Dream Labs Exits Stealth: Launches Dream-1 Robot Control Model, Closes $90M Seed Round

What happened

According to SiliconANGLE's October 19 report, the round was jointly led by B Capital and Kindred Ventures, with participation from A.Capital, Unusual Ventures, Wing VC, SV Angel, Scribble VC and angel investor Adrian Aoun. B Capital's Daisy Cai said the team did some of the defining work in robot learning at Nvidia, and now anyone with an API key can build on it.

Dream-1 treats robotic skills as video-prediction problems: trained on what the company calls human egocentric training data — videos of people doing physical work such as stacking shelves and packing boxes — it learns what should logically happen next, then decides the physical action the robot should take. The company says video data is far easier to accumulate than machine-specific datasets, and the learned intelligence transfers across different hardware.

The company says Dream-1 works with any bimanual robot equipped with parallel grippers, and getting started requires only a video feed plus natural-language instructions. Co-founder and CEO Joel Jang said the hardware is ready, and what has been missing is a brain any robot can use. Beyond the API, Dream Labs also launched Fine-Tuning-as-a-Service (submit video demonstrating a new task and get a customized Dream-1 within hours) and a browser-accessible Robot Playground (type a command and watch a live robot respond in real time). The company is working with more than 20 robotic hardware companies whose arms sell for between $2,000 and $10,000.

Key facts

1. Funding:a $90 million seed round, jointly led by B Capital and Kindred Ventures.
2. Model:Dream-1, a universal robot foundation model that turns robot-skill learning into a video-prediction problem (the company claims it controls bimanual robots it has never seen, on tasks it was never shown).
3. Pricing:the Dream API costs $3 per hour, with a lightweight version at $0.75 per hour.
4. Demo:Robot Playground — type a command in the browser and watch a live robot respond in real time.
5. Partners:more than 20 robotic hardware companies, with arms priced between $2,000 and $10,000.
6. Team:co-founder and CEO Joel Jang; investors say team members did defining work in robot learning at Nvidia.