[FINANCIAL] AI / Foundation Models ·
World Models · Seed Round
— Closed March 10, 2026
Yann LeCun's AMI Labs Raises $1.03B Seed —
Europe's Largest Seed Round Ever — to Build AI World Models
The father of convolutional neural networks departs Meta and
bets $1B that the future of AI belongs to "world models" — systems
that reason and plan about physical environments — rather than next-token
prediction. Co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital,
and Bezos Expeditions, with NVIDIA, Toyota, Samsung, Eric Schmidt, and Jeff
Bezos among backers.
DEAL SNAPSHOT
|
ROUND SIZE $1.03 Billion |
POST-MONEY VALUATION ~$3.5 Billion (pre-money
basis) |
|
ROUND TYPE Seed (Europe's largest
ever seed round) |
CO-LEAD INVESTORS Cathay Innovation,
Greycroft, Hiro Capital, HV Capital, Bezos Expeditions |
|
CLOSED March 10, 2026 |
STRATEGIC INVESTORS NVIDIA, Toyota, Samsung,
Temasek, SBVA |
THE PARTIES
|
ACQUIRER / LEAD INVESTOR Cathay Innovation +
Greycroft + Hiro Capital + HV Capital + Bezos Expeditions (co-leads) +
NVIDIA, Toyota, Samsung, Temasek, Eric Schmidt, Jeff Bezos, Mark Cuban
(participants) Five VC firms
co-led the round: Cathay Innovation (Paris/SF), Greycroft (NY), Hiro Capital
(London), HV Capital (Munich), and Bezos Expeditions (Bezos's personal
investment vehicle). Strategic investors include NVIDIA, Toyota, Samsung,
Temasek (Singapore), and SBVA. Celebrity tech investors Eric Schmidt (former
Google CEO), Jeff Bezos (Amazon founder), and Mark Cuban also participated. |
COMPANY AMI Labs (Advanced Machine
Intelligence) Paris,
France-based AI research company co-founded by Yann LeCun (former Chief AI
Scientist at Meta, Turing Award winner, father of CNNs) and Alexandre Lebrun
(founder of wit.ai, sold to Facebook). Founded early 2026. Team of 12 at time
of raise; LeCun stated AMI will hire 20-30 people "in the very short
term." Focus: building AI systems based on "world model"
architectures that maintain persistent memory, plan through complex problems,
and reason about physical environments. |
THE INVESTMENT THESIS: WORLD MODELS VS.
LLMS
Yann LeCun has
been the most prominent critic of the large language model (LLM) paradigm — the
architecture underlying ChatGPT, Claude, and Gemini. His core argument:
predicting the next token in a text sequence cannot produce genuinely
intelligent systems, because intelligence requires understanding physical
causality, planning under uncertainty, and building internal representations of
how the world works. These capabilities require "world models" — AI
architectures that simulate physical environments and reason about
consequences.
AMI Labs is
founded specifically to commercialize world model AI. The company plans to
build systems capable of reasoning and planning in complex real-world settings,
with initial target applications in manufacturing, robotics, aerospace, and
biomedical sectors — all domains where physical-world understanding is more
valuable than text generation.
"Current AI approaches based on predicting the next word or pixel
will not produce broadly capable intelligent agents. World models are the next
step toward AI that genuinely understands and navigates the physical
world."
— Yann LeCun,
Co-founder & Chief AI Scientist, AMI Labs
MARKET CONTEXT
The $1.03B seed
round — Europe's largest ever seed financing — reflects a striking bet by
investors that the LLM paradigm, while commercially dominant today, will be
displaced or supplemented by world model architectures over a 5-10 year
horizon. CEO Alexandre Lebrun predicted that "world models will be the
next buzzword — in six months, every company will call itself a world model to
raise funding." The round makes AMI the most richly valued seed-stage AI
company in European history, valued at approximately $3.5B before generating
any revenue.
ADVISORS
Company
Legal Advisor (AMI Labs): Jones Day
(advised on company formation and the $1.03B financing)
No financial
advisors were disclosed. The round was assembled through direct VC and
strategic investor relationships.