Advanced Machine Intelligence Labs (AMI Labs), the AI startup founded by New York University professor Yann LeCun, who served as chief AI scientist at Facebook parent Meta, has secured more than $1 billion (about 1.47 trillion won) in investment.
AMI Labs said on the 10th (local time) that it raised $1.03 billion to develop and commercialize a next-generation AI model called a "world model" that can understand, reason about, and plan in the physical world.
AMI Labs' pre-money valuation was pegged at $3.5 billion (about 5.15 trillion won). The first funding round included Bezos Expeditions, the investment firm of Amazon founder Jeff Bezos, France's Cathay Innovation, the United Kingdom's Hiro Capital, Singapore sovereign wealth funds Temasek, Korea's SBVA (formerly SoftBank Ventures), and Nvidia, the Financial Times (FT) reported.
According to data analytics platform Dealroom, AMI Labs' fundraising is the second-largest after U.S. AI startup Thinking Machines Lab, which raised $2 billion last year.
AMI Labs is launching with offices in Paris, New York, Singapore, and Montreal. Alexandre Lebrun, founder of French Healthtech startup Nabla, will lead the company as chief executive officer (CEO), and LeCun will serve as chair of the board. Laurent Solly, who served as Meta's vice president for Europe, will join as chief operating officer (COO).
LeCun, regarded as one of the world's four leading AI scholars, said last year that within 3 to 5 years, the large language models (LLMs) that underpin the AI we use will become outdated and "world models" that understand the physical world will become mainstream.
LeCun has pointed out the limits of LLMs. He argued that LLMs, which are driving the current Generative AI boom, predict the next word and therefore cannot reach human-level intelligence. In other words, they do not properly understand the real world. As an alternative to LLMs, he proposed "world models," in which AI, like people, directly observes the world to understand and predict the physical laws of reality.
The "world model" that LeCun has researched since Meta is expected to be used in fields such as robots, Autonomous Driving cars, manufacturing, and transportation.