$50M · Industrial AI + P-1 AI

What they're building

P-1 AI builds Archie, an AI 'engineer' for the physical world — currently at roughly the level of a junior mechanical and electrical engineer. Archie fits into existing engineering teams and uses the same complex tools human engineers use (CAD, schematics, simulation), handling multi-physics and spatial reasoning: distilling design drivers from requirements, generating product concepts, and doing first-order analysis. The tech stack is a custom agentic harness, a structured design representation, continual skills learning, and small custom post-trained models (SFT and RLVR) trained on proprietary semi-synthetic engineering datasets. They start in data-center cooling and critical power, then expand to automotive and aerospace/defense. A lightweight version, Archie Solo, is in private preview.

Why this matters

NEA leading, plus adding Jeff Immelt (ex-GE chairman/CEO) to the board and Lila Tretikov (ex-Microsoft deputy CTO) as observer, signals this is being built for real industrial buyers, not a demo. The angel list is elite AI (Jeff Dean, Peter Welinder). CEO Paul Eremenko says interest from major industrial firms 'has been extraordinary.' The macro thesis, per Eremenko's Jan 14, 2026 Fortune commentary: 'China graduates 1.3 million engineers per year, versus just 130,000 in the U.S. We need AI to bridge the gap.'