AMI Labs (Advanced Machine Intelligence) – AI World Models
AMI Labs (Advanced Machine Intelligence) builds world models — AI systems that learn abstract representations of the physical world from sensor data to support reliable reasoning, planning, and safe action in real environments. Founded in late 2025/early 2026 by Turing Award winner Yann LeCun with CEO Alexandre LeBrun, the company targets foundational technology for embodied and agentic applications in robotics, industrial automation, and healthcare.
- Status: Seed stage (one of Europe’s largest seed rounds)
- Focus: World models for physical-world understanding, controllability, and safe agentic behavior
Core Data Grid
| Funding Round | Lead Investors / Notable Backers | Total Raised (approx.) | HQ Location | Industry Sector | Estimated Team Size | Key Partners / Validation (if material) |
|---|---|---|---|---|---|---|
| Seed | Co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital, Bezos Expeditions; additional from NVIDIA, Temasek, Samsung, Toyota Ventures, Bpifrance + angels (Jeff Bezos, Mark Cuban, Eric Schmidt, Tim Berners-Lee) | $1.03 billion at $3.5B pre-money valuation | Paris, France (hubs in New York, Montreal, Singapore) | Frontier AI / World Models / Physical & Embodied AI | Not publicly disclosed (growing team of researchers and engineers across four global hubs) | Elite investor syndicate including NVIDIA and Bezos Expeditions; open research and collaboration model; active global hiring |
AMI Labs Leadership & Structural Breakdown
Key Leadership:
- Yann LeCun, Co-founder & Executive Chairman: Turing Award winner; former Chief AI Scientist at Meta (FAIR); leading advocate for world models and JEPA architecture. LinkedIn
- Alexandre LeBrun (Alex LeBrun), CEO: Serial AI entrepreneur and co-founder/former CEO of Nabla (medical AI); previously collaborated with LeCun at Meta FAIR; École Polytechnique alumnus. LinkedIn
- Key technical roles: Saining Xie (Chief Science Officer), Pascale Fung (Chief Research and Innovation Officer), Mike Rabbat (VP World Models) — senior researchers with deep ex-Meta FAIR pedigrees. Laurent Solly serves as COO
Primary Competitors:
- Figure AI — Humanoid robotics company advancing embodied AI and physical interaction.
- Covariant — Robotics foundation models focused on real-world manipulation and warehouse automation.
- 1X Technologies — Embodied AI agents designed for safe operation in human environments.
Core Use Cases & Market Problem:
- Enabling safe, long-horizon planning and prediction in robotics and industrial automation where current AI lacks grounded understanding of physics and causality.
- Supporting reliable agentic systems in healthcare assistive devices, wearables, and process control that require real-time world modeling rather than language-only reasoning.
- Addresses the concrete limitation that dominant AI approaches excel at text but struggle with physical prediction, controllability, and safe deployment in unstructured real-world settings.
Plain English Explanation
AMI Labs develops AI systems called world models that build an internal understanding of how the physical world behaves by learning from cameras and other sensors. These models let AI agents predict the results of their actions and plan accordingly, with built-in safety constraints — moving beyond today’s language-focused AI toward machines that can reliably operate in factories, homes, or clinical environments.
Target Customers & Adoption Context
Robotics developers, industrial automation providers, large manufacturers, and healthcare technology companies that need AI capable of safe physical interaction and long-term planning. The primary friction removed is the gap between current AI’s strong language performance and its weak ability to understand, predict, and act reliably in noisy, continuous real-world sensor environments where errors carry high costs.
Capital & Traction Signals
Closed a $1.03 billion seed round in March 2026 — one of the largest early-stage AI financings globally and Europe’s largest seed — at a $3.5 billion pre-money valuation. The syndicate combines strategic depth (NVIDIA, Toyota Ventures, Samsung) with high-conviction capital (Bezos Expeditions) and strong regional players. CEO Alexandre LeBrun brings prior AI startup success via Nabla. The founding technical team includes multiple senior ex-Meta FAIR researchers in scientific leadership roles. Operations launched immediately across four global hubs with active hiring of researchers and engineers. Emphasis on open publications and collaboration. No public customer deployments or revenue figures disclosed yet, consistent with a frontier research lab model at this stage.
Investor Lens
AMI Labs enters the market at a clear inflection in the 2026 capital cycle, as allocator attention shifts from pure language model scaling toward physical and embodied intelligence — the domain where current approaches show persistent gaps in planning, causality, and safe real-world operation. The company’s world model thesis directly targets this bottleneck, positioning it as potential foundational infrastructure for reliable agentic systems.
Validation is exceptionally high: direct scientific leadership from Turing Award winner Yann LeCun, concentration of proven ex-Meta FAIR research talent, demonstrated entrepreneurial execution by CEO Alex LeBrun, and one of the strongest investor groups assembled in recent AI (strategic hardware participation from NVIDIA alongside patient capital from Bezos Expeditions).
Momentum is evident in the immediate multi-continent operational footprint, rapid assembly of research and engineering talent, and alignment with macro tailwinds in robotics capital expenditure, industrial automation, and demand for safer embodied agents. Toyota Ventures and NVIDIA involvement signal credible downstream pathways in automotive and physical systems.
Watchpoints from an allocator perspective include the long development timelines typical of foundational AI research before production deployments, execution risk in achieving controllable and scalable real-world performance, and competition from both well-capitalized robotics platforms and internal efforts at major labs. As a research-first organization, the path from capital to near-term commercial traction remains to be proven.
One-sentence synthesis: The rare alignment of LeCun’s foundational research authority, a technically deep and operationally experienced founding team, and a globally distributed investor syndicate with both strategic and patient capital creates substantial asymmetric potential for AMI Labs to shape the world models layer of physical AI infrastructure, contingent on delivering early demonstrations of reliable agentic performance in priority domains over the next 12–24 months.
Last Updated: June 2026
Sources:
- https://amilabs.xyz/ (company site and team details)
- https://techcrunch.com/2026/03/09/yann-lecuns-ami-labs-raises-1-03-billion-to-build-world-models/ (funding and team coverage)
- https://sifted.eu/articles/yann-lecun-ami-labs-meta-funding-round-nvidia (investor and European context)
- Yann LeCun and Alexandre LeBrun LinkedIn/X announcements (March 2026)
- https://futurumgroup.com/insights/yann-lecuns-ami-raises-1bn-seed-round-is-the-world-model-era-finally-here/ (valuation and backers)
- Preqin / company structure references for leadership roles
