Rankings/Advanced Machine Intelligence (AMI Labs)
Advanced Machine Intelligence (AMI Labs) logo

Advanced Machine Intelligence (AMI Labs)

NativePrivate

Yann LeCun's AI-native world models lab

CEO
Alexandre LeBrun
Founded
2025
HQ
Paris, France
Total Funding
$1.0B
Employees
51-200
Flag a data issue
Foundation Models

Insufficient data

We don't have enough public data on Advanced Machine Intelligence (AMI Labs) to score it yet (3 of 18 signals; 6 needed). It stays on the AI250 and will be ranked once more data comes in.

Products & Offerings

AMI World Models (JEPA-based)

Research-stage world models trained on real-world sensory data that aim to reason, plan and stay controllable for physical-world uses such as robotics and industry.

ModelResearchEmerging

Executive Summary

What they do

  • Research lab building JEPA-based world models that learn from real-world sensory data
  • Aims for AI with persistent memory, reasoning, planning, controllability and safety
  • Targets industry, robotics, healthcare and physical AI rather than text chatbots

Why it matters

  • Raised $1.03B seed in March 2026, Europe's largest seed round reported
  • Led by Turing laureate Yann LeCun, a leading critic of LLM-only approaches
  • Backed by Bezos Expeditions, Nvidia, Samsung, Cathay Innovation, Greycroft and HV Capital

Best fit

  • Robotics and industrial firms seeking early research partnerships on physical AI
  • Healthcare AI teams exploring world models, following the Nabla relationship
  • Enterprise R&D groups tracking alternatives to large language model architectures

Watch-outs

  • No commercial product yet; leaders frame it as long-horizon research
  • Founders suggested a saleable product could be roughly five years away
  • Licensing, pricing and enterprise support terms are entirely undefined so far

Latest News

Analyst VerdictMonitor

AMI Labs has a top-tier research team and more than $1B of funding, but sources found no product to evaluate or deploy, and its leaders describe commercialization as years away. Enterprise buyers should track its research output and any early partner programs, especially in robotics, industrial and healthcare settings, rather than plan procurement now. Buyers with strategic R&D interest in physical AI may want to open a relationship early.

Signal Provenance

Latest observation date and source for each signal category

Signal CategoryLast UpdatedSource
Funding Signals
Last funding round, Funding stage
Oct 31, 2026Montclare research agents
Hiring Velocity
Headcount, Headcount growth
NeverMontclare research agents
Technical Output
Research papers, GitHub activity
NeverMontclare research agents
News Sentiment
News mentions, News sentiment
Nov 1, 2026Montclare research agents
Partnership Signals
Enterprise partnerships, Distribution partners
Oct 29, 2026Montclare research agents
Revenue Indicators
ARR estimate, Revenue growth
Nov 3, 2026Montclare research agents
Leadership Signals
CEO tenure, Founder-led
NeverMontclare research agents
Competitive Position
Category rank, Market share
NeverMontclare research agents

Risks

Pre-product research lab with no generally available commercial offering reported
High
Commercial timeline may stretch to roughly five years, per leadership statements
High
JEPA world-model approach is technically unproven at scale against frontier LLMs
Medium
Crowded world-model field includes World Labs, Google DeepMind and Nvidia Cosmos
Medium
Heavy reliance on LeCun, who also holds an NYU professorship
Low
Feedback