Advanced Machine Intelligence Secures $1.03 Billion to Build World-Understanding AI
March 15, 2026, 9:39 am
Location: United States, California, San Francisco
Employees: 11-50
Founded date: 2015
Advanced Machine Intelligence (AMI) secured an unprecedented $1.03 billion in seed funding. This landmark investment values the AI startup at $3.5 billion. AI pioneer Yann LeCun co-founded AMI. The company is forging a new path beyond current large language models (LLMs). AMI develops "world models." These systems are designed to truly comprehend the physical world. They integrate continuous sensor data, advanced reasoning, and robust planning capabilities. AMI targets building highly reliable, controllable, and safe AI. Key applications span industrial process control, complex automation, advanced robotics, and critical healthcare systems. This significant capital influx signals a major evolution in artificial intelligence research and its real-world application.
A new titan emerges in artificial intelligence. Advanced Machine Intelligence (AMI) just closed a staggering $1.03 billion seed funding round. This monumental investment signals a bold shift. It values the nascent AI startup at $3.5 billion pre-money. This is one of the largest seed rounds in European history. The capital infusion will accelerate AMI's mission. It aims to develop a new class of AI systems. These systems will truly understand the physical world.
The funding round attracted a powerful consortium of investors. SBVA, Greycroft Partners, Cathay Innovation, and Hiro Capital participated. HV Capital, Bezos Expeditions, Nvidia, and Temasek also contributed. Notable individual investors include Mark Cuban, Eric Schmidt, and Xavier Niel. Toyota Ventures, New Legacy Ventures, and Samsung provided strategic backing. This diverse group reflects confidence in AMI's ambitious vision. It underscores a collective belief in the future of world-model AI.
At the helm is Yann LeCun. He is a celebrated AI pioneer. LeCun previously served as Chief AI Scientist at Meta. He co-received the 2018 ACM A.M. Turing Award. His leadership brings immense credibility. LeCun has long advocated for AI systems that move beyond mere pattern recognition. AMI represents his pursuit of machines that grasp reality. Alexandre LeBrun joins LeCun as co-founder and CEO. LeBrun previously led the healthcare AI startup Nabla. Nabla is now AMI’s first partner.
AMI's core innovation lies in its "world models." This approach fundamentally differs from today's dominant large language models (LLMs). LLMs excel at predicting the next word or token. They demonstrate powerful capabilities in language understanding and generation. However, their understanding remains largely superficial. They do not genuinely comprehend the physical world. They rely on statistical correlations.
Real-world environments are complex. They involve continuous, high-dimensional, and often noisy data. Think of cameras, lidar, or various industrial sensors. Generative architectures, common in LLMs, struggle with this unpredictability. They might generate plausible outcomes. But they lack a deep understanding of cause and effect. They often fall short in scenarios demanding precise control and safety. AMI tackles these limitations head-on.
AMI's architecture centers on self-supervised learning. It uses Joint Embedding Predictive Architectures (JEPA). This allows systems to learn abstract representations of sensor data. They filter out unpredictable details. Their predictions occur in this representation space. This enables robust learning from vast, unlabeled real-world data. It moves beyond token-based predictions.
World models are designed for agentic systems. They predict the consequences of actions. This is a critical capability. An AI agent can then plan action sequences. It works towards a specific goal. Safety guardrails are built into the planning process. This ensures controllable and reliable operation. Such capabilities are vital for physical world interaction.
The applications for AMI's technology are transformative. Industrial process control stands to benefit significantly. AI can optimize complex manufacturing lines. It can manage critical infrastructure. Automation across various sectors will see new levels of intelligence. Robots operating in dynamic environments demand true world understanding. AMI's systems will enhance their autonomy and safety.
Healthcare is another crucial domain. AI needs high reliability and controllability. AMI can power advanced diagnostic tools. It can assist in surgical robotics. Wearable devices could gain deeper context awareness. They could provide proactive health interventions. The focus remains on AI where safety and reliability are paramount.
AMI operates with a global footprint from inception. Its main office is in Paris, France. Additional strategic locations include New York City, Montreal, and Singapore. This international presence facilitates diverse talent acquisition. It enables access to varied research ecosystems. This broad operational base supports AMI's global ambitions.
The scale of AMI's seed funding is unprecedented. It highlights the industry's readiness for new AI paradigms. Current generative AI has shown immense potential. Yet, it reveals inherent limitations in physical world understanding. AMI's world models offer a promising path forward. They represent a quest for deeper machine intelligence.
This venture is not just about funding. It is about pioneering the next generation of AI. It addresses fundamental challenges in AI development. It aims for systems that are truly intelligent. They will perceive, reason, and act in the physical world. AMI positions itself at the forefront of this evolution. The future of AI may well be shaped by world models. This seed round marks a monumental beginning.
A new titan emerges in artificial intelligence. Advanced Machine Intelligence (AMI) just closed a staggering $1.03 billion seed funding round. This monumental investment signals a bold shift. It values the nascent AI startup at $3.5 billion pre-money. This is one of the largest seed rounds in European history. The capital infusion will accelerate AMI's mission. It aims to develop a new class of AI systems. These systems will truly understand the physical world.
The funding round attracted a powerful consortium of investors. SBVA, Greycroft Partners, Cathay Innovation, and Hiro Capital participated. HV Capital, Bezos Expeditions, Nvidia, and Temasek also contributed. Notable individual investors include Mark Cuban, Eric Schmidt, and Xavier Niel. Toyota Ventures, New Legacy Ventures, and Samsung provided strategic backing. This diverse group reflects confidence in AMI's ambitious vision. It underscores a collective belief in the future of world-model AI.
At the helm is Yann LeCun. He is a celebrated AI pioneer. LeCun previously served as Chief AI Scientist at Meta. He co-received the 2018 ACM A.M. Turing Award. His leadership brings immense credibility. LeCun has long advocated for AI systems that move beyond mere pattern recognition. AMI represents his pursuit of machines that grasp reality. Alexandre LeBrun joins LeCun as co-founder and CEO. LeBrun previously led the healthcare AI startup Nabla. Nabla is now AMI’s first partner.
AMI's core innovation lies in its "world models." This approach fundamentally differs from today's dominant large language models (LLMs). LLMs excel at predicting the next word or token. They demonstrate powerful capabilities in language understanding and generation. However, their understanding remains largely superficial. They do not genuinely comprehend the physical world. They rely on statistical correlations.
Real-world environments are complex. They involve continuous, high-dimensional, and often noisy data. Think of cameras, lidar, or various industrial sensors. Generative architectures, common in LLMs, struggle with this unpredictability. They might generate plausible outcomes. But they lack a deep understanding of cause and effect. They often fall short in scenarios demanding precise control and safety. AMI tackles these limitations head-on.
AMI's architecture centers on self-supervised learning. It uses Joint Embedding Predictive Architectures (JEPA). This allows systems to learn abstract representations of sensor data. They filter out unpredictable details. Their predictions occur in this representation space. This enables robust learning from vast, unlabeled real-world data. It moves beyond token-based predictions.
World models are designed for agentic systems. They predict the consequences of actions. This is a critical capability. An AI agent can then plan action sequences. It works towards a specific goal. Safety guardrails are built into the planning process. This ensures controllable and reliable operation. Such capabilities are vital for physical world interaction.
The applications for AMI's technology are transformative. Industrial process control stands to benefit significantly. AI can optimize complex manufacturing lines. It can manage critical infrastructure. Automation across various sectors will see new levels of intelligence. Robots operating in dynamic environments demand true world understanding. AMI's systems will enhance their autonomy and safety.
Healthcare is another crucial domain. AI needs high reliability and controllability. AMI can power advanced diagnostic tools. It can assist in surgical robotics. Wearable devices could gain deeper context awareness. They could provide proactive health interventions. The focus remains on AI where safety and reliability are paramount.
AMI operates with a global footprint from inception. Its main office is in Paris, France. Additional strategic locations include New York City, Montreal, and Singapore. This international presence facilitates diverse talent acquisition. It enables access to varied research ecosystems. This broad operational base supports AMI's global ambitions.
The scale of AMI's seed funding is unprecedented. It highlights the industry's readiness for new AI paradigms. Current generative AI has shown immense potential. Yet, it reveals inherent limitations in physical world understanding. AMI's world models offer a promising path forward. They represent a quest for deeper machine intelligence.
This venture is not just about funding. It is about pioneering the next generation of AI. It addresses fundamental challenges in AI development. It aims for systems that are truly intelligent. They will perceive, reason, and act in the physical world. AMI positions itself at the forefront of this evolution. The future of AI may well be shaped by world models. This seed round marks a monumental beginning.