Anthropic's $6 Billion Decart Deal Targets AI Compute Costs

The report also said Decart builds world models, systems that simulate the physical world for applications such as autonomous driving. Its Lucy AI model is used in eCommerce to let customers try on apparel and accessories virtually.

Decart emerged from stealth in October 2024 with $21 million in seed funding, describing itself as "an efficiency-focused AI research lab." A $32 million Series A followed in December 2024, and a $300 million Series B was announced in May. Its product lines include DOS, an inference and training stack that lets agents and reasoning models run smarter and faster, and Oasis, a world model for physical AI. The reported price is roughly 17 times the $353 million Decart has raised across those rounds.

Chief executive and co-founder Dean Leitersdorf said in a May press release that world models move AI from the virtual world to the physical world. "Once systems can understand and operate within the physical world, the range of what becomes possible expands dramatically," he said, "from robotics and autonomous systems to entirely new forms of commerce and live experience."

In a separate report Thursday, Bloomberg said investors expect Anthropic to pursue a valuation of at least $2 trillion in an October initial public offering, roughly two months away. A listing at that valuation would be the largest stock-market debut in history.