Robots Train in Factories First

The winners in home robotics are likely to be built in factories and military programs first, because that is where robots can earn money while still needing human help. In a warehouse or on an assembly line, teleoperators can step in when the robot gets stuck, the customer still gets usable work, and every intervention becomes training data. That makes B2B and government the practical on ramp to the eventual consumer product, not a side market.

• Industrial sites are easier training grounds than homes. Figure’s BMW deployment ran daily shifts, loaded more than 90,000 parts, and contributed to over 30,000 X3 vehicles, which shows why factories are valuable, repeatable places to harden robots before sending them into messy kitchens and bedrooms.
  2 figure 6 bmwgroup
• Government and industrial buyers tolerate supervised rollouts that consumers will not. Foundation has explicitly focused on logistics, manufacturing, defense, and security first, and described teleoperation as essential because a human can keep the line moving when the model mispredicts, while the company captures those edge cases for retraining.
  3 sacra 4 sacra
• Consumer ambition is still visible underneath the B2B sequencing. Figure partnered with Brookfield to access 100,000 residential units for home environment data, and 1X opened NEO pre orders with first U.S. deliveries starting in 2026, showing that enterprise deployments fund and train the system while home products remain the long game.
  5 figure 7 1x 8 1x

Over the next few years, the category is set to split into two stages. First, robots become dependable enough to earn their keep in factories, warehouses, and defense settings. Then the companies with the biggest deployment base, the most intervention data, and the fastest manufacturing lines will be the ones best positioned to ship a consumer robot that feels more like an appliance than a science project.