- Dream Labs has raised $90M in seed funding led by B Capital and Kindred Ventures.
- Its Dream API lets developers run one model on robots from different makers.
- Dream-1 handles tabletop tasks only; mobile robots and dexterous hands come later.
Low-cost robots now sell for a few thousand dollars, yet teaching one a single task can still take specialists months, according to Dream Labs. The San Francisco startup, founded in 2026 by former NVIDIA robotics researchers, says its model can get a new robot doing useful work in minutes.
The company has raised $90 million in seed funding, led by B Capital and Kindred Ventures, with participation from A.Capital, Unusual Ventures, SV Angel, Wing VC, Scribble VC, and Adrian Aoun, among others.
It announced the round on October 19 alongside Dream API, which it calls the first publicly available API that turns a robot’s camera feed and a plain-English instruction into motor commands, even for robots it has never seen.
Joel Jang, the co-founder and CEO, was a senior research scientist at NVIDIA’s GEAR Lab, where he led the world-model team for Project GR00T. His team’s DreamZero paper, published in February 2026, described a 14-billion-parameter world action model that adapted to an entirely new robot with 30 minutes of play data.
A model that learns from people
Dream-1 is trained on video of people performing everyday tasks. It predicts what happens next and works out the robot’s resulting action. Dream Labs says that because the model learns from video rather than from one machine, its skills transfer across hardware, and that Dream-1 can control any robot with grippers zero-shot.
“The hardware is ready. What’s been missing is a brain any robot can use. When you get into a rental car, you don’t relearn how to drive. You find the controls and your past driving experience does the rest. Dream-1 approaches a new robot the same way, so a new robot can do useful work in minutes, not months. And instead of edited demos, we’re giving people a working model they can test themselves, mistakes included,” said Jang.
Developers can connect a robot through the Dream API, which runs Dream Labs’ models in data centres worldwide, or send a small batch of demonstration data to a fine-tuning service that returns a tailored model within hours.
A browser-based Robot Playground lets anyone type commands to a robot in the company’s San Francisco office and watch it respond live and unedited. Forbes reported that Dream-1 is priced from $3 an hour, with a cheaper model at 75 cents.
Dream-1 handles tabletop tasks such as picking up and placing objects and folding laundry. Support for wheeled mobile robots and dexterous hands is planned. Jang told Forbes that a pair of dexterous hands costs about $100,000 and that he expects capable ones at $5,000 to $10,000 by 2027.
A crowded race with a different customer
Dream Labs is not the only company selling intelligence for other people’s hardware.
Skild AI raised about $1.4 billion in January 2026, led by SoftBank, at a valuation above $14 billion, and said it reached about $30 million in revenue in 2025 after starting from zero. Rhoda AI, which also trains on video, raised $450 million at a $1.7 billion valuation in March. Genesis AI emerged from stealth with a $105 million seed in July 2025, so a $90 million seed is large but not unique here.
FieldAI announced $405 million across two rounds in August 2025. Figure, which builds its own humanoids, closed a Series C above $1 billion at a $39 billion valuation in September 2025.
Dream Labs is aiming at a different buyer. The robots it supports cost about $2,000 to $10,000, and it works with more than 20 hardware partners.
Daisy Cai, general partner at B Capital, said, “A capable robot arm has gone from the price of a luxury car to the price of a high-end laptop. What it needs now is intelligence,” she said. “This team did some of the defining work in robot learning at NVIDIA. Now anyone with an API key can build on it.”
What the $90M has to prove
Dream Labs will spend the money primarily on compute, to expand the tasks and robots Dream-1 handles and to support more users. Its biggest rivals are spending there too. Figure signed a $3.5 billion AI cloud deal with Nscale in September, and humanoid startups had raised $8.6 billion in 2026 by late July, 1.8 times the 2025 total, according to Dealroom data cited by TFN.
The market Dream Labs points to is large. The International Federation of Robotics counted a record five million industrial robots in factories in 2025, up 9%, with more than 600,000 installed that year, up 11%. But those are industrial machines, and Dream-1 has only served tabletop robots so far.
Jang’s pitch is a working model people can test, mistakes included. The next test is whether those mistakes get rarer as more robots connect, and whether renting out a brain is enough when rivals own the hardware, the data, or both.
