World Labs CEO Dr. Fei-Fei Li and SceniX Co-Founder Yunzhu Li have highlighted the significant data scarcity challenges faced in robotics during a discussion on their collaboration following World Labs’ acquisition of SceniX in July 2026. They noted that, unlike language models, which benefit from abundant online data, robotics suffers from a lack of data for training and evaluation. Their joint efforts are focusing on creating digital worlds that can generate data for robots, potentially replacing costly and unsafe real-world data collection. This approach aims to unlock scalable learning and evaluations essential for advancements in robotics.

SceniX: SceniX is a robotics company specializing in high-fidelity simulation and real-to-sim techniques for training and evaluating robots. It recently joined World Labs, bringing expertise in simulation that supports scalable robot learning without relying solely on real-world data. The integration enables early results in building worlds that train robots for safer and more efficient development.
Yunzhu Li: Yunzhu Li is the co-founder of SceniX, contributing to advancements in simulation technologies for robotics. Following SceniX’s integration into World Labs, he has participated in discussions on overcoming data bottlenecks in robot training and evaluation. His work focuses on creating digital environments that facilitate scalable, safe robot learning and assessment.
World Labs: World Labs is an AI startup focused on spatial intelligence, developing world models that enable machines to perceive, generate, and interact with virtual and physical environments. The company recently acquired SceniX to expand its capabilities into embodied AI and robotics training. This move aligns directly with its efforts to address data limitations in robotics through simulated digital worlds.
Dr. Fei-Fei Li: Dr. Fei-Fei Li is the co-founder and CEO of World Labs, with a background as a leading AI researcher and professor at Stanford. She has been actively discussing the unique data challenges in robotics compared to language models and how world models can unlock scaling laws through environmental simulation. Her recent announcements highlight the integration of SceniX’s team to advance spatial intelligence for physical robot interaction.

Acquisition: World Labs announced the acquisition of SceniX in July 2026 to combine simulation expertise with spatial intelligence for embodied AI applications.
Robotics Development: Discussions emphasize that robotics faces unique data scarcity issues in both training and evaluation, distinct from the abundant internet data available for language models.
World Models Application: Early results from the World Labs and SceniX collaboration demonstrate the use of modeled digital environments to generate data that replaces costly or unsafe real-world collection for robot learning.