Enterprise AI, in production.
    Four desks, one beat.

    Retail · News

    Dyna Robotics Lands $120 Million to Advance Embodied AI

    With deployments in hotels, gyms, and laundromats, the startup positions its DYNA-1 model for broader adoption

    Last updated

    Dyna Robotics Lands $120 Million to Advance Embodied AI

    When Lindon Gao, York Yang, and Jason Ma founded Dyna Robotics in 2024, they were not entering unfamiliar territory. Gao and Yang had previously built Caper AI, a company that applied computer vision and hardware to create AI-powered shopping carts, which sold to Instacart in 2021 for $350 million. Ma, a former DeepMind research scientist, had spent his career developing foundation models for robotics. Their new venture was designed around a straightforward conviction: advances in AI would only be fully realized when machines could perform useful physical work outside the lab.

    In just over a year, the Redwood City-based startup has raised more than $140 million, including a $120 million Series A announced in September. Investors range from early-stage venture firms like CRV and First Round Capital to corporate funds from Nvidia, Amazon, Salesforce, Samsung, and LG. The company is now valued above $600 million.

    From Demonstration to Deployment

    The company’s central product is DYNA-1, a robotics foundation model that powers stationary robotic arms. Unlike traditional robots that rely on scripted instructions, DYNA-1 is designed to generalize across environments. According to the company, the system achieved more than 99% uptime during a 24-hour continuous operation test. Within six months of launch, its robots were running up to sixteen hours a day in hotels, restaurants, laundromats, and gyms, performing tasks such as folding napkins, towels, and clothing.

    Jason Ma described the company’s approach: “Scalable real-world robot learning systems need to master and generalize many manipulation skills. To achieve the best performance on complex tasks, Dyna's foundation models are developed to enable general world understanding while learning from the models' own experience for rapid online learning”.

    by AIM Media HouseFree · No paywall

    Keep reading. It's free.

    Create a free Deployed account to finish this story and every story on the site. Choose the briefings you want with it:

    or

    No credit card. One click to unsubscribe. Already a reader? Sign in

    About the author

    Deployed

    Related stories