Decart Raises $300M to Integrate Real-Time World Models with Amazon’s Chips
May 18, 2026 – 1:58 pm
Image by: Decart
Radical Ventures led the funding round, with Nvidia, Sequoia, Benchmark, Adobe, and Toyota alongside. Andrej Karpathy, Michael Eisner, and the Nintendo family are among the angel investors. Total raised now exceeds $450 million.
Decart, an AI research lab developing real-time video and world models, announced today that it has secured $300 million in new funding led by Radical Ventures. This round brings Decart’s total raised to over $450 million, with Nvidia, Atreides Management, Valor Equity Partners, Adobe Ventures, Toyota Ventures, and eBay Ventures joining existing investors Sequoia Capital, Benchmark, and Zeev Ventures.
The impressive list of angel investors includes Andrej Karpathy, co-founder of OpenAI and former Tesla AI head, Michael Eisner, ex-Disney CEO, the Nintendo family, and gaming investor Moritz Baier-Lentz.
Decart focuses on media, gaming, and infrastructure applications, aligning with its target use cases. They offer:
- DOS (Decart Optimization Stack): An inference and training platform that runs across Nvidia GPUs, Google TPUs, and Amazon Trainium, delivering 1,600 tokens per second for agentic inference—a significant improvement over the industry average of around 200. It also supports full-HD video inference at up to 100 frames per second.
- Lucy: A "world model for immersive experiences" that responds to user input in under 30 milliseconds and is deployed in virtual try-on, live streaming, and dynamic in-video advertising.
- Oasis: A parallel product for physical AI aimed at robotics and autonomous systems, built on Decart’s original real-time Minecraft-style demo in October 2024.
The announcement highlights Decart’s partnership with Amazon. Decart is one of the first companies to deploy real-time AI models at this scale on AWS Trainium, with its Lucy2 model running on Trainium3, achieving over 80% Model FLOPS Utilization.
Decart CEO Dean Leitersdorf emphasizes the significance of world models in bridging the gap between virtual and physical worlds, arguing that language models "fundamentally operate in text" and lack understanding of real-world physics.
Decart’s funding trajectory is clear:
- December 2024: $32 million Series A at a $500 million valuation.
- August 2025: Fortune reported $100 million raised at a $3.1 billion valuation.
- Present: Total funding exceeding $450 million.