TechCrunch’s Russell Brandom reports on the unusual secrecy enveloping world model startups, a field he explored as moderator of a panel at the All In conference. The two most prominent players are Yann LeCun’s AMI Labs and Fei-Fei Li’s World Labs, both of which have attracted significant funding and buzz but rank low on near-term monetization. At its core, world modeling aims to automate spatial intelligence, with potential applications spanning robotics, interactive video, and advanced self-driving systems. However, when Brandom pressed for concrete commercialization details, the answers grew vague.

AMI Labs co-founder and VP of World Models Michael Rabbat declined to discuss specifics, saying, “We’ll talk about it when we’re ready to talk about it.” In a follow-up email, he clarified: “We’re still in a research and building phase, so we’re not talking publicly about any product plans or timeline.” Given that AMI is less than a year old, such reticence is understandable, but Brandom found this caginess extends across the entire world-modeling space. World Labs‘ Marble, likely the most developed product in the field, showcases capabilities ranging from media creation and explorable video game environments to CGI effects and robotics use cases, yet the platform appears designed more to demonstrate potential than to ship a focused product.

The secrecy even reaches upstream suppliers. Physicl, a data supplier for world model companies, has CEO Alex de Vigan who knows his company’s data is being used but remains in the dark about the specifics. “I wish they would tell us more. We could build more useful data if we knew what they were working on,” de Vigan told Brandom.

The root cause, Brandom argues, is the sheer versatility of world models as a concept. The simplest form is a navigable map of the world, similar to AI models in self-driving cars. The same approach that helps a Waymo navigate traffic could also enable a humanoid robot to carry boxes or transform short video footage into an explorable environment. AMI has already explored manufacturing, biomedicine, robotics, and AI software for doctors through its Nabia partnership, but presumably will concentrate on only a subset. With abundant fundraising available, there is little pressure to commit early.

Competitive dynamics further incentivize silence. If AMI publicly announced a breakthrough like a humanoid OpenClaw or a next-generation rendering system, other labs — including rival world model companies, neolabs, and even OpenAI and Anthropic — would suddenly swarm the space. The easy money that enables stealthy development also funds potential competitors once the path to market becomes clear. Since competition is inevitable, delaying it as long as possible makes strategic sense. Brandom compares this to Cixin Liu’s dark forest scenario: if you don’t know who else is in the woods, it’s best not to attract attention.

While no one doubts viable businesses can be built on world model tech, the combination of open-ended applications, abundant capital, and intense competitive threat keeps the industry unusually tight-lipped about its actual products and timelines. The article, published September 18, 2026, offers a snapshot of a field that remains conceptually exciting yet commercially opaque.