Ex-NVIDIA researchers’ Dream Labs raises $90M to give low-cost robots a shared AI brain

Dream Labs has pulled in $90 million in seed funding and is making a simple pitch: instead of teaching every robot from scratch, rent it a brain that already knows how to work across machines. The round was led by B Capital and Kindred Ventures, with backing from A.Capital, Unusual Ventures, SV Angel, Wing VC, Scribble VC and Adrian Aoun, among others. The startup came out of the gate on October 19 with Dream API, which it says is the first public API that can turn a robot’s camera feed and a plain-English prompt into motor commands, even on hardware it has never seen before.

That’s the bet behind Dream-1, the company’s current model. Right now it handles tabletop tasks like picking up and placing objects and folding laundry. Mobile robots and dexterous hands are coming later. Dream Labs says the model is trained on video of people doing everyday tasks, so it learns what should happen next and then turns that into action. Because it learns from video rather than from one specific machine, the company says it can transfer across hardware and control any robot with grippers zero-shot.

The pitch comes from people who know the field from the inside. CEO Joel Jang was a senior research scientist at NVIDIA’s GEAR Lab and 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 a new robot with 30 minutes of play data. Jang says the aim is to get a new robot doing useful work in minutes, not months, and to let people test the model themselves, mistakes included.

Dream Labs is also trying to make the product easy to touch. Developers can connect a robot through the Dream API, which runs the company’s 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 pricing from $3 an hour for Dream-1, with a cheaper option at 75 cents.

The company is entering a crowded market, but not the same one as the humanoid builders. Its robots cost about $2,000 to $10,000, and it works with more than 20 hardware partners. B Capital’s Daisy Cai framed the point bluntly: hardware has become cheaper, and intelligence is the missing piece. Dream Labs says the new money will go mostly to compute, so it can expand the tasks and robots Dream-1 supports and serve more users. That’s the hard part. Selling the brain is neat. Proving it’s the part people keep paying for is harder.