After 20+ robotics startup interviews a week, a clear pattern: data, neointegrators, world models, hardware
lukas_m_ziegler · x · 2026-09-10
The author, who interviews 20+ emerging robotics companies every week, found 80% of them fall into one of four buckets:
- Data collection (of various types)
- Vertically specific deployment models ("neointegrators")
- World models — software brains for robots
- Components manufacturing — scaling hardware production
He poses the open question he first raised four months ago: what are the real moats in physical AI today?
More from Venture
- AI valuations look like peak crypto; lab revenue still two orders of magnitude from trillions — allTheYud · 2026-09-10
- Lattice's 3rd employee breaks down the playbook that hit $50M ARR with 75% inbound — thedealdirector · 2026-09-10
- Lightfield raises $47M Series A led by a16z to rebuild CRM as a world model for agents — socialwithaayan · 2026-09-10
- 2026 Google Ranking Factors survey: 131 SEO experts score 13,000 data points, AI factors added — VeryWellVersed · 2026-09-10
- Type launches as a shared AI workspace for teams, raises $4M from Lerer Hippeau and others — Scobleizer · 2026-09-10
- Salesforce in talks to buy AI customer research startup Listen Labs for around $2B — ditzikow · 2026-09-10