NVIDIA Jetson Orin NX
The NVIDIA Jetson Orin NX is a compact, SO-DIMM form-factor edge AI compute module designed for robotics, autonomous machines, and embedded vision applications. Manufactured by NVIDIA, it is available in 8 GB and 16 GB memory configurations and pairs an Ampere-architecture GPU with an 8-core Arm Cortex-A78AE CPU to deliver high-throughput on-device inference without relying on cloud connectivity. Positioned within NVIDIA's broader Jetson Orin family, the Orin NX targets developers and system integrators who need a balance of performance and power efficiency in a small footprint. It is commonly used in applications such as industrial inspection, autonomous mobile robots (AMRs), drone navigation, and smart edge devices where real-time AI inference is critical.

Overview and Use Cases
The NVIDIA Jetson Orin NX is part of NVIDIA's Jetson Orin platform, a line of edge AI modules intended to bring data-center-class AI performance to embedded and robotics deployments. The Orin NX specifically occupies a mid-tier position in the family, offering a step up from the Jetson Orin Nano while remaining more power-efficient and cost-accessible than the flagship Jetson AGX Orin.
Typical deployment scenarios include:
- Autonomous mobile robots (AMRs) in warehouses and logistics facilities
- Industrial machine vision and quality inspection systems
- Unmanned aerial vehicles (UAVs) requiring onboard perception
- Smart cameras and edge inference appliances
- Medical imaging devices and portable diagnostic tools
The module's SO-DIMM form factor allows it to be swapped into carrier boards designed for earlier Jetson modules, easing hardware iteration for developers.
Key Technical Details
The Jetson Orin NX is available in two primary SKUs:
- 8 GB variant: Fewer active GPU streaming multiprocessors and CPU cores enabled, lower maximum power envelope
- 16 GB variant: More GPU and CPU resources unlocked, higher peak AI performance
Core architectural highlights (as publicly documented by NVIDIA):
- GPU: NVIDIA Ampere architecture with dedicated Tensor Cores for INT8/FP16 inference
- CPU: Up to 8-core Arm Cortex-A78AE v8.2 64-bit processor
- AI Performance: Reportedly up to 100 TOPS (tera-operations per second) for the 16 GB variant, as stated in NVIDIA's official product materials
- Memory: LPDDR5 unified memory shared between CPU and GPU
- Connectivity: PCIe, USB, CSI camera interfaces, Gigabit Ethernet, and MIPI support via carrier board
- Power: Configurable TDP modes, typically ranging from approximately 10 W to 25 W depending on configuration
The module runs NVIDIA's JetPack SDK, which bundles CUDA, cuDNN, TensorRT, and the DeepStream SDK, enabling a familiar software stack for AI developers.
Comparison to Related Products
Within the Jetson ecosystem, the Orin NX sits between the entry-level Jetson Orin Nano and the high-end Jetson AGX Orin. It offers more AI compute than the Nano series while consuming less power and costing less than the AGX Orin, making it a practical choice for mid-complexity robotics tasks.
Compared to competing edge AI modules such as the Qualcomm Robotics RB5 platform or the Google Coral Dev Board, the Jetson Orin NX benefits from NVIDIA's mature CUDA software ecosystem and broad community support, though competing platforms may offer advantages in specific low-power or cost-sensitive scenarios.
Market Context and Target Buyers
The Jetson Orin NX is positioned as a professional-grade component for:
- Robotics OEMs building products that require certified, production-ready AI compute
- Research institutions prototyping autonomous systems
- System integrators deploying edge AI in industrial or commercial environments
As of public reporting, the module is available through NVIDIA's authorized distributors and partners. Pricing varies by region and volume; NVIDIA has publicly listed suggested prices for the modules, with the 8 GB variant being more accessible than the 16 GB. Exact current pricing should be confirmed through official channels.
Deployments and Ecosystem
The Jetson Orin NX is supported by a wide range of third-party carrier boards from manufacturers such as Seeed Studio, Auvidea, and Connect Tech, enabling rapid integration into custom hardware designs. NVIDIA's Isaac ROS framework and Isaac Sim simulation environment are optimized for Jetson hardware, making the Orin NX a natural fit for ROS 2-based robotics development.
Notable application areas reported in the industry include warehouse automation, agricultural robotics, and autonomous inspection drones, though specific customer deployments are not always publicly disclosed.
Future Outlook
As edge AI workloads grow in complexity—driven by advances in large vision-language models and multi-sensor fusion—modules like the Jetson Orin NX are expected to see continued adoption in next-generation robotics platforms. NVIDIA's ongoing investment in the JetPack SDK and Isaac ecosystem suggests sustained software support. Future generations of Jetson hardware are anticipated, though NVIDIA has not publicly committed to specific successor timelines as of available reporting.
Related videos
Related entries
RobotUniversal Robots UR5e
The Universal Robots UR5e is a six-axis collaborative robot arm (cobot) belonging to Universal Robots' e-Series product line. Designed for light-to-medium industrial and laboratory tasks, it is widely used in assembly, pick-and-place, machine tending, quality inspection, and lab automation workflows. Universal Robots, a Danish company and a subsidiary of Teradyne, is one of the most recognized names in the collaborative robotics market. The UR5e is programmed using Universal Robots' PolyScope graphical interface on a teach pendant, making it accessible to operators without deep robotics expertise. Its built-in force/torque sensing, tool-center-point control, and a broad ecosystem of certified end-effectors and accessories (the UR+ platform) have made it a popular mid-range cobot choice across manufacturing, electronics, food handling, and research sectors.
17 views
RobotUnitree Z1
The Unitree Z1 is a compact, six-degrees-of-freedom (6-DOF) robotic arm developed by Unitree Robotics, a Chinese robotics company headquartered in Hangzhou. Designed for both standalone research and integration onto mobile robotic platforms, the Z1 targets universities, robotics developers, and automation engineers who need a lightweight, capable manipulator at a relatively accessible price point. The arm features joint torque sensing, position and force control modes, and an open programming interface, making it well-suited for tasks such as object manipulation, pick-and-place operations, and human–robot interaction research. It is notably compatible with Unitree's own quadruped and humanoid platforms, enabling loco-manipulation experiments on mobile bases.
16 views
RobotQuicktron M100
The Quicktron M100 is a heavy-duty autonomous mobile robot (AMR) belonging to Quicktron Robotics' M-Series product line. It is designed for demanding material handling tasks in warehouses, distribution centers, and manufacturing environments, using an integrated lift mechanism to transport shelves and pallets autonomously across facility floors. Quicktron Robotics, a company with roots in China and a global commercial presence, positions the M100 as a high-capacity solution for operations that require moving heavier loads than lighter AMR models can accommodate. The M100 targets logistics operators and manufacturers seeking to automate goods-to-person or pallet-movement workflows at scale.
16 views
RobotDash
Dash is an award-winning educational robot designed for children, produced by Wonder Workshop. It responds to voice commands and environmental sounds, and is capable of moving, dancing, and producing audio, making it an engaging hands-on learning tool for classrooms and homes alike. Dash supports progressive coding education through Wonder Workshop's companion apps—including Blockly-based visual programming and the Wonder app—along with a curriculum library aligned to STEM standards. It has received recognition from educators and industry observers as an accessible entry point for teaching computational thinking to children as young as six.
16 views

