Meta Vet Yann LeCun’s AI Startup Pulls in $1 Billion

The company announced the new funding Tuesday (March 10), saying it would help it build artificial intelligence (AI) systems that “understand the world, have persistent memory, can reason and plan, and are controllable and safe.”

AMI (Advanced Machine Intelligence) was founded by Yann LeCun, who stepped down from Meta late last year.

According to its website, AMI is developing what it calls “world models” that “learn abstract representations of real-world sensor data, ignoring unpredictable details, and that make predictions in representation space.”

“Action-conditioned world models allow agentic systems to predict the consequences of their actions, and to plan action sequences to accomplish a task, subject to safety guardrails,” the company added.

In an interview with Reuters, LeCun said the company’s target customers are organizations running complex systems, such those in the automotive, aerospace, biomedical, manufacturing and pharmaceutical sectors. In time, AMI’s tech could find its way to consumers, he added.

“What consumers could be interacting with is a domestic robot,” LeCun said. “You need a domestic robot to have some level of common sense to really understand the physical world.”

As PYMNTS wrote last year, LeCun is among a growing number of AI scientists who have shifted into entrepreneurial roles, along with former OpenAI Chief Technology Officer Mira Murati (founder of Thinking Machines) and Stanford professor Fei-Fei Li, who co-founded World Labs to advance spatial intelligence AI and achieved unicorn status soon after its debut.

“These researcher-founders share a conviction that AI’s next breakthroughs won’t come from scaling transformers but from systems capable of perceiving causality, spatial relationships and physical context,” the report said. “Their companies, many built on open-research principles, are attracting record funding rounds as investors shift capital toward scientist-led innovation.”

Meanwhile, Meta is reportedly building a new applied AI engineering organization that will work with the company’s Superintelligence Lab to make AI models faster

The new organization will be headed by Maher Saba, who is a vice president in Meta’s Reality Labs division, per a report last week by the Wall Street Journal (WSJ).

“Building great models isn’t just about researchers and compute; it requires real-world data, feedback and evals,” Saba wrote in a memo to employees, per the WSJ. “This creates the flywheel that turns a strong model into a leading one. Lately, we’ve seen some excellent gains from reinforcement learning and post-training and we believe we have a real opportunity to move faster and pull ahead if we double down on these efforts.”