Veeda AI raises $90M for world models
Archive item — written before sources were shown.
Sanja Fidler, ex-Nvidia research lead, raised over $90M seed for Veeda AI, led by Khosla and Radical, to build multimodal world models for training physical AI.
Toronto-based Veeda AI raised more than $90 million in seed funding, led by Khosla Ventures and Radical Ventures, one of the largest seed rounds in Canadian history and struck about three months after the company was founded. Veeda is run by Sanja Fidler, who led AI research at Nvidia before leaving to start it, with co-founders Huan Ling as chief scientist and Zan Gojcic as CTO, both former Nvidia research colleagues. The company builds multimodal “world models” that simulate the physical world, aiming to generate scalable simulated environments where embodied AI agents and robots can be trained through repeated interaction.
What it means for you
World models are the training-data play for physical AI: instead of collecting real-world robot data, which is slow and expensive, you generate limitless realistic simulation and let agents practice in it. A $90 million seed, at that size and speed, is investors betting that whoever supplies the simulation layer sits upstream of the entire robotics stack, much as data labelers sit upstream of language models. The pedigree matters here; Fidler’s group at Nvidia did foundational work on exactly this problem. It is the same applied-AI thesis drawing capital this week, from Rillet’s $1B valuation for AI-native accounting to the inference-hardware buildout behind Etched’s $700M round. If your work touches robotics or embodied agents, watch which world-model provider ships usable environments first.
- 01Sanja Fidler's world model startup Veeda AI raises $90M in seed fundingsiliconangle.com · primary reporting
- 02Veeda AI raises $90M+ seed backed by Khosla and Radical to build world models for physical AIunite.ai · reporting
