
Deepak Pathak
Co-founder and CEO · Skild AI
Joint tenth: dedicated action models offer strong humanoid options; transfer and commercial claims remain material dependencies.
Analysis
Established position
Skild identifies Pathak as co-founder/CEO. NVIDIA’s partner case describes Skild training on humanoids and testing humanoid locomotion and manipulation; these are interested-partner reports, not independent validation. Skild’s financing and operations supply resources, but warehouse revenue is not humanoid revenue. Skild AI — 2026-01-14; NVIDIA — undated/current record; NVIDIA — 2026-09-10; Skild AI — 2026-09-10; Skild AI — 2026-04-15
Analysis
Strategic thesis
Pathak can supply reusable robot-action intelligence for humanoid programmes, giving Skild substantial options if its cross-body learning transfers into humanoid manipulation and whole-body systems while the company retains useful data rights and dependable operational integration. Skild’s action policies address physical manipulation rather than only language or driving. Humanoid use requires body-specific control, balance and hand adaptation; warehouse deployments are not relabelled as humanoid deployments.
Analysis
Explicit assumption
Skild’s transferable robot-action layer can integrate with humanoid bodies, with technically useful and contractually permitted learning across embodiments.
Analysis
Strongest counter-case
High reported growth can mask pilot economics, services or customer concentration. Acquiring hardware operations complicates a universal-model thesis, and a partner account is not independent confirmation of broad autonomy. Confidence: Commercial and generalisation premises rely heavily on vendor/partner reporting; revenue quality and integration could change the placement.
Analysis
Forward scenarios through 2030
Base: Skild’s robot-action research supports selected humanoid integrations with substantial body-specific engineering. Upside: Transferable manipulation becomes a widely chosen humanoid intelligence layer. Downside: Services-heavy revenue and integration complexity weaken the universal platform thesis.
Analysis
What to monitor
- 01Named humanoid integrations and repeated useful manipulation.
- 02Cross-body learning gains with explicit data rights.
- 03Operational integration and economics of acquired robotics activities.
Analysis
Research boundaries
Reported financing; current cash and contractual obligations undisclosed. Founder CEO with cofounder president and financing governance. The full capital, talent, compute, data, manufacturing, distribution, architecture, integration, dependence and governance record is structured in the snapshot.
Skild Brain S1: company reported cross embodiment action model.
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