Robot Ducks: 20,000 Units Reserved After Viral Keynote Moment

A flock of small, waddling robots became the centerpiece of a viral video during Nvidia’s GTC Berlin conference. The clip showed CEO Jensen Huang interacting with the devices, which performed programmed behaviors such as sitting, falling over to simulate sleep, and executing somersaults on voice commands. The footage captured the robots making mechanical noises while moving across the stage, drawing immediate attention from the audience and online viewers.

The surge in interest translated quickly into commercial numbers. By the day following the keynote, preorders for the Microducks exceeded 20,000 units, according to Clem Delangue, head of Hugging Face. He noted on social media that the demand may make the device the year’s best-selling robot, potentially attracting an adult audience as well as children. Promotional material released by the company features adults engaging with the units, reinforcing their appeal beyond the traditional toy market.

The Microducks are the product of Pollen Robotics, a startup acquired by Hugging Face in 2025. Hugging Face itself was purchased by Nvidia in September, placing the robot line under the GPU manufacturer’s corporate umbrella. This marks the second robotics product from the team, following the Reachy Mini. The devices are open source, allowing for community modification and development.

Technically, the robots were trained using Nvidia’s simulation stack. Developers taught the virtual models basic motor skills like walking, sitting, and standing, which are then transferred to the physical units for shipping. Pollen stated that owners can program new tricks and share these custom behaviors with other users, creating a collaborative ecosystem for the hardware.

Delangue expects the units to arrive in time for the Christmas holiday season. The rapid preorder growth suggests the Microducks are positioned as a top contender for holiday gift sales. The devices combine open-source software with consumer-friendly hardware, leveraging Nvidia’s infrastructure to deliver pre-programmed actions while maintaining user flexibility.