Analysis
A TechCrunch-moderated panel on world models at the All In conference exposed how little the category's two most prominent companies -- Yann LeCun's AMI Labs and Fei-Fei Li's World Labs -- are willing to say about their actual business plans, according to TechCrunch and SiliconANGLE.
What A World Model Actually Is
World models are AI systems built to understand and simulate physical, spatial environments rather than just generate text or images -- the underlying technology needed for more capable robotics, interactive video generation, and self-driving systems that reason about 3D space rather than pattern-matching on 2D camera frames. Both AMI Labs and World Labs have positioned themselves as leaders in this category, and both have raised money on the strength of that positioning rather than shipped commercial revenue at meaningful scale.
“The caginess itself is a data point worth taking seriously: companies with a clear, defensible go-to-market story rarely need to stay this quiet about it for this long.”
Evasive On Stage And In Writing
When pressed on the panel about what AMI Labs is actually building toward commercially, co-founder and VP of World Models Michael Rabbat responded, "We'll talk about it when we're ready to talk about it." He later clarified by email that the company is "still in a research and building phase, so we're not talking publicly about any product plans or timeline." AMI Labs, founded by Meta AI chief scientist Yann LeCun, raised $1.03 billion in March 2026 -- a massive round for a company that, per this exchange, still cannot articulate a public product roadmap six months later.
World Labs Has Shipped Something, At Least
World Labs is further along commercially: the company launched Marble, a multimodal generative world model that creates explorable, spatially cohesive 3D environments from images, video or text prompts, with demos spanning media creation, game environments and CGI effects. World Labs was founded in 2024 by Fei-Fei Li alongside Justin Johnson, Christoph Lassner and Ben Mildenhall, emerging from stealth with $230 million at a $1 billion valuation. It is now reportedly in talks to raise new funding at a $5 billion valuation -- backed by a investor list that includes Nvidia, AMD, Fidelity, Autodesk, Emerson Collective and Sea -- a 5x valuation increase in under two years.
The Numbers In Context
A $1 billion-plus round for AMI Labs with no public product roadmap, and a potential 5x valuation step-up for World Labs built substantially on Marble's early demos rather than disclosed revenue, both reflect how much of the current world-model funding wave is priced on the category's promise -- robotics, spatial reasoning, next-generation content creation -- rather than on proven unit economics. That is a materially different risk profile than most late-stage AI infrastructure rounds this year, which increasingly point to specific revenue run-rates as justification, the way Temporal's $550 million round leaned on a disclosed $250 million ARR figure.
What To Watch
Whether World Labs' reported $5 billion round closes at that valuation, and whether AMI Labs breaks its silence with any concrete product announcement in the next few months, will be the real test of whether investor enthusiasm for world models is tracking genuine commercial progress or simply extending on reputation and capital availability. The caginess itself is a data point worth taking seriously: companies with a clear, defensible go-to-market story rarely need to stay this quiet about it for this long.