San Francisco-based AI robotics startup Dream Labs has raised $90 million in seed funding led by B Capital, with backing from Kindred Ventures. The round also included participation from A.Capital, Unusual Ventures, SV Angel, Wing VC, Scribble VC, and angel investor Adrian Aoun, among others.
Founded in 2026 by former NVIDIA researchers Joel Jang, Yunhao Ge, Scott Reed, and You Liang Tan, Dream Labs develops AI foundation models designed to enable robots from different manufacturers to perform physical tasks. Jang serves as the company’s CEO.
Dream Labs’ core technology is Dream-1, a World Action Model that processes visual information and natural-language instructions to generate robot actions. The model learns from video data and is designed to transfer capabilities across different robotic systems, reducing the need for extensive machine-specific training.
Alongside the funding announcement on October 19, 2026, Dream Labs introduced Dream API, a publicly accessible interface that allows developers to connect robotic hardware to Dream-1. The platform also offers fine-tuning services for adapting the model to specific tasks and a Robot Playground for interacting with robots remotely.
Dream-1 initially focuses on robotic manipulation using two-armed systems equipped with parallel grippers. The company has demonstrated applications involving object handling, sorting and other everyday manipulation tasks. Dream Labs describes the model as a research beta and notes that its performance varies across robots and tasks.
Dream Labs is working with more than 20 hardware partners. The company plans to use the new funding primarily to expand its computing resources, support additional robot configurations and tasks, and scale access to its technology.
