NVIDIA Jetson Nano
by NVIDIA
Small, powerful computer delivering 472 GFLOPs for modern AI in a $99 module, designed for embedded applications and AI IoT at the edge
See https://developer.nvidia.com/embedded/jetson-nano
Features
Processing Power:
- GPU: NVIDIA Maxwell architecture with 128 CUDA cores
- CPU: Quad-core ARM Cortex-A57 processor
- Memory: 4GB 64-bit LPDDR4 at 1600MHz (25.6 GB/s bandwidth)
- Storage: 16GB eMMC 5.1
- Power consumption: 5-10 watts
AI Capabilities:
- 472 GFLOPs performance for modern AI algorithms
- Runs multiple neural networks in parallel
- Processes high-resolution sensors concurrently
- Supports image classification, object detection, segmentation, and speech processing
Media & Video:
- Video encode: Up to 250MP/sec (4K@30fps HEVC)
- Video decode: Up to 500MP/sec (4K@60fps HEVC)
- Camera: 12 MIPI CSI-2 lanes supporting up to 1.5 Gb/s per pair
- HDMI 2.0 and eDP 1.4 display outputs
Connectivity:
- Gigabit Ethernet, M.2 Key E
- 4x USB 3.0, USB 2.0 Micro-B
- GPIO, I2C, I2S, SPI, UART interfaces
- Dimensions: 69.6mm x 45mm (260-pin edge connector)
Software & Development:
- JetPack SDK with accelerated libraries
- Deep learning, computer vision, graphics, multimedia support
- NVIDIA Isaac platform for robotics
- NVIDIA Riva for AI speech
- NVIDIA Metropolis for video analytics
Superpowers
The Jetson Nano brings modern AI to the edge at an accessible price point, making it ideal for:
- Embedded AI developers building edge AI applications without cloud dependency
- Robotics engineers creating autonomous robots with real-time vision and navigation (e.g., JetBot open-source robotics kit)
- Computer vision projects requiring real-time inference for surveillance, smart cities, manufacturing, and retail
- IoT innovators developing intelligent gateways, home robots, and embedded analytics
- Educators and makers learning AI/ML development with affordable, production-ready hardware
Real-world applications:
- Autonomous mobile robots and AGVs in warehouses and logistics
- Real-time object detection and tracking for retail analytics
- Smart camera systems with on-device AI inference
- Agricultural monitoring and automation
- Human-robot interaction systems using pose estimation and gesture recognition
Key advantages:
- Low power consumption (5-10W) enables deployment in power-constrained environments
- Compact form factor fits embedded and mobile applications
- Handles high-resolution video streams in real-time
- Active developer community with extensive project examples
- Production-ready with industrial-grade reliability
Pricing
- Developer Kit: $99 USD (1KU+)
- Jetson Nano Module: Available for production deployment
- Next-gen option: Jetson Orin Nano Super ($249) offers up to 67 TOPS (142x performance improvement)
2025 Status: While newer Orin series offers higher performance, the original Jetson Nano remains relevant for educational purposes, budget-conscious projects, and legacy applications where 472 GFLOPS is sufficient.
Getting Started
Development Kit includes:
- Jetson Nano module with heatsink and fan
- Developer carrier board with ports and interfaces
- JetPack SDK with CUDA, cuDNN, TensorRT libraries
- Access to NVIDIA developer forums and documentation
Popular starter projects:
- JetBot: Open-source DIY robotics kit
- Object detection: Real-time YOLO/SSD implementations
- Pose estimation: Human pose tracking with PoseNet
- Smart cameras: NVR systems with AI analytics