World model AI companies are keeping secrets

During a panel discussion at the All In conference on September 20, 2026, moderated by Russell Brandom, the secretive nature of the artificial intelligence sector came into sharp focus. Prominent players such as Yann LeCun's AMI Labs and Fei-Fei Li's World Labs have successfully accumulated massive amounts of buzz and funding, yet they currently rank remarkably low on the monetization scale while fiercely guarding their strategic directions.

At their core, world models focus on automating spatial intelligence, opening up lucrative avenues ranging from robotics and interactive video to sophisticated autonomous driving systems. Although World Labs' Marble represents one of the most advanced products in this domain—capable of generating explorable environments for video games and CGI effects—the platform still appears primarily designed to showcase raw capabilities rather than immediate commercial output.

Stock photo for illustration only, not from the actual event

This shroud of secrecy extends far beyond the core development teams to reach their supply chain partners. Alex de Vigan, CEO of data supplier Physicl, noted on the sidelines of the same conference that while his firm's data is clearly utilized in ongoing developments, he remains completely in the dark about the exact nature of the final applications, stating:

"I wish they would tell us more. We could build more useful data if we knew what they were working on"

Alex de Vigan

Part of this mystery stems from the immense versatility inherent to world models. The same foundational modeling approach utilized for autonomous navigation can theoretically assist humanoid robots with physical labor or transform video footage into interactive environments. AMI Labs, for instance, has already explored manufacturing, biomedicine, robotics, and medical software through its Nabia partnership, though pursuing all avenues simultaneously remains impractical.