Introduction
Jetson AGX Thor is Nvidia's new "robot brain" developer kit priced at $3,499; it enables building robots that run advanced AI models and have expanded memory capacity
Context
Quick definition: Jetson AGX Thor is a robotics module based on Nvidia's Blackwell GPU generation, designed to run AI models on robots and edge devices
Main features
The Jetson Thor kit uses Blackwell architecture and is reported to be 7.5× faster than the previous generation. Modules include 128 GB of memory to support generative AI, large language models and visual models needed for environment perception
Pricing and availability
Nvidia lists the developer kit at $3,499 and said first kits ship next month. For production, Thor T5000 modules will be sold, with a volume price of $2,999 per module for orders above 1,000 units
Who uses it and why it matters
Companies including Agility Robotics, Amazon, Meta and Boston Dynamics use Jetson chips, and Nvidia has invested in startups such as Field AI. The module supports prototyping and production of smarter robots by running complex models on-device
The problem / Challenge
Robotics remains a small portion of Nvidia's revenue (about 1%), so scaling requires software-hardware integration, cost efficiencies and proven use cases to justify production investment
Solution / Approach
Nvidia's approach is to provide infrastructure and associated software: developer kits for prototyping followed by production modules, lowering technical barriers for companies integrating generative AI into robots
"We do not build robots, we do not build cars, but we enable the whole industry with our infrastructure computers and the associated software"
Deepu Talla, Vice President of Robotics and Edge AI / Nvidia
Market impact
Nvidia's leadership sees robotics as its largest growth opportunity outside AI; the combined automotive and robotics unit reported $567 million in quarterly sales, up 72% year over year, indicating fast growth even if the segment is currently small
Technical considerations and limits
The Blackwell GPU, 128 GB memory and ability to run LLMs make Thor suitable for complex tasks, but commercial success depends on software ecosystems, unit costs and industrial demand; current figures do not guarantee immediate mass adoption
Conclusion
Jetson AGX Thor is a meaningful step to enable on-board generative AI in robots; pricing and Nvidia's go-to-market show a strategy to convert GPU leadership into robotics opportunities
FAQ
Short answer: Jetson AGX Thor is a Nvidia robotics module aimed at running complex AI models directly on robots
The developer kit retails at $3,499; production Thor T5000 modules cost $2,999 each for orders above 1,000 units
Nvidia said the first developer kits will ship next month
Agility Robotics, Amazon, Meta and Boston Dynamics are among the users; Nvidia has invested in Field AI
128 GB enables running large and visual models that interpret surroundings, essential for humanoid robots
No, it currently accounts for about 1% of total revenue but shows rapid growth
No, Nvidia supplies hardware and software infrastructure that other companies use to build robots
- How much does Jetson AGX Thor cost?
- When will the first kits ship?
- Which companies use Jetson modules?
- Why is 128 GB memory significant?
- Is robotics a major revenue source for Nvidia?
- Does Nvidia build robots?