NVIDIA Jetson is a family of embedded AI computing modules purpose-built for running deep-learning inference at the edge. The current Orin generation includes seven SKUs sharing one Ampere-architecture System-on-Chip: the entry-level Jetson Orin Nano Super (up to 67 TOPS at 7-25 W, $249 dev kit), the mid-range Orin NX (up to 157 TOPS), the flagship AGX Orin 64 GB (275 TOPS at 15-60 W), and the latest AGX Thor 128 GB designed for the largest VLA and humanoid workloads. All modules run NVIDIA’s JetPack SDK, a Linux-for-Tegra environment with CUDA, TensorRT, cuDNN, and the full Isaac robotics stack. This consistency means policies trained in Isaac Sim or Isaac Lab can be deployed directly to Jetson hardware with hardware-acceleration of vision, language, and control models. Jetson modules support multi-camera CSI input, GMSL automotive cameras, GigE, PCIe, USB 3.2, and broad sensor compatibility through partner carrier boards. Used in Boston Dynamics Spot, Agility Digit, Unitree, ANYbotics, and hundreds of other commercial robots. NVIDIA reports a Jetson developer ecosystem of over a million developers, 6,000 customers (including 2,000 startups), and 150 hardware partners.
Family of embedded AI computing modules (Orin Nano, Orin NX, AGX Orin, AGX Thor) running deep-learning inference on robots. Top-end AGX Orin delivers 275 TOPS of AI performance at 15-60 W; the de-facto standard for on-board AI in commercial robotics.
NVIDIA Jetson is a family of embedded AI computing modules purpose-built for running deep-learning inference at the edge. The current Orin generation includes seven SKUs sharing one Ampere-architecture System-on-Chip: the entry-level Jetson Orin Nano Super (up to 67 TOPS at 7-25 W, $249 dev kit), the mid-range Orin NX (up to 157 TOPS), the flagship AGX Orin 64 GB (275 TOPS at 15-60 W), and the latest AGX Thor 128 GB designed for the largest VLA and humanoid workloads. All modules run NVIDIA’s JetPack SDK, a Linux-for-Tegra environment with CUDA, TensorRT, cuDNN, and the full Isaac robotics stack. This consistency means policies trained in Isaac Sim or Isaac Lab can be deployed directly to Jetson hardware with hardware-acceleration of vision, language, and control models. Jetson modules support multi-camera CSI input, GMSL automotive cameras, GigE, PCIe, USB 3.2, and broad sensor compatibility through partner carrier boards. Used in Boston Dynamics Spot, Agility Digit, Unitree, ANYbotics, and hundreds of other commercial robots. NVIDIA reports a Jetson developer ecosystem of over a million developers, 6,000 customers (including 2,000 startups), and 150 hardware partners.