China just crossed 10,000 mass-produced humanoid robots while Silicon Valley is still doing demos.

The Summary

  • Agibot, a Shanghai-based startup, hit 10,000 manufactured humanoid units, doubling production from 5,000 just three months prior
  • This is manufacturing velocity, not lab prototypes—Agibot is building humanoids at scale while Western competitors iterate on hardware
  • China is treating humanoid robotics like it treated solar panels and EVs: as an infrastructure play, not a moonshot

The Signal

The number matters less than the trajectory. Agibot went from 5,000 to 10,000 units in 90 days. That's not startup tinkering. That's industrial capacity ramping. The company announced the 5,000-unit milestone in late December, which means they're doubling production roughly every quarter. At this rate, they'll hit 40,000 by year-end.

Compare that to the U.S. humanoid landscape. Figure's robots are impressive but measured in dozens, maybe low hundreds deployed. Tesla's Optimus has headlines but unclear production numbers. Boston Dynamics built something incredible and got bought by Hyundai. The American approach treats humanoids as frontier tech. The Chinese approach treats them as manufacturing output with predictable unit economics.

This is the playbook China ran on EVs and solar: subsidize early, build overcapacity, drive down unit costs, export globally once you've achieved price dominance. Agibot isn't selling these robots as luxury research tools. They're positioning for commodity hardware at scale. When humanoids become cheap enough for warehouse deployment, factory floor work, and eventually consumer applications, the company with 100,000 units in the field wins. Not the one with the best TED talk.

The agent economy needs bodies. Software agents are proliferating, but the physical world still requires physical manipulation. Whoever controls the hardware layer controls the economics of automation. Right now, that's looking like a Chinese advantage built on manufacturing throughput, not just engineering elegance.

The Implication

If you're building in the agent economy, start planning for cheap Chinese humanoid hardware the same way you plan for cheap Chinese displays and batteries. It's coming. The question is whether Western robotics companies can compete on production velocity or if they'll end up licensing software to run on Agibot chassis.

For workers, this is the canary. When humanoids hit commodity pricing, the economic case for human labor in physical tasks compresses fast. The transition window is shorter than most people think.

Source: The Information