musk has been unusually blunt about his own program. "This is going to be the hardest product to scale manufacturing that we've ever made at Tesla, because everything on the robot is new," he told investors on Tesla's second-quarter earnings call. He added that "there is no supply chain."
Boston Dynamics, owned by Hyundai, is unlikely to pursue a stock listing next year, a senior Hyundai executive said this month. Its Atlas humanoid has not been deployed at scale and the company remains unprofitable.
China's market has already run the experiment. Shares in Unitree, a robot maker, soared more than fivefold on their August 19 Shanghai debut, then fell 55% from the peak and have struggled to recover. Regulators have since used informal guidance to hold back a queue of humanoid IPOs.
Real deployments do exist. Siemens trialled a robot for logistics tasks at its Erlangen plant, and BMW is deploying a humanoid in car and battery production. Dave Grant, CEO of PickNik Robotics, questioned the whole form factor in a report from the consulting firm Arthur D. Little, asking "Why would we limit it to mimic the human form factor...?" Human-level dexterity, autonomy outside controlled settings and unresolved liability and insurance questions remain open problems.
The IFR's own conclusion is that most humanoids remain at the prototype or pilot stage, while conventional industrial robots are still cheaper for factory work. Read plainly, the world's robotics story in 2026 is not about humanoids at all. It is about the slow, steady build-out of machines that already work.
Drones face a different bottleneck
Automation's other fast-moving segment is stuck on paperwork rather than physics. The FAA's forthcoming proposal would let certified drone operators fly beyond the visual line of sight, ending a long and often costly waiver process. Reuters reported that 15 states, including New York, California, Massachusetts, Wisconsin and Michigan, plus Harris County, Texas, sued the agency on September 28 over its environmental review of the rules. The case was filed in the US Court of Appeals for the Second Circuit.
California Attorney General Rob Bonta said the "FAA's programmatic environmental assessment falls far short of properly analyzing the environmental impacts that drone package delivery operations may have." The FAA's review modeled noise from 1,150 average deliveries out of a single hub, with drones flying under 400 feet and avoiding parks, campgrounds and trailheads.
Researchers estimate per-delivery costs could fall to as low as $2 by 2034. That is the prize on the table while the lawyers argue.
What to watch next
The install numbers will keep coming from China and the United States, and Europe's recovery will hinge on the auto sector. The interesting question is what the humanoid vendors report when their own pilot data matures: whether the 90,000 shipments Bank of America projects for this year materialize as working machines or as another round of demos. Our desk covers robotics and drones as the second wave of deployments gets audited, and the archive holds earlier coverage of the sector's funding rounds and regulatory fights.
For now the balance sheet is clear. Five million robots punch clocks every day, and the industry's most-hyped product has yet to prove it can hold a shift.