Chinese robot makers including Unitree Robotics, AgiBot, Galbot, Engine AI, and UBTech Robotics are among the first to use Nvidia’s new Jetson AGX Thor, launched on August 25, 2025.

The Jetson AGX Thor is a computing platform designed to power robots in sectors such as manufacturing, logistics, healthcare, agriculture, and retail.

It uses Nvidia’s Blackwell GPU and supports 128GB of memory, with the company saying it can deliver up to 2,070 FP4 teraflops of AI computing power at 130 watts.

Nvidia said the new system can run multiple AI workflows in real time, enabling more advanced human-robot interactions.

🔗 Source: South China Morning Post

🧠 Food for thought

1️⃣ China’s robotics surge reflects coordinated industrial strategy paying dividends

China’s adoption of Nvidia’s latest robotics platform demonstrates how targeted government initiatives can rapidly transform market dynamics.

The country has become the world’s largest installer of industrial robots and surpassed the U.S. in robot density, reaching 470 robots per 10,000 manufacturing employees by 20232.

This surge stems from China’s “Made in China 2025” initiative, which specifically targets 70% domestic production of industrially-sold robots by 20253.

The strategy appears to be working, as China’s robotics market is projected to grow 23% annually from $47 billion in 2024 to $108 billion by 20281.

What makes this significant is the speed of execution. The adoption of Nvidia’s advanced robotics platform by multiple Chinese firms like Unitree, AgiBot, and UBTech suggests industry-wide capability building rather than isolated company initiatives.

2️⃣ Nvidia navigates geopolitical tensions by focusing on edge computing rather than data center chips

Nvidia’s success with Chinese robotics companies highlights how the company has strategically pivoted to maintain market access despite U.S.-China tech tensions.

The Jetson AGX Thor platform operates within a 130-watt power envelope while delivering 2,070 teraflops of AI computing power, positioning it as edge computing rather than the high-powered data center chips that face export restrictions1.

This approach allows Nvidia to serve the Chinese market through robotics applications while avoiding the regulatory challenges that have affected its H20 data center chips1.

Recent Nvidia developments