Loading
Loading
The market spans platform owners, model builders, low-cost hardware players, and companies already learning in real deployment.
Market context
Capital is flowing into humanoids, robot foundation models, and autonomy stacks. What matters now is not whether the money is real, but which companies can turn narrative into deployment.
Platform layer (GR00T, Isaac, Omniverse, Jetson Thor)
Trying to attach models, simulation, digital twins, and edge compute to existing industrial channels through partners like ABB, FANUC, and KUKA.
General-purpose humanoid (Figure 02)
Factory deployment is the near-term test. The question is whether pilot work turns into reliable, repeatable commercial operation.
Humanoid robotics
The home narrative is far ahead of current proof. Factory use remains the more credible milestone to watch first.
Robot foundation models (pi0)
A bet that one model family can generalize across many robot bodies and tasks. The open question is how much embodiment-specific data is still required.
Autonomous vehicles
Still the clearest example of physical AI working at scale in the real world, with all the operational complexity that comes with it.
Low-cost humanoid (G1, H1)
Chinese supply-chain speed and pricing pressure matter because they push the hardware cost floor lower for the whole category.
Warehouse humanoid (Digit)
A narrower logistics focus in environments structured enough to learn from and measure clearly.
Consumer humanoid (NEO)
The home remains the hardest operating environment: emotionally compelling, operationally messy, and difficult to service.
Advanced locomotion (Atlas, Spot)
Years ahead on physical capability and field learning. The commercial question is how much of that lead converts into repeatable deployments.
Gemini Robotics, RT-2
Strong research output around VLA and spatial reasoning. The product question is how quickly that turns into deployed systems.