At the recent All In conference, discussions around world models—a type of AI focused on automating spatial intelligence—highlighted the secretive nature of companies developing this technology. Key players such as Yann LeCun’s AMI Labs and Fei-Fei Li’s World Labs have attracted attention and investment but remain tight-lipped about specific products or timelines.

World models aim to create AI systems capable of understanding and navigating physical environments, with applications ranging from robotics and interactive video to advanced self-driving vehicles. Despite this wide scope, concrete commercialization plans remain unclear. Michael Rabbat, AMI Labs’ VP of World Models, emphasized that the company is still in a research and development phase and is not ready to disclose product details.

World Labs’ Marble platform demonstrates some capabilities, including media creation and interactive environments for gaming and CGI, but these appear more like proofs of concept than market-ready solutions. Even suppliers like Physicl, which provides data to these companies, acknowledge limited insight into the end uses of their contributions.

The broad applicability of world models—from autonomous navigation to biomedical AI—adds to the ambiguity. AMI Labs is exploring diverse sectors including manufacturing, healthcare, and robotics, but has not committed publicly to any specific direction.

Industry observers note that the ample funding available reduces pressure to focus narrowly, allowing companies to explore multiple avenues while maintaining secrecy. Revealing breakthroughs prematurely could invite competition from other startups and established AI firms like OpenAI and Anthropic.

This cautious approach resembles a strategic "dark forest" scenario, where companies avoid drawing attention to protect their innovations until they are ready to compete openly. As the field matures, clearer commercial applications and competitive dynamics are expected to emerge.