China leads humanoid deployment
Sankaet Pathak, CEO of Foundation, on why humanoids win in robotics
The real signal is that China is no longer proving humanoids in demos, it is proving them in factories and on production lines. Shipping 1,000 plus units means Chinese players have crossed from lab prototypes into repeatable manufacturing, field support, and customer workflows. That matters because the winner in humanoids is likely the company that learns fastest from deployed robots, and China has a clear head start in collecting that real world operating data while U.S. companies are still mostly in pilot mode.
-
The advantage is not just lower robot prices. Chinese companies sit close to a dense component base, with Asia controlling 63% of key humanoid parts and enabling robots built about 65% cheaper than non Chinese supply chains. That makes it easier to iterate hardware, replace failed parts, and scale output fast.
-
For U.S. startups, the immediate race is less about importing Chinese robots and more about matching the learning loop. Foundation describes the core battle as getting robots into factories, using teleoperation when models fail, labeling those interventions, and feeding that data back into better action models. More shipped robots means more edge cases captured every day.
-
The leading U.S. companies are still centered on narrow industrial jobs because those are the easiest places to prove ROI. Figure, Apptronik, Agility, and Foundation are targeting factories and warehouses where a humanoid can step into existing aisles and workstations without a 12 to 18 month facility retrofit. China reaching volume first suggests that practical deployment, not polished demos, is setting the pace.
The next phase is a scale fight between Chinese manufacturing volume and U.S. software and customer integration. If Chinese companies keep shipping in the thousands while American firms deepen footholds in domestic industrial and defense settings, the market will split into two lanes, one led by low cost hardware volume, the other by protected, high value deployments where data quality and reliability matter most.