The Jetson AGX Thor is Nvidia’s most powerful robotics processor, costing customers $3,499 to start developing on the new device, as the company competes more intensively to carve out a future business in the robotics arena. This groundbreaking new Nvidia chip, called the robot brain, will run on the most advanced Blackwell chip, which is fully optimized to scale, and will astonish Oren Yosef with astounding humanoid robot and self-driving car applications with leading-edge artificial intelligence processing.
Jetson AGX Thor targets robotics revolution
Nvidia introduced its new chip module on Monday, aimed at operating in robots, and it is available in developer kit form and costs $3,499. The chip, known by the company as a robot brain, is scheduled to appear next month as the company moves forward to provide customers with the first kits to create sophisticated robotic systems.
The negotiations between Nvidia and Thor companies to sell Thor T5000 modules will be in the commercially available robots when the companies develop the modules using the developer kit. To those companies that need more than 1,000 Thor chips, Nvidia will charge a price of $2,999 per module, where scale robotics deployment will become more cost-effective.
How next-generation robotics is fueled by Blackwell architecture
The Jetson Thor processors are based on the latest generation graphics processors, Blackwell, developed by Nvidia, which are also found in AI chips and video games. Nvidia claimed its Jetson Thor processors have a 750 percent improvement compared with previous-generation systems, which means they can run generative AI models, such as large language models and visual models capable of understanding the world around them.
CEO sees the greatest growth opportunity in robotics
Nvidia CEO Jensen Huang has stated that robotics is the biggest non-artificial intelligence growth opportunity, and the company has seen its overall revenues multiply more than three times over the last two years. We do not manufacture robots, we do not manufacture cars, but we can help the entire industry with our infrastructure computers, and the accompanying software, said Deepu Talla, who is the vice president of robotics and edge at Nvidia.
The Jetson Thor chips come with 128GB of memory, which is necessary to support large AI models that humanoid robots need. This large memory capacity allows the processing of complex representations of visual and language coverage required in the autonomous under-decision of dynamic environments.
The Nvidia Jetson chips have been used in organisations such as Agility Robotics, Amazon, Meta, and Boston Dynamics. Nvidia has also invested in robotics companies, including Field AI, which shows how it is committed to the robotics ecosystem beyond the provision of hardware.
There is an automotive market extension that manufactures automobiles
Nvidia added Jetson Thor is also intended to make fully autonomous vehicles, particularly those made in China as well. The company markets its car chips as Drive AGX, and though similar to robotics chips, the company has tailored the operating system, Drive OS, to be employed in cars.
Recently, Nvidia merged its business departments under the same line, which was the automotive and robotics division. Robotics amounted to only approximately 1 percent of total sales at the company, but saw a 72 percent year-on-year growth of just over 567 million in quarterly sales in May. Yet robotics is a modest but fast-growing endeavor at Nvidia, with this enterprise comprising less than 1 percent of overall revenue since it began shipping a few new robot chips in 2014.
Jetson AGX Thor is not only a new product, but a new direction reflecting Nvidia’s stance on robotics as an upcoming AI market unit. Having 7.5 times faster performance, with a memory capacity of 128GB and the support of key industry players such as Amazon and Meta, this robot brain can help Nvidia to conquer the new robotics revolution.