Artificial intelligence startup Advanced Machine Intelligence said Tuesday it has raised $1.03 billion in new funding at a $3.5 billion pre-money valuation, as the company pursues a different path to AI development focused on reasoning, planning, and so-called “world models.”
The company was founded by renowned AI scientist Yann LeCun, who previously served as chief AI scientist at Meta Platforms. The financing marks one of the largest early-stage investments in an AI startup and positions AMI as a high-profile experiment in LeCun’s long-standing view that today’s large language models alone cannot produce truly intelligent machines.
The round was co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital, and Bezos Expeditions, the investment vehicle of Jeff Bezos.
LeCun’s AMI is poised to challenge the dominance of large language models.
The funding places AMI at the center of an emerging debate within the AI research community about whether systems based primarily on large language models can eventually achieve human-level reasoning.
Most of the industry’s recent breakthroughs have come from models trained to predict the next word or pixel in a sequence, the core technique behind tools developed by companies such as OpenAI and Google. While these systems have demonstrated impressive capabilities in generating text, code, and images, LeCun has argued they remain fundamentally limited.
Speaking in an interview, he said that prediction-based models alone will not lead to broadly capable intelligent agents.
AMI instead plans to build AI systems based on “world models,” architectures designed to simulate how the physical world works. These models aim to allow machines to reason about cause and effect, anticipate outcomes, and plan actions in dynamic environments.
According to LeCun, such systems would move AI closer to the type of common-sense reasoning humans rely on when interacting with the real world.
Industrial Applications First
The startup is initially targeting enterprise and industrial customers that operate complex systems where reasoning and planning capabilities could deliver significant productivity gains.
Potential clients include manufacturers, automakers, aerospace companies, biomedical firms, and pharmaceutical groups that manage intricate processes or large-scale simulations.
“We want to become the main provider of intelligent systems, regardless of what the application is,” LeCun said.
Industry analysts say the strategy is a reflection of a broader trend among AI startups to focus first on high-value enterprise deployments rather than consumer applications, where regulatory scrutiny and safety concerns are often greater.
A Potential Path Toward Robotics
While AMI’s early efforts will focus on industrial use cases, LeCun said the technology could eventually enable more advanced consumer products, particularly robotics.
A key challenge for robots operating in human environments is the ability to understand and reason about the physical world — something current AI systems struggle to do reliably.
“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.”
Researchers have long argued that combining perception, reasoning, and planning capabilities could unlock a new generation of autonomous machines capable of performing everyday tasks in homes and workplaces.
Although LeCun left Meta at the end of 2025 after more than a decade leading its AI research efforts, his new venture may still collaborate with the technology giant.
He said discussions were underway about potentially deploying AMI’s technology in smart wearable devices developed by Meta, including its Ray-Ban smart glasses.
Meta has been intensifying its investment in artificial intelligence as competition across the technology sector accelerates. In June 2025, the company reorganized its AI efforts under a new division known as Meta Superintelligence Labs, led by former Scale AI chief executive Alexandr Wang.
The restructuring was part of a broader effort to accelerate development of large language models and next-generation AI systems that could power products across Meta’s ecosystem.
A Different Bet On The Future Of AI
AMI’s creation pops from LeCun’s long-held belief that the industry’s current trajectory may not deliver artificial general intelligence without new architectural breakthroughs. Rather than relying solely on ever-larger datasets and computing power, he argues that AI systems must learn internal representations of the world that allow them to reason about events and plan actions.
The new funding gives AMI substantial resources to pursue that vision at a time when investment in AI infrastructure and research is accelerating worldwide.
Although it’s not certain that the company’s approach can produce a viable alternative to today’s dominant AI models, the scale of the funding round suggests investors are willing to back competing ideas about how the next generation of intelligent systems will be built.
