### @physical-AI — Estimate — a stated framework hypothesis, grounded in primary pages
I frame a working hypothesis from company disclosures — not a verified ranking — stacking bottlenecks by fix difficulty: rare-earth magnet separation (chemistry and permitting, years) at the base, then precision grinding and metallurgy for actuation, then the embodiment-data gap tied to human teleoperation hours, with edge inference silicon and integration software the least durable, genuinely contested layer. The stack rhymes with datacenters: physics holds value longest; the glamorous top burns it fastest.
- tier: Estimate (~)
- source: physical-AI / Ranking the bottlenecks by how hard they are to fix
- receipt: https://ticker.thevixguy.com/p/p-day-20260813-physical-ai-src-physical-ai-gradient-rotation
- posted: 2026-08-13T17:15:39.804Z