Yasho Singh

Notes on the physical bottlenecks of AI: compute, power, and the supply chains beneath them.

I’m a startup operator in San Francisco. I started on the floor of an industrial transformer factory at sixteen, sourced GPUs for Nuro’s autonomous vehicles through the 2022 chip shortage, and spent three years investing in early stage startups at Catalyst, where I led six deals and helped build two companies from scratch.

The Series

No. 1
Silicon, Power, Steel
Why the AI bottleneck keeps sinking deeper
YS
Yasho Singh
NO. 1 · NOVEMBER 2026
Silicon, Power, Steel
Each time the AI bottleneck moves, it lands somewhere money can’t fix as quickly.
No. 2
In preparation
NO. 2 · FORTHCOMING
The next constraint, in progress.

About

I studied Industrial Engineering at Georgia Tech because of a childhood question: how did ancient armies run expeditions and supply chains without maps? I’ve been chasing versions of that question since, from factory floors to autonomous vehicles to venture capital.

These days I write about the infrastructure under AI. Each piece in the series is a long, careful attempt to understand one constraint.

  • Catalyst by WellstarINVESTOR · 2023–26
  • NuroSUPPLY CHAIN · 2022
  • Honeywell · CumminsENGINEERING
  • Georgia TechINDUSTRIAL ENG.

Contact

If you’re building in compute, inference, or physical AI, I’d love to chat.

hello@yashosingh.comX · @SinghYashorajLinkedIn