Skild AI hits $100M ARR in 10 months, explains robotics data trade-offs
deepakpathak · x · 2026-09-11
Robotics foundation model company Skild AI announced $100M ARR within 10 months of starting deployments, now working on factory lines, construction sites, in kitchens and data centers — cleaning, welding, building and cooking. Co-founder Deepak Pathak shared his views at AUTONOMOUS:
- Robotics has been treated as a hardware problem for 70 years, but the missing piece is a general brain; the LLM recipe of more data and bigger models doesn't transfer because robotics has no internet to train on.
- Robotics data quality has three axes: scalability, diversity, and proximity to the robot.
- Four data sources each fail on their own: robot self-collected data is best-targeted but scales slowest (physics runs in real time); teleoperation is the industry default with almost no diversity; simulation is fast but every task needs a hand-built environment; human video is abundant but furthest from the robot (different body, forces invisible).
- There is no single "golden path" — Skild trains one brain across every form factor using all four sources combined.
Related event: Skild AI Hits $100M ARR Just 10 Months After First Deployment(36 posts)→
More from Companies & People
- Anthropic files S-1 with SEC, formally kicking off its IPO process — minchoi · 2026-10-26
- Silicon Valley breakdown: five Stanford-area giants worth ~$18T — FinanceYF5 · 2026-10-26
- User suspects X's bundled Grok subscription boosted tweet reach amid algorithm change rumors — talkaboutdesign · 2026-10-26
- "RIP Claude + Chrome": developer flags the end of Anthropic's browser integration — damienghader · 2026-10-26
- a16z podcast: Diode pairs AI PCB design with Brooklyn factory to revive US hardware manufacturing — MattPerault · 2026-10-26
- Vercel Ships 12 New Products at Its 2026 Conference as It Races Rivals for the Agent Entry Point — FinanceYF5 · 2026-10-26