Physical Intelligence's new robot brain, π0.7, demonstrated compositional generalization by cooking...
The AMW Read
Updates the robotics player map with a demonstration of scaling laws (cross.§B) and mentions a prospective $11B valuation (cross.§D) that would trigger high-tier capital cycle significance.
NoveltySignificance
Robotics · Player MapScaling LawsCapital Cycles
Physical Intelligence's new robot brain, π0.7, demonstrated compositional generalization by cooking a sweet potato in an air‑fryer it had seen only twice in its 903 M‑step training data. With a simple verbal walkthrough, task success jumped from 5% to 95%, showing real‑time coaching can replace costly retraining. The model matched specialist systems on coffee‑making, laundry‑folding and box‑assembly, hinting at a scaling law similar to LLMs. If this trajectory holds, robot deployment could accelerate before the next funding round pushes the company’s valuation toward $11 B.
