Humanoid robots are nearly ready for the factory floor. The price tag is not.
The technology behind warehouse robots has come a long way. The business case still needs work, and here is why cost, not artificial intelligence, is the real obstacle.
Key points
- Humanoid robots from leading developers currently cost several hundred thousand dollars per unit, making most deployments financially impractical.
- Specialised parcel-sorting robots achieve up to 3,000 picks per hour; the best humanoid demonstrations reach around 1,000 picks per hour in controlled conditions.
- BMW is expanding its humanoid pilot, moving from the Figure 02 to the newer Figure 03 model at its Spartanburg, South Carolina plant.
- A perspective published by The Robot Report argues Tesla's manufacturing expertise gives it a head start on reaching commercially viable humanoid prices.
- The FCC has moved to ban Chinese humanoid robot imports over national security concerns, which may limit one route to lower-cost units.
Humanoid robots, machines shaped roughly like people that can handle physical tasks in human-designed spaces, can now do things that looked like science fiction a few years ago. They pick boxes, navigate warehouses, work alongside people on assembly lines. BMW is already running a pilot at its South Carolina plant, and the results were good enough that the company is upgrading to the newer Figure 03 model.
The technology moved faster than the spreadsheet did.
What is actually stopping the rollout?
The short answer is money. One leading developer prices its humanoid robots at several hundred thousand dollars each. To make that spend pay off, a factory would need to run the machine across multiple shifts, around the clock, before the numbers begin to break even. And if you already run that kind of operation, the same budget often buys a purpose-built machine that does a single job faster and pays off sooner.
The gap is easy to picture. In parcel sorting, the best humanoid robots hit roughly 1,000 picks per hour under ideal conditions. Specialised sorting robots, built for exactly that task and nothing else, run between 1,300 and close to 3,000 picks per hour. That is a meaningful difference when you are deciding where to spend capital.
| Robot type | Picks per hour (approx.) |
|---|---|
| Humanoid (best demo conditions) | ~1,000 |
| Specialist sorter (low end) | ~1,300 |
| Specialist sorter (high end) | ~3,000 |
Is the underlying technology actually ready?
Mostly, yes. Vision systems, physical hardware and the notoriously tricky hands have all improved to the point where robots handle real work in real warehouses. Agility Robotics, a company we covered in July earning money from humanoid deployments while bigger names were still demoing, has proven results at warehouse scale. The industry has moved on from "can they do the job?" to "can we afford them?"
Human oversight still matters. The smartest deployments treat people as remote supervisors who step in only when a robot hits an unusual situation, rather than operators controlling every move. One person can monitor 50 or more robots at once that way, and every correction becomes training data.
What happens next?
Price comes down with production volume. Build more units, each one gets cheaper. Straightforward logic, but it requires genuine large-scale manufacturing experience, not just research expertise.
The Robot Report's argument is that Tesla has an edge here because building electric vehicles already put it through years of brutal supply-chain and production challenges. That knowledge transfers to its Optimus humanoid. Meanwhile, lower-cost Chinese humanoid models have shown what factory-scale production can do to the price, though the FCC's recent move to restrict Chinese humanoid imports adds real uncertainty to that picture.
The beat reporters' read: the AI story is being oversold. Manufacturing discipline is what separates the companies that will ship at scale from those that will keep releasing impressive demos. Watch which humanoid developers announce supplier agreements and production targets, not which ones post the slickest warehouse footage.
For most factories right now, the specialised machine still wins on cost and speed. Humanoids are a genuine option for businesses willing to absorb the premium and the wait.
Common questions
Do humanoid robots replace human workers entirely?
Not in current deployments. Human supervisors still oversee multiple robots and step in when something unexpected happens, so the technology shifts roles rather than eliminating them outright.
Why would a company choose a humanoid robot over a regular industrial robot?
Humanoids can, in theory, handle many different tasks in spaces built for people, without custom installations. The trade-off is cost and speed; specialised machines are still faster and cheaper for high-volume single tasks.
