As robots become more capable and autonomous, a major challenge remains: enabling real-time decision-making at the edge. NVIDIA's latest innovation, Jetson Thor, tackles that challenge head-on.
Launched as part of NVIDIA’s broader push into Physical AI, Jetson Thor is a compact, high-performance computing module designed specifically for robotic systems. Unlike traditional embedded systems that rely on cloud connectivity for heavy AI tasks, Jetson Thor brings massive compute power directly onto the robot.
🤖 Generative AI, On the Robot
Jetson Thor supports transformer models and large language models (LLMs), allowing robots to interpret natural language commands, process sensor fusion, and engage in multi-modal reasoning—all without sending data off-device. This edge capability means lower latency, higher privacy, and more robust autonomy.
Powered by NVIDIA’s Blackwell GPU architecture, Jetson Thor delivers up to 2,070 FP4 TFLOPS and includes 128 GB of memory within a compact form factor that operates between 40–130W. Compared to its predecessor, Jetson Orin, that’s a 7.5× increase in AI compute and 3.5× improvement in efficiency.
🌍 Broad Ecosystem, Real-World Use Cases
Jetson Thor is not just about speed. It integrates deeply with NVIDIA’s existing software stacks:
Isaac ROS and Isaac Sim for robot development and simulation
Metropolis for vision-based applications
Holoscan for real-time sensor and signal processing
The platform supports cutting-edge models like DeepSeek, LLaMA, and NVIDIA’s own GR00T. Its power and versatility make it ideal for industries ranging from logistics and healthcare to manufacturing and precision agriculture.
✨ A Shift in Physical AI
This marks a meaningful evolution in how robots understand and interact with the world. Instead of relying on remote inference, robots powered by Jetson Thor can:
Navigate dynamic environments
Interpret complex language prompts
Make decisions with minimal delay
Physical AI is no longer a distant concept—it’s being actively shaped by edge computing breakthroughs like Jetson Thor.
🎓 NVIDIA Courses at Rose City Robotics
In partnership with NVIDIA, Rose City Robotics now offers an NVIDIA-Certified Associate: Generative AI Certification Prep course. This program, available at no cost to high school and college students and faculty, provides access to foundational AI tools and hands-on learning resources recognized across industry.
Rose City Robotics is an NVIDIA University Ambassador for Portland State University. Our leadership helps bring NVIDIA-backed education to more learners in the Pacific Northwest, aligning directly with RCR’s mission of access, equity, and real-world readiness.
Whether it's understanding neural networks or building AI-driven robotics systems, these resources provide students with a critical bridge into emerging technologies.
The edge is here. And with tools like these, the frontier of robotics is getting closer to home.
#RoboticsEducation #EdgeAI #JetsonThor #NVIDIA #PhysicalAI #GenerativeAI #RoseCityRobotics
Samantha Town
Events and Outreach
Sam helps shape our communication and operational strategies with a conscientious emphasis on human relationships and impact. Sam graduated from the University of Oregon and grew up working in a family business in Oregon wine country.