AI startup AMI Labs, valued at $3.5 billion, is scouting for industrial partners in South Korea to advance robotics. The company aims to move beyond language models by developing 'world models' that help robots safely interact with physical environments.
AMI Labs, the AI startup co-founded by industry veteran Yann LeCun, is shifting its focus toward physical-world applications. CEO Alexandre LeBrun recently visited Seoul, South Korea, to engage with potential partners in the robotics, semiconductor, and manufacturing sectors. The startup, which is currently in its pre-product stage, is prioritizing the development of 'world models'—a technology designed to help AI systems predict and understand physical environments rather than just processing text.
Moving Beyond Language Models
While the current market enthusiasm centers on Large Language Models (LLMs) that predict text, LeBrun argues these tools struggle when applied to physical tasks. He noted that current robotic systems are often limited to static, repetitive routines and lack an intuitive understanding of their surroundings. AMI Labs aims to bridge this gap by creating AI that can anticipate physical consequences, such as understanding gravity or spatial awareness, to improve the safety and effectiveness of robots in real-world settings like homes and factories.
Strategic Expansion into Asia
AMI Labs' search for partners in South Korea is driven by the country's strong industrial base and government-backed initiatives for AI and robotics. The company requires access to real-world environments to train its models, as LeBrun compares AI trained only on data to a doctor who has read books but lacks clinical practice. South Korea's existing infrastructure in chip manufacturing and data center development makes it a logical target for AMI Labs as it attempts to bring its technology out of laboratory settings.
Financial Context and Investor Expectations
Investors are closely monitoring the company's progress, given its high valuation. In March 2026, AMI Labs raised $1.03 billion, reaching a pre-money valuation of $3.5 billion. Despite this capital infusion, the company has not yet launched a commercial product. The startup faces the challenge of translating significant funding into a functional, safe, and scalable technology for industrial use. For shareholders and investors in the broader AI space, the timeline for a product release and the success of these industrial partnerships will be key monitorables in the coming quarters.
While the company has secured backing from firms like SBVA, the transition from theory to physical reality involves significant execution risks. The ability to create robots that can operate safely in open, unpredictable environments remains an unresolved problem across the global AI industry. As AMI Labs continues its development, its ability to prove its 'world models' in manufacturing environments will determine whether it can justify its current private valuation and compete in the evolving robotics market.