# Nvidia CEO showcases open-source Microduck companion robots as holiday preorders surpass 20,000

Jensen Huang’s GTC keynote demonstration of the AI-trained waddling duck robots sparks a viral consumer rush for Hugging Face and Pollen Robotics hardware.

- Canonical: https://reviewblip.com/news/nvidia-ceo-showcases-open-source-microduck-companion-robots-as-holiday-preorders-surpass-20-000/
- Published: 2026-10-24T20:00:58.654Z
- Updated: 2026-10-24T20:00:58.654Z
- Author: ReviewBlip
- Category: Consumer Robotics & Edge AI
- Type: News

![Microduck robotic companion duck demonstrating autonomous locomotion and voice response](https://reviewblip.com/editorial-images/6c46ba94d082f3c604ad23566fbba4d61abda1a8ab54c29433e8bf21e1ae0210)

Image credit: TechCrunch / Pollen Robotics / Hugging Face / Nvidia — consumer robotics hardware documentation

A keynote demonstration by Nvidia chief executive Jensen Huang at the company’s European GTC conference in Berlin has ignited a commercial frenzy for open-source robotics, sending preorders for the newly revealed Microduck companion robot past 20,000 units within twenty-four hours. Developed by French robotics laboratory Pollen Robotics and open-source AI platform Hugging Face—which Nvidia acquired in September 2026—the palm-sized bipedal robots combine on-device neural voice models with autonomous balance control.

During the live stage presentation, Huang interacted with a flock of the miniature robotic ducks as they responded dynamically to voice commands, performed coordinated somersaults, vocalized expressive synthesizer chirps, and simulated sleep routines by collapsing flat before righting themselves. Hugging Face chief executive Clément Delangue confirmed that the immediate surge in demand guarantees delivery ahead of the 2026 winter holiday season, positioning the open-hardware device as an unexpected consumer tech hit spanning both children and adult hobbyists.

The hardware architecture of the Microduck relies entirely on open-source specifications, featuring 3D-printable chassis files, off-the-shelf micro-actuators, and dual front-facing tracking cameras. The robots were trained using Nvidia’s Isaac Sim physics engine and reinforcement learning frameworks, allowing simulated synthetic ducks to master locomotion, balance recovery, and spatial orientation across thousands of virtual environments before flashing the resulting lightweight neural policies directly onto consumer units.

Pollen Robotics designed the Microduck with an open community ecosystem, enabling buyers to train custom physical tricks and voice responses via Hugging Face’s LeRobot platform and distribute behavioral weights to other owners. With consumer robotics historically constrained by brittle cloud dependencies and prohibitive pricing, the Microduck’s sub-$200 bill of materials and fully local edge inference represent a landmark inflection point for affordable physical AI in everyday households.

## Sources
- [\[src-microduck-robot\] Hottest toy for Christmas? Nvidia CEO plays with adorable Microduck robots, 20,000 sell](https://techcrunch.com/2026/10/23/hottest-toy-for-christmas-nvidia-ceo-plays-with-adorable-microduck-robots-20000-sell/)