Short answer: Not yet, at least not at scale. This week Xiaomi said its humanoid robots worked a full 8-hour shift in its car factory, and AGIBOT livestreamed six humanoids running a convenience store for 24 hours. Both are real progress, but neither company released the numbers, like success rates or how often humans stepped in, that would show the robots can reliably do a person’s job.
Key takeaways
- Xiaomi posted video on November 3 of humanoid robots sorting car parts for a full 8-hour shift. It did not say how many parts they sorted, their success rate, or how often a human stepped in.
- AGIBOT ran six G2 humanoids through a 24-hour convenience store livestream on November 2 to 3. Reporting notes the event does not prove the robots worked without charging, maintenance or human help.
- Ordinary factory robots are where the real money is moving: North American companies ordered 10,989 robots worth $723.5 million in the third quarter of 2026, up 24.8% in units, according to A3 data.
- Neither company said the robots replaced any workers. For now, humanoids are tests, not headcount cuts.
Did Xiaomi’s humanoid robots really work a full factory shift?
What happened: On November 3, Xiaomi posted footage of its full-size humanoid robots working an 8-hour task at the “Center Console Side-Cover Sorting Workstation” in its electric car factory. The robots pull plastic side covers from bins, carry them to storage racks and slot them in place. Xiaomi says the robots can “autonomously detect and correct anomalies,” meaning they notice and fix their own mistakes, and that they coordinated with the production line’s control system. Electrek reported that the video carries an “Autonomous, 100x Speed” label, so viewers see a sped-up version of the shift.
This is the latest step in a year of testing. According to Electrek, a Xiaomi humanoid worked as an “intern” at its Beijing car factory in March, installing nuts with a 90.2% success rate over three hours, fast enough to meet that station’s 76-second cycle time. By July, Xiaomi said that rate was up to 98% and added two harder stations, including this sorting job. Xiaomi CEO Lei Jun has said the company plans to deploy humanoid robots in large numbers across its factories within five years, Electrek reported.
Why it matters: An 8-hour shift is the unit that matters for jobs. Short demos show a robot can do a task once. A full shift tests whether it can keep doing it, the way a worker must. If humanoids can run whole shifts on their own, factory tasks that were too varied for traditional robot arms become candidates for automation.
The other side: Xiaomi did not say how many parts the robots sorted, what the success rate was over eight hours, how many times a human had to step in, or how the robots stayed powered all shift. Electrek’s editor wrote that the demo would be impressive if fully autonomous, “But we’ve been disappointed by humanoid robot demos before, and Xiaomi left out the numbers that would let anyone check.” Note too that 98% success sounds high, but it still means about 1 miss in every 50 tries. At a 76-second cycle, an 8-hour shift is roughly 379 cycles, so that rate would leave about 7 or 8 errors per shift for someone, or something, to catch.
Can humanoid robots run a store?
What happened: Chinese robotics company AGIBOT ran a livestream called “Robots Never Clock Out” from November 2 to November 3. Six of its G2 humanoid robots worked as “convenience store trainees,” in the company’s words, stocking shelves, making coffee and hot food, greeting customers and clearing trash across a lunch rush, evening rush, late-night duty and breakfast prep, Interesting Engineering reported.
Why it matters: Retail and food service employ millions of people, and their work is messy: odd-shaped items, crowded aisles, customers who interrupt. A long livestream shows more than a polished clip, including pauses, retries and recoveries.
The other side: Interesting Engineering noted that the 24-hour figure refers to the livestream and “does not establish that each robot operates continuously without charging, maintenance, or human assistance.” The promotional material it reviewed also did not spell out an intervention policy or a scoring system for task success. Some coverage, including a widely shared TechSpot headline, said two of the six robots did not perform well enough to “keep their jobs.” We could not open that article to verify the details, so we leave those results out.
How do this week’s robot tests compare?
| Test | What was shown | What was not disclosed |
|---|---|---|
| Xiaomi factory shift (Nov. 3) | Humanoids sorting car parts for 8 hours; video shown at 100x speed | Parts sorted, 8-hour success rate, human interventions, power source |
| Xiaomi nut install (March to July) | 90.2% success over 3 hours, later 98%; met a 76-second cycle time | Results over a full shift; error handling |
| AGIBOT store (Nov. 2 to 3) | Six G2 robots stocking, cooking, greeting, cleaning over 24 hours | Store location, scoring system, intervention policy, charging breaks |
Are companies actually buying more robots?
Yes. North American companies ordered 10,989 robots valued at $723.5 million in the third quarter of 2026, according to Association for Advancing Automation (A3) data reported by Robotics & Automation News. Units rose 24.8% and value rose 26.1% from a year earlier, the strongest quarter of 2026 so far. Through nine months, orders reached 28,984 robots, up 9.6%. The data are broken out by industry, like carmakers and electronics, and the coverage we read did not single out humanoids.
| Industry | Change in robot units ordered, Q3 2026 vs. Q3 2025 |
|---|---|
| Semiconductors and electronics/photonics | +97.6% |
| Automotive components | +90.4% |
| Plastics and rubber | +67.7% |
| Life sciences, pharma, biomedical | +38.7% |
| Metals | +37% |
| Automotive OEMs (carmakers) | +7.5% |
| Food and consumer goods | -23.7% |
Collaborative robots, or “cobots,” designed to work next to people rather than behind safety cages, made up 16.7% of units ordered, with 1,831 units, up about 56%. “What stands out in the third quarter is not simply the magnitude of the growth, but how broadly it was distributed across the market,” said Alex Shikany, executive vice president at A3. The lesson for workers: the robot arriving at your workplace next year is more likely a cobot or a fixed arm than a humanoid.
The other side: why humanoid demos still matter
It is easy to dismiss demos, and skepticism is healthy. But the trend line is real. Xiaomi went from a 3-hour test in March to a full 8-hour shift in November, and its reported success rate climbed from 90.2% to 98% in four months. Companies are also moving to longer, livestreamed tests that expose failures instead of hiding them. If that pace continues, the gap between demo and deployment could close faster than skeptics expect.
The counterpoint is cost and reliability. A factory does not need a robot that works on a good day. It needs one that works every day, at a lower total cost than a person, with clear safety records. None of this week’s announcements included price, uptime or safety data. Until they do, humanoids remain a research and marketing story more than a hiring story.
The TrustNoRobot Risk Check
| Question | Our read on humanoid robots and jobs |
|---|---|
| Exposure | Moderate over time for repetitive factory station work and simple store tasks like stocking. Low today for jobs with lots of customer contact, judgment or unusual items. |
| Evidence | Weak. Company videos and partial success rates, with no independent data on interventions, uptime or cost. No layoffs tied to humanoids were announced. |
| Speed | Slow to moderate. Testing is speeding up, and Xiaomi talks about large-scale use within five years, but that is a plan, not an outcome. |
| Cushion | Fair. Robots still need people to supervise, fix and handle exceptions. Traditional robots and cobots are the nearer-term change most workers will see. |
What this means for you
If you work on a factory line or in a store, a humanoid robot is not likely to take your shift this year. The more useful question is whether your tasks are repetitive, timed and done at one station. Those are the tasks companies are testing first. Skills that get more valuable as robots spread include robot supervision, maintenance, troubleshooting and handling the exceptions machines cannot.
When you see the next robot demo, ask four questions: How long did it run? What was the success rate? How often did a human step in? What does it cost? If a company will not answer, treat the demo as marketing.
To check your own job, try the Job Exposure Explorer and read our guide on how to measure AI risk to your job. For more on robots at work, see our earlier looks at humanoid robots in factories, warehouse robots at Amazon and robotaxis and driver jobs.
Frequently asked questions
Are humanoid robots replacing factory workers in 2026?
Not at scale. Companies like Xiaomi are testing humanoids at individual stations, but neither Xiaomi nor AGIBOT said its robots replaced any workers this week.
Can a humanoid robot work a full 8-hour shift?
Xiaomi says its robots did, sorting car parts for 8 hours. It did not release success rates, part counts or how often humans stepped in, so the claim cannot yet be checked.
How many robots are companies buying?
North American companies ordered 10,989 robots worth $723.5 million in the third quarter of 2026, up 24.8% in units from a year earlier, according to A3 data. Collaborative robots, or cobots, made up 16.7% of those units.
How can I tell if a robot demo is real?
Look for run time, success rate, number of human interventions and cost. Labels like “autonomous” mean little without those numbers.
Sources
- Electrek, “Xiaomi’s humanoid robots work a full 8-hour shift in its EV factory” (Nov. 4, 2026)
- Interesting Engineering, “Watch: AGIBOT puts six humanoid robots through a 24-hour convenience store shift” (Nov. 3, 2026)
- Robotics & Automation News, “North American robot orders increase 25 percent in third quarter of 2026” (Nov. 2, 2026), reporting A3 data
Leave a Reply