Decart, a frontier AI research lab focused on real-time video and world models, announced a $300 million funding round led by Radical Ventures as the company expands its infrastructure platform for low-latency AI systems. New investors in the round included NVIDIA, Adobe Ventures, Toyota Ventures, eBay Ventures, Atreides Management, and Valor Equity Partners. Existing backers Sequoia Capital, Benchmark, and Zeev Ventures also participated. Individual investors included Andrej Karpathy, Michael Eisner, the Nintendo family, and gaming investor Moritz Baier-Lentz. This funding round was raised at nearly a $4 billion valuation.
Founded in 2023 and headquartered in San Francisco with operations in New York City and Tel Aviv, Decart is building a vertically integrated AI stack spanning world models, inference systems, and optimization infrastructure. The company said the funding will support the expansion of three core product lines: DOS, Lucy, and Oasis.
The company said DOS 2.0, the latest version of its Decart Optimization Stack, is designed to accelerate inference and training workloads for large language models, agentic AI, video systems, and world models across NVIDIA GPUs, Google TPUs, and Amazon Trainium chips. According to Decart, DOS 2.0 can deliver more than 1,000 tokens per second for agentic inference and support full-HD video and world model inference at up to 100 frames per second.
Decart said Lucy, its immersive world model platform, powers real-time experiences across gaming, e-commerce, streaming, and advertising, while Oasis focuses on simulations for robotics and autonomous systems. The company added that Lucy responds to user input in under 30 milliseconds and is already deployed with enterprise customers across multiple industries.
The company also highlighted a strategic collaboration with Amazon Web Services, which includes a commercial agreement and a joint go-to-market initiative. Decart said it is among the first companies deploying real-time world models on Amazon’s Trainium chips at scale, with Lucy2 currently running on Trainium3 hardware.
Decart said it has now raised more than $450 million to date.
KEY QUOTES:
“World models are the key to moving AI from the virtual world into the physical world. Language models have taken us far, but they fundamentally operate in text, they don’t understand how the physical world behaves. To move beyond that, AI needs to model environments that persist, react, and follow real-world dynamics – it needs to be able to understand structure and physics. Once systems can understand and operate within the physical world, the range of what becomes possible expands dramatically, from robotics and autonomous systems to entirely new forms of commerce and live experience.”
Dean Leitersdorf, CEO & co-founder, Decart
“Decart is the kind of partner we love working with: technologically fearless, fast-moving, and deeply innovative. They were among the first in the world to optimize on Trainium3 and have embraced our newest software libraries. The results speak for themselves. Their latest model, Lucy2, exceeds 80% Model FLOPS Utilization, meaning more of the chip’s raw power is doing real, productive work.”
Nafea Bshara, VP, Amazon’s Annapurna Labs
“Decart is the engine powering the future of real-time media and physical AI. Decart makes State of the Art real-time video economically viable at massive scale. While the AI industry has considered this technological achievement years away, Decart has built a software platform that combines proprietary chip optimization, SOTA world models, and customer API that has closed the gap between research and self-serve product platform in a way no one else has.”
Jordan Jacobs, Managing Partner, Radical Ventures
