Beyond LLMs: AI pioneer Yann LeCun’s new venture AMI raises €890 million to build “world model” AI systems

Advanced Machine Intelligence (AMI),  a new Paris-based startup founded by the former Chief AI Scientist at Meta and co-recipient of the 2018 ACM A.M. Turing Award, Yann LeCun, has raised nearly €890 million ($1.03 billion) at a €3 billion ($3.5 billion) pre-money valuation. 

AMI is developing a new kind of AI systems that comprehend the world, possess persistent memory, and are capable of reasoning, planning, and remaining controllable and safe.

This is among the largest Seed rounds ever raised by a European company. It was co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital, Bezos Expeditions, and various other investors and angels worldwide.

AMI is additionally backed by a group of long-term international investors and strategic partners, such as Toyota Ventures, New Legacy Ventures, Temasek, SBVA, NVIDIA, Mark Cuban, Association Familiale Mulliez, Groupe industriel Marcel Dassault, Sea, and Alpha Intelligence Capital.

Additional notable participation comes from Eric Schmidt, Aglaé Lab, ZEBOX Ventures, Artémis, Xavier Niel, Publicis Groupe, Samsung, Bpifrance Digital Venture, Jim Breyer, Tim & Rosemary Berners-Lee, and Mark Leslie.

Alexandre LeBrun, the co-founder and CEO of Paris-based healthcare AI startup Nabla, will join AMI as its co-founder and CEO, while transitioning from CEO of Nabla to Chief AI Scientist and Chairman. Nabla is AMI’s first partner.

According to AMI, real-world data is continuous, high-dimensional, and noisy, whether it is obtained through cameras or any other sensor modality. It believes that generative architectures, trained through self-supervised learning to anticipate the future, have achieved remarkable success in language understanding and generation. However, since much real-world sensor data is unpredictable, generative methods often fall short.

LeBrun mentioned in his LinkedIn post, “While the solutions we’ve shipped have (I hope) had a positive impact and helped people, they all relied on “shortcuts”. Generative architecture trained by self-supervised learning intelligence; they don’t genuinely understand the world. Predicting tokens, though powerful, works best for discrete and low-dimensional tasks like information retrieval, summarisation, coding, and mathematics.

“However, factories, hospitals, and robots operating in open environments demand AI that grasps reality. And reality is not tokenised: it’s continuous, noisy and high-dimensional. Despite their immense power, I do not believe that generative architectures are the path to achieving this true understanding. It is time to move beyond shortcuts and work on a foundational solution.”

This is where world data comes in. “AMI is developing 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 said in a statement announcing the funding.

AMI claims that it will enhance AI research and develop applications focused on reliability, controllability, and safety, particularly in industrial process control, automation, wearable devices, robotics, healthcare, and more. It has been operating in Paris, New York, Montreal and Singapore from day one.

The post Beyond LLMs: AI pioneer Yann LeCun’s new venture AMI raises €890 million to build “world model” AI systems appeared first on EU-Startups.