Decart raises $300M to run real-time world model on Amazon chips

Decart has raised $300 million in a funding round led by Radical Ventures, bringing the AI research lab’s total capital raised past $450 million. The two-year-old company, which specializes in real-time video and world models, announced the deal on Monday, with a high-profile group of backers including Nvidia, Sequoia Capital, Benchmark, Adobe, and Toyota Ventures. New investors Atreides Management, Valor Equity Partners, and eBay Ventures also joined, alongside returning supporters Sequoia, Benchmark, and Zeev Ventures.

The angel investors offer a telling clue about Decart’s strategic positioning. The cap table features Andrej Karpathy, co-founder of OpenAI and former head of AI at Tesla, alongside former Disney CEO Michael Eisner, the Nintendo family, and gaming investor Moritz Baier-Lentz. This blend of media, gaming, and infrastructure heavyweights, rather than pure software engineering names, maps directly to the use cases Decart’s products are now targeting.

Decart sells three core offerings. The Decart Optimization Stack (DOS) is an inference and training platform that works across Nvidia GPUs, Google TPUs, and Amazon Trainium. The company claims it delivers 1,600 tokens per second for agentic inference, compared to an industry average of around 200, plus full-HD video inference at up to 100 frames per second. Lucy is a “world model for immersive experiences” that responds to user input in under 30 milliseconds and is now deployed in virtual try-on, live streaming, and dynamic in-video advertising. Oasis targets physical AI applications, including robotics and autonomous systems, building on the real-time Minecraft-style demo that went viral in October 2024.

The Amazon partnership stands out as the most significant commercial detail. Decart describes itself as one of the first companies to deploy real-time AI models of this scale on AWS Trainium, with its Lucy2 model running on Trainium3. Nafea Bshara, vice president of Amazon’s Annapurna Labs, noted in a statement that Lucy2 exceeds 80% Model FLOPS Utilization, meaning more of the chip’s raw power is doing productive work.

CEO Dean Leitersdorf framed world models as “the key to moving AI from the virtual world into the physical world,” arguing that language models “fundamentally operate in text” and “don’t understand how the physical world behaves.” The company’s funding trajectory reflects its rapid growth. Decart closed a $32 million Series A at a $500 million valuation in December 2024, just four months after its $21 million seed round. Fortune reported in August 2025 that the company had raised $100 million at a $3.1 billion valuation. Today’s round pushes total funding above $450 million.

Leitersdorf and co-founder Moshe Shalev have been building the company since 2023. Their thesis, articulated most clearly on the Sequoia podcast, is that vertically integrated optimization, rather than larger models, is the missing layer of the real-time AI stack. The investor mix follows a recognizable Nvidia-equity pattern. The chipmaker has committed over $40 billion in AI equity in 2026 alone, typically taking a stake in companies that then sign long-term GPU commitments, with some GPU revenue flowing back to Nvidia. Decart, whose DOS stack runs across all three major chip families, presents a sharper test of that pattern. It is both a Nvidia portfolio asset in equity terms and a public reference point for Amazon Trainium and Google TPU in customer terms.

The end-markets align with the angel list. Lucy is in production with retailers and streaming platforms for virtual try-on and dynamic advertising. Oasis is being positioned for robotics and autonomous-vehicle simulation. Gaming and live-experience use cases sit underneath both. The wider industry context, as the conversation around AI-generated media shifts from possibility to procurement decisions, centers on real-time, low-latency inference at production scale as the bottleneck most existing video models have not solved.

Decart did not disclose run-rate revenue, the new valuation, or the hyperscaler customers licensing DOS. The next public milestones include DOS 2.0, announced alongside the funding, plus Lucy 2.5 and Oasis 3, both expected in the coming weeks.