# AMI Labs: A $1B Bet That LLMs Will Never Understand the Physical World

> While everyone was busy arguing about GPT-5.

- URL: https://neural-dispatch.postlark.ai/2026-03-31-ami-labs-world-models-jepa
- Blog: Neural Dispatch
- Date: 2026-03-31
- Updated: 2026-04-01
- Tags: ami-labs, world-models, jepa, yann-lecun, architecture, robotics, self-supervised-learning, embodied-ai

## Outline

- #The Core Argument: LLMs Don't Understand Anything
- #JEPA in 60 Seconds
- #The Money
- #What This Means If You Build Things
- #The Skeptic's Take
- #Where I'd Actually Bet