Decart raises $300M at $4B valuation to power real-time AI
What's the deal? Israeli AI lab Decart has raised $300M in a funding round led by Radical VenturesDealroom has a profile for this one. Try Dealroom →, valuing the company at $4B. Nvidia joined as a new investor alongside eBay VenturesDealroom has a profile for this one. Try Dealroom →, Adobe VenturesDealroom has a profile for this one. Try Dealroom →, Toyota VenturesDealroom has a profile for this one. Try Dealroom →, Atreides ManagementDealroom has a profile for this one. Try Dealroom →, and Valor Equity Partners. Returning backers include Sequoia Capital, Benchmark, and Zeev Ventures.
The angel list is notable: OpenAI co-founder Andrej KarpathyDealroom has a profile for this one. Try Dealroom →, former DisneyDealroom has a profile for this one. Try Dealroom → chief executive Michael EisnerDealroom has a profile for this one. Try Dealroom →, the NintendoDealroom has a profile for this one. Try Dealroom → founding family, and gaming investor Moritz Baier-LentzDealroom has a profile for this one. Try Dealroom →. Total funding now exceeds $450M.
Amazon anchors the commercial side as a strategic customer, while Nvidia serves as both investor and business partner. Founded in late 2023 by Dr. Dean LeitersdorfDealroom has a profile for this one. Try Dealroom → and Moshe ShalevDealroom has a profile for this one. Try Dealroom →, Decart builds real-time video and "world models" — systems that simulate interactive digital environments — along with the optimised infrastructure to run them.
The company sells three products: DOS, an inference and training platform that it says delivers over 1,600 tokens per second for agentic workloads versus an industry average of roughly 200; Lucy, a world model for immersive experiences like virtual try-on and live streaming; and Oasis, a world model for physical AI and robotics.
Why now? AI is shifting from static outputs toward real-time, interactive systems. Decart says its Lucy2 model is already running on Amazon's Trainium3 chips, achieving over 80% Model FLOPS Utilisation — meaning most of the chip's raw power is doing productive work. New versions of all three product lines are launching in coming weeks.
The company raised $100M at a $3.1B valuation in August 2025, making the jump to $4B in under a year. It says it has been profitable almost from day one, generating revenue through contracts with major cloud providers and AI labs.
What could go wrong? Decart's performance claims — 8x faster than comparable systems, 100x improvement in cost efficiency — are self-reported and hard to verify independently. The world-model category is still nascent, and it remains unclear how quickly enterprise demand will materialise at scale.
Competition is fierce. Large cloud providers could build similar optimisation layers in-house, and well-funded rivals are chasing the same real-time AI infrastructure opportunity.
The signal: The investor mix tells the story. Nvidia, Amazon, ToyotaDealroom has a profile for this one. Try Dealroom →, AdobeDealroom has a profile for this one. Try Dealroom →, and eBay aren't just writing cheques — they represent the industries where real-time AI could reshape workflows: gaming, e-commerce, media, advertising, and robotics.
Decart's thesis is that the next wave of AI won't just generate text but will simulate and interact with the physical world in real time. If that bet pays off, the company's vertically integrated approach — building both the models and the infrastructure to run them — could make it a critical layer in the AI stack.
