Technopolis

Tuesday, July 21, 2026

AI & Models

AMI Labs targets South Korea to bring physical AI out of the lab

Yann LeCun's AMI Labs is seeking hardware partnerships in South Korea to develop physical AI, while rejecting industry buzzwords like AGI.

AMI Labs, the AI startup co-founded by Yann LeCun after he left Meta, raised $1.03 billion in March at a $3.5 billion pre-money valuation. CEO Alexandre LeBrun spoke to TechCrunch while in Seoul last week for The International Conference on Machine Learning, where he was scouting for local industrial partners, global companies and researchers to help move the company’s world models out of the lab. The startup remains pre-product, with no timeline set; LeBrun said only that AMI would offer a surprise once it is ready.

LeBrun avoids terms like AGI (Artificial General Intelligence) or superintelligence when describing AMI’s technology, arguing both are poorly defined. “We never used the word AGI. And I just noticed that nobody is using it anymore; they switched to superintelligence,” he said. He argues AI remains highly limited and unintelligent in the physical world, even as hardware advances race ahead of software understanding. A world model incorporates physics to predict the next state of the physical world, unlike a large language model (LLM), which predicts the next word; LeBrun frames the two as complementary rather than replaceable. World models are expected to have a large impact on robotics, and almost every industry that touches the real world could eventually utilize robotics built on them, he said. But robots are not safe right now, and no existing solution addresses this — LeBrun pointed to an incident in which a robot at a public event ended up approaching and kicking a child. Drawing on his time at Nabla, the AI health startup he previously founded, LeBrun said large language models cover only 1% of healthcare needs, with the rest depending on real-world experience.

AMI’s interest in Korea is backed by SBVA (a venture capital firm and one of AMI’s backers in Asia), whose CEO JP Lee has pushed the company to expand there. LeBrun pointed to South Korea’s advanced hardware-heavy sectors, which the first wave of AI barely touched, and to the country’s history as an early adopter, noting Korea was the fastest adopter of the internet 25 years ago. Lee highlighted Seoul’s June plan to mobilize some $880 billion for chips, AI data centers and physical AI, and cited homegrown internet players like Naver and Kakao as evidence of how quickly Korean developers adopt and adapt new tools.

Why it matters

AMI Labs, the world model startup co-founded by Yann LeCun, is rejecting popular industry buzzwords like AGI and superintelligence while seeking hardware and robotics partnerships in South Korea to move its physical AI models out of the lab, a bet that almost every industry touching the real world could eventually rely on such robotics.