Skip to content

Documentation / Market

02 / 22

Physical AI is becoming an industry

Talent and capital are moving from software models into machines that can act in the world.

AI is moving out of software and into machines that can see, decide and act, drawing engineers from frontier model labs, autonomous driving programmes and long-running robotics research groups. Capital is following that shift, but the supporting stack remains fragmented.

Context / Talent migrationElite AI talent is moving into physical AI at an accelerating pace
  • Tesla2003PublicHumanoid / OptimusElon Musk, Milan KovacPayPal · SpaceX
  • 2014$125M+Home humanoidBernt BørnichUniv. of Oslo · Halodi
  • 2016~$1.0BIndustrial humanoidJeff Cardenas, Nick PaineUT Austin · NASA
  • 2022$1.7B+Humanoid platformBrett AdcockArcher · Vettery
  • 2023$1.7B+Robot foundation modelDeepak Pathak, Abhinav GuptaMeta AI · CMU
  • Physical Intelligence2024$1.1BRobot foundation modelKarol Hausman, Chelsea Finn, Sergey LevineGoogle · Stanford · Berkeley
  • 2024$300M+Home robotsKyle Vogt, Paril Jain, Luke HoloubekCruise · Twitch · Tesla
  • 2025$1.03BWorld models / physical AIYann LeCun, Alexandre LeBrunMeta FAIR · Nabla
  • 2026$1.7BIndustrial roboticsTravis KalanickUber · CloudKitchens

Every team still has to assemble its own data, evaluation, hardware and deployment loop. That repeated work slows the field even as model and hardware capability improve.