
Karol Hausman
Co-founder and CEO · Physical Intelligence
Joint tenth: reusable manipulation research can influence humanoid brains, but it is not a demonstrated complete humanoid controller.
Analysis
Established position
The Generalist’s March article identifies Hausman as co-founder and CEO. PI’s April π0.7 material reports generalisation progress; February partner accounts discuss commercial workflows and continued human interventions. Those accounts are evidence of claimed use, not independent certification of general autonomy. The Generalist — 2026-03-17; Physical Intelligence — 2026-04-16; Physical Intelligence / Weave / Ultra — 2026-02-24
Analysis
Strategic thesis
Hausman can develop the reusable manipulation intelligence needed by humanoids through dedicated robot-action models, creating strategic options across manufacturers while requiring separate integration for balance, locomotion, dexterous hands and reliable operation in each humanoid body. PI’s policies already map perception and instructions into robot action. Extending manipulation to humanoids is a concrete technical transfer thesis; the reviewed bimanual and mobile-arm results do not establish a complete humanoid controller.
Analysis
Explicit assumption
PI’s manipulation policies can be incorporated into humanoid systems with manageable hand, balance and whole-body adaptation costs.
Analysis
Strongest counter-case
Commercial pressure can force specialisation and dilute general learning. OEMs or larger platforms may retain data and distribution, capturing more power than a technically impressive model supplier. Confidence: The robot-action research is identifiable, but humanoid hand/whole-body integration, adoption and associated data rights are pivotal to this narrower power thesis.
Analysis
Forward scenarios through 2030
Base: PI’s action research informs selected humanoid manipulation systems with specialist integration. Upside: Reusable manipulation policies become a preferred intelligence component across humanoid manufacturers. Downside: Specialisation and partner dependence prevent broad strategic capture.
Analysis
What to monitor
- 01Humanoid hand and whole-body integrations with disclosed intervention costs.
- 02Customer retention and operational responsibility.
- 03Data sharing and model portability across embodiments.
Analysis
Research boundaries
Private financing; available cash and robot budget undisclosed. Cofounder and investor governance; shared scientific leadership. The full capital, talent, compute, data, manufacturing, distribution, architecture, integration, dependence and governance record is structured in the snapshot.
Physical Intelligence π 0.7: research release; Physical Intelligence π 0.6: partner workflows.
Sources were accessed on September 13, 2026. Undated profiles establish only the current reviewed record; future product, capacity or delivery statements remain plans. Exact budgets, data rights and private delegation are not assumed from the title.
Analysis