Sources: Veeda, which is developing world models and is led by ex-Nvidia VP of AI Research Sanja Fidler, raised a $90M+ seed and is backed by Khosla and Radical
The company is led by former Nvidia AI Research VP Sanja Fidler and is focused on control systems for physical AI, distinguishing its stated application from generic AI software funding.
First-order effects
Veeda gains more than $90 million in reported seed funding to build its world-model and physical-AI control-system effort, with Khosla Ventures and Radical Ventures backing the company.
Khosla Ventures now has reported exposure to both Veeda and Embo, two companies pursuing world models for robotics.
Second-order effects
Embo must compete with a newly well-capitalized peer for the attention of investors and prospective robotics partners as its own large seed round remains in talks.
The reported financing raises the capital benchmark for early companies building world models tied to physical systems, rather than application-layer AI products.
Third-order effects
If similarly sized early rounds persist, physical-AI infrastructure may consolidate around a small set of exceptionally funded labs before commercial deployments establish clear technical winners.
Khosla’s participation across reported world-model financings signals a broader investor strategy of building exposure to competing approaches within a foundational AI category.
The trend: World-model startups for robotics are attracting frontier-lab-scale seed financing as investors place early bets on the control layer for physical AI.
I'm more excited than ever about simulation in robotics. One of the main challenges in robotics today is getting enough real-world data to meet the massive data needs of modern robotics. @FidlerSanja, @HuanLing6, and @ZGojcic, who worked on robotics simulation at NVIDIA, just
Today, I'm incredibly excited to introduce Veeda AI, which I co-founded with @FidlerSanja and @ZGojcic . Throughout my career, I've been fascinated by one question: how can visual generative models impact the physical world? My journey has taken me from images to video to
The era of Physical AI is coming, and I firmly believe that embodied intelligence will not emerge from imitation learning alone. Just as humans first learn by observing the world and then build deeper understanding through interaction, the breakthrough moment for Physical AI will
Having collaborated with you guys for so long on world modeling, I am extremely excited to see this come to life. World models have many cool applications. Robotics is arguably one of the most exciting you could have picked! Congrats @ZGojcic, @HuanLing6, and @FidlerSanja!!!
Proud to be backing @VeedaAI as they come out of stealth today. They're building the training ground robotics needs: somewhere robots can fail and learn before the stakes are real. @FidlerSanja, @ZGojcic, and @HuanLing6 spent years tackling some of simulation's hardest problems