Alexandre LeBrun, CEO of AMI Labs, the world-model startup co-founded with Turing Award winner Yann LeCun, has a simple reason for refusing to describe his company’s technology as AGI or superintelligence: he doesn’t think the words mean anything.
“We never used the word AGI. And I just noticed that nobody is using it anymore; they switched to superintelligence. Next time we’ll switch to something else,” LeBrun said in an interview. “There’s no good definition. What is superintelligence? I don’t know. It’s not a very useful word.”
AMI Labs raised US$1.03 billion (approximately £790 million) in March 2026 at a US$3.5 billion pre-money valuation, making it one of the best-funded AI startups without a publicly released product. The company focuses on what LeBrun calls “world models,” AI systems that predict the next physical state of the world rather than the next token in a sequence.
“Large language models predict the next word or text; a world model predicts the next state,” LeBrun explained. “Nudge a glass off the table, and you already know it will tip and spill, that’s the intuition a world model is meant to capture.”
The world-model approach is complementary to LLMs, not a replacement. LeBrun sees LLMs handling language and reasoning while world models provide the physical intuition needed for robotics, manufacturing, and real-world interaction. He describes today’s AI systems as “a doctor trained only on textbooks and without a residency,” noting that LLMs cover “only 1% of healthcare.”
On robotics, LeBrun is blunt about the gap between hardware progress and software intelligence. “The hardware is very advanced; progress in hardware in the last few months is incredible, but there’s no brain,” he said. “Robots are not safe right now. There’s no solution for that today.”
That safety gap is central to AMI Labs’ mission. While humanoid robots have made dramatic strides in dexterity and locomotion, exemplified by Figure’s humanoids in EV assembly and recent world-first robot surgeries on live pigs, LeBrun argues that only world models can give robots the context awareness needed to operate safely around humans. A dancing robot that kicked a child, he noted, would not have happened with a system that understood physical consequences.
AMI Labs is targeting South Korea as a strategic base for industrial partnerships, drawn by its advanced hardware ecosystem in robotics, semiconductors, and manufacturing. JP Lee, CEO of SBVA (AMI Labs’ Asian backer), said: “I’ve been telling Alex and the team to come to Korea. The government has done a tremendous job funding local sovereign LLM models, but they should coexist with physical AI.”
LeBrun was in Seoul for the ICML conference scouting local industrial partners. No product or launch timeline has been announced. “We’ll make a surprise when we’re ready,” he said.
Sources: TechCrunch