Summary
Fei-Fei Li, co-founder and CEO of World Labs, and Yunu, co-founder of Scenix and assistant professor at Columbia University, join a16z for a wide-ranging conversation on spatial intelligence — the emerging AI frontier centered on systems that can perceive, reason about, and act within three-dimensional space. The episode doubles as a discussion of World Labs’ acquisition of Scenix, the robotics simulation startup that Yunu built following his postdoctoral work with Fei-Fei Li at Stanford.
The core technical argument is about the sim-to-real pipeline. Scenix’s approach maps real-world environments into a high-fidelity digital world, trains robot policies in simulation, and transfers them back to physical hardware. Yunu addresses the long-standing skepticism about simulation — that it always diverges from reality — by arguing that the goal isn’t perfect physics fidelity but capturing the essential structure of a problem, with increasing reliance on real-world data as it accumulates over time. Fei-Fei Li adds a philosophical layer: human intelligence itself is built on simulation and counterfactual reasoning, so simulation is not a workaround but a feature.
The conversation also covers World Labs’ broader thesis that spatial intelligence will power both virtual applications — VFX, gaming, design — and physical robotics, and that these are not separate product lines but expressions of the same underlying capability. For anyone tracking the convergence of world models, embodied AI, and large-scale simulation, this is a high-signal discussion from two of the field’s most prominent researchers.
📺 Source: a16z · Published July 28, 2026
🏷️ Format: Interview
