Yann LeCun’s AMI Labs Raises $1.03 Billion to Build “World Model” AI

A new artificial intelligence research company led by one of the field’s most influential scientists has secured more than $1 billion in funding to pursue a different vision for the future of AI.

The startup AMI Labs, co-founded by renowned AI researcher Yann LeCun, has raised approximately $1.03 billion at a $3.5 billion pre-money valuation. The company plans to focus on developing “world model” AI systems that aim to understand real-world environments rather than simply generate text.

The investment round was reported by TechCrunch and reflects growing interest in alternative AI architectures beyond large language models.

A Different Vision of Artificial Intelligence

AMI Labs is built around a concept that LeCun has advocated for several years: AI systems that learn from multimodal real-world data rather than relying primarily on text.

These systems, often referred to as world models, attempt to represent how the physical world works and predict how environments change over time.

Unlike traditional large language models, which learn patterns from massive text datasets, world models are designed to integrate information from many types of input such as images, video, audio, and sensor data.

Supporters of this approach believe it could help overcome limitations commonly seen in current AI systems, including hallucinations and weak reasoning about real-world events.

The Technology Behind the Project

AMI Labs’ research direction is based on Joint Embedding Predictive Architecture, often referred to as JEPA.

First introduced in 2022, JEPA is designed to help AI systems learn abstract representations of the world by predicting relationships between different pieces of data.

Rather than generating outputs word by word like many language models, JEPA-based systems attempt to build internal models of how environments function and use those models to make predictions.

Researchers working on the architecture believe this approach could enable more reliable reasoning and more efficient learning compared with current generative AI methods.

Major Investors Back the Project

AMI Labs’ funding round attracted a wide range of investors from both the technology and venture capital sectors.

The round was co-led by several investment firms, including:

• Cathay Innovation
• Greycroft
• Hiro Capital
• HV Capital
• Bezos Expeditions

Additional participants include technology companies and global investment groups such as:

• NVIDIA
• Samsung
• Temasek
• Toyota Ventures

Several well-known technology figures also joined the round as individual investors, including web inventor Tim Berners‑Lee, entrepreneur Mark Cuban, venture capitalist Jim Breyer, and former Google CEO Eric Schmidt.

A Research-Focused AI Startup

Unlike many AI startups racing to release commercial products, AMI Labs plans to operate more like a research institution.

CEO Alexandre LeBrun has emphasized that the company is not trying to launch a product immediately or generate rapid revenue.

Instead, the focus will be on long-term foundational research aimed at developing new AI architectures.

This means the company may spend years experimenting with world models before releasing commercially viable applications.

The strategy reflects LeCun’s long-standing belief that the current generation of language-based AI models is only an early step toward more advanced systems capable of understanding the world.

Global Research Hubs

AMI Labs plans to operate across several international research centers.

The company’s headquarters will be located in Paris, while additional research hubs are planned in:

• New York City
• Montreal
• Singapore

This distributed structure is intended to help the company recruit top AI researchers from major academic and technology centers around the world.

LeCun himself continues to teach at New York University, which will help connect the startup with academic research communities.

Healthcare as an Early Testing Ground

Although the company does not plan to launch products immediately, it has already identified potential applications for its research.

One early collaboration involves Nabla, where LeBrun serves as chairman.

Healthcare presents a particularly important testing environment for AI reliability, since hallucinated or incorrect outputs from AI systems can have serious consequences in clinical settings.

World models could potentially help reduce such errors by enabling AI systems to reason more accurately about real-world data.

Open Research Strategy

In line with LeCun’s long-standing advocacy for open science, AMI Labs plans to publish research papers and release portions of its code publicly.

The company hopes that sharing its work will encourage broader collaboration across the global AI research community.

LeBrun predicts that “world models” could become the next major focus in AI research, though he also expects many startups to adopt the term primarily as a fundraising strategy.

AMI Labs aims to differentiate itself by concentrating on fundamental research rather than short-term product development.

A Long-Term Bet on the Future of AI

The launch of AMI Labs reflects a growing debate within the AI community about the direction of the field.

While large language models dominate today’s AI landscape, researchers including LeCun argue that systems capable of understanding the physical world will ultimately be necessary to reach more advanced forms of intelligence.

With more than a billion dollars in funding and a global research network, AMI Labs is positioning itself to explore that possibility over the coming years.