The line that built Model S now builds robots

Converting a real automotive assembly line is a much stronger commitment than building a pilot cell — and a much more expensive one to reverse.

Tesla is converting the Fremont Model S and X line, which ended production in May, to build Optimus, with the V3 reveal timed to the start of manufacturing.

Why the line matters more than the robot

A pilot cell is a research expense. A converted automotive assembly line has tooling, a supply chain and trained staff attached, and unwinding it means writing all three off. That makes this a materially harder commitment than any demo, and harder commitments are better evidence than better demos.

The warning attached is worth as much as the target. Musk's stated goal is a million units a year at Fremont; his stated caveat is that initial output will be slow and the rate essentially unpredictable, citing roughly 10,000 unique parts on an entirely new line. That is a supply-chain problem before it is a robotics problem, and it is where every manufacturing ramp in history has actually lived.

Revealing at production is a change of posture

Timing the V3 reveal to the start of manufacturing rather than ahead of it means the thing shown is the thing being built. That invites a direct comparison with what eventually ships — a comparison the industry's demo-first convention has generally been arranged to avoid.

Meanwhile, volume is elsewhere

Unitree ships more humanoids than any Western competitor at roughly a tenth of the price, following 5,500-plus shipments last year and targeting 10,000 to 20,000 this year. Figure 03 has passed 1,000 units at about one robot per hour.

The price gap does more work than the unit counts. A platform at a tenth of the cost is not competing for the same purchase decision — it is competing for the labs, universities and integrators who would otherwise build nothing. That is how ecosystems get seeded, and ecosystems are what eventually determine which platform the software targets.

One robot per hour is the more meaningful Figure number, because a rate extrapolates and a cumulative total does not. It also quietly concedes something: an hour per unit is assembly, not manufacturing, and the distance between those two words is most of the distance to volume.

The metric nobody publishes

Unit counts are the least informative number in this sector, which is exactly why they are quoted. Deployment duration, task breadth and intervention rate would tell you far more. Every one of these companies tracks all three internally. None of them publish it, and the silence is the most reliable signal available.

Electrek — Tesla pushes Optimus V3 reveal later this year — again → · Technology.org — Humanoid robots in 2026: what is actually deployed → · Interesting Engineering — The humanoid robot reality check →