New World Model Startups Tap Investor Appetite for Robotics AI
Two new startups are joining a wave of world model ventures that have attracted billions in investor funding over the past year. Dream Labs, founded this month by Joel Jang, a former Nvidia research scientist who worked on Project Groot, is seeking tens of millions in initial funding. One World AI, founded by NYU professor and Google DeepMind researcher Sherry Yang, is targeting $100 million. Both startups are capitalizing on investor appetite for foundation models that simulate physics and object interaction, capabilities seen as foundational for robotics development.
Two new startups are joining a wave of world model ventures that have attracted billions in investor funding over the past year. Dream Labs, founded this month by Joel Jang, a former Nvidia research scientist who worked on Project Groot, is seeking tens of millions in initial funding. One World AI, founded by NYU professor and Google DeepMind researcher Sherry Yang, is targeting $100 million. Both startups are capitalizing on investor appetite for foundation models that simulate physics and object interaction, capabilities seen as foundational for robotics development.
- Dream Labs, founded by ex-Nvidia researcher Joel Jang, is raising tens of millions for world model development after his work on Nvidia's Project Groot
- One World AI, led by NYU professor and Google DeepMind scientist Sherry Yang, is targeting $100 million in funding for world model research
- World models, which approximate physics and human-object interaction, are attracting major investor interest alongside existing efforts from Fei-Fei Li's World Labs and Yann LeCun's AMI Labs
- Both startups represent a broader trend of researchers leaving established AI labs to launch ventures in the world models space
World models are emerging as a critical research direction for embodied AI and robotics, with major funding flowing to the space from both established players and new entrants. The entry of experienced researchers from Nvidia and Google DeepMind signals that the field has moved beyond early exploration into a competitive commercialization phase. This concentration of talent and capital suggests the industry believes world models will be essential infrastructure for next-generation AI systems.
- Talent migration from established labs like Nvidia and Google DeepMind to startups is accelerating, suggesting these companies may not be moving fast enough on world models or are losing key researchers to entrepreneurial opportunities
- The funding targets (tens of millions to $100 million) indicate world model development is capital-intensive, potentially favoring well-connected founders and those with institutional backing
- Multiple well-funded teams pursuing similar objectives in world models could lead to rapid iteration and breakthroughs, or market fragmentation if differentiation remains unclear
Our Briefing
Weekly signal. No noise. Built for founders, operators, and AI-curious professionals.
No spam. Unsubscribe any time.
