the humanoid moment is real, the timeline isn't
china is shipping 36x more units. that's not a demonstration. that's a market forming.
The headline you saw: "Chinese humanoids beating US robots."
What it actually means: different timelines, different constraints, different customers.
Chinese player Unitree shipped roughly 36 times more units last year than Figure or Tesla. That's not a rounding error. That's a different market. They're not competing on the same stage.
Figure AI and others in the US are optimizing for one thing: can this robot weld a car? Can it handle industrial environments? Can it scale to millions of deployments?
Chinese players optimized for what they can sell today. Deliveries, logistics, manufacturing support. Cheaper units. Faster iteration. Get them into the field and let the data flow back.
Neither timeline is wrong. But they're racing different things.
The bet I had to make: which approach actually generates the data that matters for the next decade? I used to think it was the US approach—harder problem, bigger payoff. But 36x more units in the field means 36x more feedback cycles, 36x more edge cases, 36x more data.
Data is everything in robotics. The team that has it wins. Not because they had better engineering three years ago, but because they have better hardware understanding right now.
This is why the China narrative isn't hype. It's just market mechanics.
Are you betting on the team with the longest runway, or the team with the fastest feedback loop?