AMI Labs and World Labs keep world model plans secret, TechCrunch panel reveals

Panel lifts the lid on an opaque AI sector

According to TechCrunch, a panel on world models at the All In conference — an event the writer notes has no connection to the podcast of the same name — offered a rare look at one of the least transparent corners of the AI industry. The two names dominating the field are AMI Labs, founded by Yann LeCun, and World Labs, led by Fei-Fei Li. Both have drawn considerable attention and capital, TechCrunch reports, yet neither appears far along the path to generating revenue.

World models are, at their core, an effort to automate spatial intelligence. That same underlying idea could point toward many markets: robotics, interactive video, or more sophisticated self-driving systems.

AMI Labs keeps its cards close

Michael Rabbat, a co-founder of AMI Labs and the company's VP of World Models, sat on the panel but declined to describe what the company is actually building. Pressed on the question, he said the company would "talk about it when we're ready to talk about it." He later clarified over email that AMI is "still in a research and building phase" and is not discussing product plans or timelines publicly.

TechCrunch notes that AMI is less than a year old, which makes the reticence understandable. But the reporter found the same vagueness running through the entire field.

Marble leads the field, but only as a showcase

Per TechCrunch, the most mature product in the space is Marble from World Labs. Its demonstrations span media creation, explorable environments built for video games, and CGI effects, and there are robotics applications as well. Even so, the platform reads more like a display of capabilities than a product aimed at a defined commercial market.

Even data suppliers are left guessing

The secrecy extends beyond customers and journalists. On the sidelines of the same conference, TechCrunch spoke with Alex de Vigan, CEO of Physicl, a company that supplies data to the emerging world model business. De Vigan said he knows his company's data has proved useful to these labs, but he has no visibility into what they are building with it. "I wish they would tell us more," he said, adding that Physicl "could build more useful data" with a clearer picture of the labs' work.

Silence as strategy

Part of the fog, TechCrunch argues, comes from how flexible the concept is. In its simplest form, a world model is a navigable map of the world, not unlike the models that steer self-driving cars. But the techniques that let an autonomous vehicle move through traffic could also guide a humanoid robot carrying boxes, or turn a few minutes of video into an explorable environment. AMI has already signaled activity across manufacturing, biomedicine, robotics and doctor-facing AI software through its Nabia partnership — a spread broad enough that outsiders cannot tell which of them, if any, is the true focus.

There is also a commercial logic to saying nothing. As long as fundraising stays easy, TechCrunch observes, there is little pressure to commit to a single application. And announcing a concrete product — a humanoid robot platform, say, or a next-generation Hollywood rendering system — would instantly make the space interesting to other world model companies, the so-called neolabs, and even OpenAI and Anthropic. The same pools of capital that let a lab build quietly are available to potential rivals once a path to market becomes visible.

TechCrunch likens the standoff to a "dark forest" scenario drawn from Cixin Liu's fiction: when you cannot tell who else is out there, the safest move is to avoid attracting attention.

Why it matters

The world model sector is absorbing significant investment while disclosing almost nothing about products, timelines or target markets. Suppliers, partners and investors are left evaluating demos rather than roadmaps — and even the companies' own data vendors are guessing at the end use. On TechCrunch's reading, this opacity is not merely a habit but a deliberate strategy to delay competition. The moment any lab reveals a viable commercial path, well-funded rivals can move in quickly, which suggests the field could shift from quiet to crowded in a very short window.