LeCun left Meta, started AMI Labs, and is betting world models beat LLMs for real AI. Here's what that actually means, what the research shows, and why it matters for where AI tooling goes next. Quick context if you haven't been following: Yann LeCun is one of the three "godfathers of deep learning" (the Turing Award crew alongside Hinton and Bengio), spent 12 years running Meta's AI research lab FAIR, and has been publicly, loudly skeptical of LLMs basically the entire time they became the dominant paradigm. Think of him as the guy in your Discord who keeps saying "yeah but have you actually read the architecture paper" -- except he's usually right, and now he's raised a billion dollars. In November 2025 he left Meta. By March 2026, his new lab AMI Labs (pronounced ah-mee , French for friend, cute) closed a $1.03B seed round at a $3.5B valuation . Largest seed round in European startup history. Backers include Bezos Expeditions, NVIDIA, Samsung, Toyota Ventures, and Tim Berners-Lee personally.…