Wired: Yann LeCun AMI Startup Raises $1 Billion for AI World Models

AMI's $3.5 Billion Valuation and European Base

The funding round, which values AMI at $3.5 billion, was co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital, and Bezos Expeditions. Notable backers include Mark Cuban, former Google CEO Eric Schmidt, and French telecommunications executive Xavier Niel.

AMI will operate globally with offices in Paris, Montreal, Singapore, and New York, where LeCun continues his role as a New York University professor while leading the startup. The company's European headquarters in Paris positions it within the EU's AI regulatory framework, which could influence how AMI approaches model development and enterprise partnerships.

Technical Challenge to Language Model Scaling

LeCun argues that extending large language models to achieve human-level intelligence is "complete nonsense," instead advocating for AI systems that understand physical environments through world models. AMI aims to build systems with persistent memory, reasoning capabilities, and controllable safety features.

The startup plans to target manufacturing, biomedical, and robotics companies with significant data resources. LeCun cited aircraft engine optimization as a potential use case, where AMI could create realistic world models to help manufacturers improve efficiency and reliability.

Industry Positioning and Open Source Strategy

AMI's cofounders include former Meta executives Michael Rabbat (director of research science), Laurent Solly (VP of Europe), and Pascale Fung (senior director of AI research). Alexandre LeBrun, former Nabla CEO, serves as AMI's CEO, while former Google DeepMind researcher Saining Xie leads as chief science officer.

The startup plans to develop open source technology, with LeCun arguing that AI is too powerful for any single private company to control. This approach contrasts with the increasingly closed development models at major AI labs and aligns with European preferences for transparent, auditable AI systems.

Enterprise Applications and Partnership Strategy

LeCun left Meta in November 2024 after concluding that world model research would progress faster outside the social media company's consumer-focused business model. He plans to work with enterprise partners including Toyota and Samsung before developing what he calls a "universal world model" for general intelligence applications.

While Meta is not an investor in AMI, LeCun indicated potential collaboration opportunities, including using AMI's world models to power assistants in Meta's smart glasses products.

The AMI launch signals a significant bet against the scaling hypothesis that drives current AI investment, with implications for European enterprises evaluating between language-focused and physically-grounded AI approaches. LeCun's track record as a Turing Award winner and AI pioneer gives weight to his criticism of language model limitations, though AMI must still prove that world models can deliver practical enterprise value at scale.

Original source: Wired reported on Yann LeCun's $1 billion AMI startup launch and its focus on AI world models.