Back to Tutorials
Introduction to Jetson

NVIDIA Jetson
The Platform for Physical AI

Explore Developer Kits Software Stack Product Page
NVIDIA Jetson family of developer kits and modules
Overview

NVIDIA Jetson™ is a powerful platform for developing innovative edge AI and robotics solutions across industries. It delivers the compact, energy-efficient modules and developer kits with a robust AI software stack you need to deploy next-generation physical AI solutions.

Physical AI Platform

NVIDIA Jetson brings accelerated AI compute to the edge for robotics, autonomous machines, industrial systems, cameras, and other embedded applications.

Unified Software

JetPack provides a common software foundation across Jetson platforms, including Jetson Linux, CUDA, TensorRT, multimedia APIs, and developer tools.

From Devkit to Production

Start development on a Jetson developer kit, then move to a production Jetson module integrated with the carrier board and enclosure for your product.

Software

Software Stack

Jetson Linux, formerly Linux for Tegra (L4T), is NVIDIA's official board support package for Jetson, comprising the kernel, UEFI bootloader, NVIDIA drivers, flashing tools, and an Ubuntu-based root filesystem. Use this path for developer kits, reference platforms, and production software built directly on the NVIDIA-supported BSP.

JetPack SDK packages Jetson Linux with CUDA, cuDNN, TensorRT, multimedia APIs, developer tools, and sample applications. JetPack 7.2 is the SDK release built on Jetson Linux r39.

One software foundation across generations. JetPack 7.2 extends the Ubuntu 24.04, Linux kernel 6.8, and NVIDIA CUDA 13 compute stack introduced with Jetson Thor to the Jetson Orin family, bringing both platforms onto a single unified software foundation.

Agentic-ready for real-world deployment. With optimized memory and performance, JetPack 7.2 adds one-command deployment of NVIDIA NemoClaw and introduces Jetson Agentic Skills, extending the latest compute stack and agentic capabilities to Jetson Orin.

Jetson Linux

Developer Kits vs Modules: What's the difference?

Jetson Developer Kits accelerate application development and prototyping with a Jetson module on a reference carrier board, ready to build on out of the box.

Jetson Modules help you get to market faster with production System-on-Modules you mount onto a carrier board designed for your end product.

Hardware

Developer Kits

Reference platforms built for prototyping, evaluation, and software development. Start here to build on Jetson.

Hardware

Production Modules

System-on-Module (SOM) SKUs for shipping products, grouped by series.

Jetson Thor Series

Blackwell GPU • LPDDR5X
Jetson T5000 module

Jetson T5000

  • • 2070 TFLOPS FP4 sparse • 2560-core Blackwell GPU
  • • 128GB LPDDR5X • 273 GB/s
  • • 14-core Arm Neoverse-V3AE • 40–130W
Module page
Jetson T4000 module

Jetson T4000

  • • 1200 TFLOPS FP4 sparse • 1536-core Blackwell GPU
  • • 64GB LPDDR5X • 273 GB/s
  • • 12-core Arm Neoverse-V3AE • 40–70W
Module page

Jetson AGX Orin Series

Ampere GPU • LPDDR5
Jetson AGX Orin 64GB module

Jetson AGX Orin 64GB

  • • 275 TOPS • 2048-core Ampere GPU
  • • 64GB LPDDR5 • 204.8 GB/s
  • • 12-core Arm Cortex-A78AE • 15–60W
Module page

Jetson Orin NX Series

Ampere GPU • LPDDR5
Jetson Orin NX 16GB module

Jetson Orin NX 16GB

  • • 157 TOPS (Super Mode) • 1024-core Ampere GPU
  • • 16GB LPDDR5 • 102.4 GB/s
  • • 8-core Arm Cortex-A78AE • 10–40W
Module page

Jetson Orin Nano Series

Ampere GPU • LPDDR5
Jetson Orin Nano 8GB module

Jetson Orin Nano 8GB

  • • 67 TOPS (Super Mode) • 1024-core Ampere GPU
  • • 8GB LPDDR5 • 102 GB/s
  • • 6-core Arm Cortex-A78AE • 7–25W
Module page
Platform Comparison

Jetson AGX Thor vs DGX Spark

Both are compact Blackwell-class systems with 128 GB unified memory and 273 GB/s memory bandwidth, but they serve different development goals.

AI Development Workflow

From model development to edge deployment

Use DGX Spark to build and iterate on generative AI models locally. Use Jetson AGX Thor to develop, validate, and deploy physical AI workloads at the edge.

Create Models

Train and experiment

Customize

Fine-tune for your use case

Run & Deploy

Test locally or move to edge

DGX Spark zone
Jetson Thor zone
Edge / Robotics Devkit

Jetson AGX Thor Developer Kit

A reference carrier board with a Jetson T5000 module attached. It lives on your desk for development and emulates the robot, drone, or autonomous machine that will eventually run a production Thor module.

Purpose
Prototype before embedding in a robot
Compute
2070 TFLOPS FP4 sparse (Jetson T5000)
Memory
128 GB LPDDR5X (273 GB/s)
Power
40–130 W
I/O highlights
QSFP28 (4×25GbE), CAN, USB-C, NVMe
Software
Jetson Linux / JetPack 7.2

Choose Jetson AGX Thor when

  • • You are building a robot, drone, AMR, or autonomous machine
  • • Your workload depends on real-time perception, sensor fusion, or physical AI control
  • • You need to prototype on a developer kit before integrating a Jetson module
  • • Your system requires edge I/O such as CAN, MIPI, GbE, or GMSL
Personal AI Workstation

NVIDIA DGX Spark

A personal AI computer. Built around the Grace Blackwell GB10 superchip, it is a finished desktop appliance with no carrier board or production module path. Designed to be the developer's local LLM and AI sandbox.

Purpose
Desktop AI dev, fine-tuning, local LLMs
Compute
1000 TOPS FP4 sparse (GB10)
Memory
128 GB LPDDR5x (273 GB/s)
Power
240 W power supply (140 W GB10 TDP)
I/O highlights
ConnectX-7 networking, USB, NVMe
Software
DGX OS (Ubuntu-based)

Choose DGX Spark when

  • • You are developing, fine-tuning, or evaluating generative AI models locally
  • • You need a compact workstation for LLMs, agents, and AI experimentation
  • • Your workflow is desktop-based rather than embedded in a physical system
  • • You want to bring cloud-style AI development closer to your desk
Related Platform

What about NVIDIA IGX?

Jetson is the right place to start for compact robots, cameras, drones, and embedded AI products. NVIDIA IGX is for industrial-grade edge AI systems that need enterprise support, security, and functional safety.

Learn more about NVIDIA IGX

Where to next?

Explore the Jetson platform, choose the developer kit or module that fits your project, and start building physical AI at the edge.

Getting Started with Jetson Browse Tutorials Explore Models Where to Buy