Advanced Machine Intelligence (AMI) has raised $1.03 billion to build and commercialize an alternative artificial intelligence model. The startup said the round came at a $3.50 billion pre-money valuation.
The financing puts fresh attention on Yann LeCun’s long-standing view that current AI methods still face core limits. LeCun founded AMI after leaving Meta at the end of 2025. He joined Meta in 2013 and created Facebook AI Research, later known as FAIR.
At Meta, he became one of the company’s most influential AI leaders. His new company now aims to build systems centered on reasoning, planning, and world models.
The funding round was co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital, and Bezos Expeditions. Other reported backers include NVIDIA, Eric Schmidt, and Tim Berners-Lee. The size of the round makes AMI one of the most closely watched new AI startups in 2026.
AMI enters the market as most large AI companies continue to invest heavily in large language models. LeCun has argued for years that next-word prediction alone cannot deliver broad human-level intelligence. He says systems also need stronger reasoning, memory, planning, and understanding of the physical world.
That position now sits at the center of AMI’s research and business strategy. The company said generative architectures trained through self-supervised learning have achieved strong language results. However, AMI argues that those approaches struggle with much real-world sensor data because much of that data remains unpredictable.
AMI says its world models will learn abstract representations of real-world data. The goal is to ignore noisy details while preserving the information needed for prediction and action. In practical terms, AMI wants AI systems that can anticipate outcomes and plan safe action sequences.
In an interview with Reuters, LeCun said AMI wants to build systems for complex real-world settings. He said current approaches based on predicting words or pixels will not produce broadly capable intelligent agents by themselves. That statement explains why AMI is focusing on a different technical path.
The company’s approach aims at environments where reliability matters more than fluent conversation alone. AMI expects early demand from organizations that run complex systems and need clearer control. Those customers include manufacturers, automakers, aerospace companies, biomedical firms, and pharmaceutical groups.
That customer list reflects a practical near-term commercial plan. Industrial and scientific settings often require models that can interpret changing conditions and act within constraints. AMI appears to see those sectors as a better fit than general consumer chat products. That last point is an inference based on the company’s named target markets.
Also Read: Zuckerberg's Meta Plans Massive AI Spending Push to Build ‘Personal Superintelligence’
LeCun said the same technology could later support consumer products. He highlighted domestic robots as one example because such machines need common sense and physical understanding. He also said AMI has discussed possible deployment with Meta for Ray-Ban Meta smart glasses.
Those comments suggest AMI wants both enterprise and consumer relevance over time. Still, the company has not announced a public product launch or revenue timeline. Reports also describe AMI as a research-heavy venture with no product and no revenue yet.
The broader significance of the round extends beyond one startup. AMI now serves as a test of whether investors will back AI systems that move beyond mainstream language-first architectures. For now, the company has significant capital, a defined thesis, and a clear list of early target industries.