# Robotis Teams Up with Nvidia to Kick Off Mass Production of 10,000 Humanoids a Year — retained record

> Retained reddit record from reddit:stocks.

Canonical: https://highsignal.app/data/records/8e9f9778da1f72e8

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:stocks)  published 2026-10-14 17:31 ingested 2026-10-15 02:40

[original source ↗](https://www.reddit.com/r/stocks/comments/1x5xiti/robotis_teams_up_with_nvidia_to_kick_off_mass/)

## retained text



```
Robotis has partnered with Nvidia to unveil real-world task capabilities for its humanoid "AI Sapiens" and has begun building a production system targeting 10,000 units annually. At the Humanoid Summit Seoul 2026 held at COEX on the 22nd, Robotis announced its open humanoid platform strategy and showcased how it used Nvidia's Cosmos Transfer to resolve a shadow misidentification problem with synthetic data. Nvidia unveiled its general-purpose robot intelligence model GR00T 1.7 and the development roadmap for next-generation GR00T 2, presenting a "data pyramid" strategy. 

Kim Byung-soo, founder and CEO of Robotis, laid out a vision in his keynote to establish humanoids as a platform where diverse applications perform multiple functions, much like smartphones did in the past. He explained that while automating 100 different tasks once required 100 different types of robots, advances in AI now allow a single robot to handle multiple tasks.

Robotis plans to significantly expand AI Sapiens production starting next year, aiming to scale humanoid manufacturing to an annual rate of 10,000 units. The company's mass production strategy rests on two pillars: high-volume hardware manufacturing and a software ecosystem that enables a single robot to rapidly learn multiple tasks.

On the hardware side, Robotis plans to sequentially release a lineup of roughly 70 actuator models beginning early next year. At the event, AI Sapiens took the stage to demonstrate boxing and squat motions, showcasing the stability of its joint actuation.

Kim compared AI Sapiens to Unitree's G1, noting that "the height is similar at around 136 cm, and the weight is the same," but emphasized that "the higher actuator efficiency gives it the advantage of more than double the battery operating time."

As humanoid production scales up, learning efficiency is expected to become even more critical. Collecting real-world task data for each individual robot requires equipment, operators, time, and space. Robotis is also pursuing a parallel approach: extracting motion from human video or text, training simultaneously across thousands of virtual environments, and then applying the results to physical robots.

https://finance.biggo.com/news/9e9f2ccb-28da-4523-a7c9-dcf8ef58c5f7
```



document fetched 2026-10-15 00:35 · canonical https://reddit.com/r/stocks/comments/1x5xiti/robotis_teams_up_with_nvidia_to_kick_off_mass
