Overview
The NVIDIA Jetson AGX Xavier 32GB is the volume configuration of the AGX Xavier family, used in thousands of deployed robots, medical devices and edge systems. Its 512-core NVIDIA Volta GPU with 64 Tensor Cores and two NVDLA engines delivers up to 32 TOPS while the module runs between 10W and 30W.
The 32GB of LPDDR4x memory and 32GB eMMC provide comfortable headroom for multi-model inference and video pipelines, and the 699-pin form factor is shared with every other AGX Xavier module, so the same carrier supports 8GB, 32GB, 64GB and Industrial variants.
Key Features
- 32 TOPS: 512-core Volta GPU with 64 Tensor Cores and 2x NVDLA engines
- 32GB LPDDR4x: 256-bit interface, 136.5 GB/s
- 8-core Carmel CPU: Arm v8.2 cores at up to 2.26 GHz
- 16-lane PCIe Gen4: Enough bandwidth for NVMe storage, capture cards and NICs
- Multi-camera input: 16 lanes MIPI CSI-2, up to 6 cameras
- 10W to 30W: Configurable power modes
- Mature software: Long-supported JetPack releases with CUDA, TensorRT and DeepStream
Technical Specifications
| Specification | Details |
|---|---|
| AI Performance | 32 TOPS |
| GPU | 512-core NVIDIA Volta GPU, 64 Tensor Cores |
| DL Accelerator | 2x NVDLA |
| CPU | 8-core Arm v8.2 Carmel (64-bit), up to 2.26 GHz |
| Memory | 32GB 256-bit LPDDR4x, 136.5 GB/s |
| Storage | 32GB eMMC 5.1 |
| Video Encode | 4x 4K60, 8x 4K30, 16x 1080p60 (H.265) |
| Video Decode | 2x 8K30, 6x 4K60, 12x 4K30 (H.265) |
| Camera | 16 lanes MIPI CSI-2, up to 6 cameras |
| PCIe | 16 lanes PCIe Gen4 |
| Networking | 10/100/1000BASE-T Ethernet |
| Display | 3 multi-mode DP 1.4 / eDP 1.4 / HDMI 2.0 |
| USB | 3x USB 3.1, 4x USB 2.0 |
| Power | 10W to 30W |
| Form Factor | 100 mm x 87 mm, 699-pin connector |
Ideal Use Cases
- Autonomous mobile robots and AGVs
- Smart cameras and video analytics gateways
- Medical and laboratory instruments
- Drones and inspection systems
- Replacement modules for AGX Xavier products in production
Why Choose This Product?
The 32GB AGX Xavier is the balanced member of the family: enough memory for most multi-model pipelines at a lower cost than the 64GB module, with an established supply history and a software stack your team already knows.
Interested? Contact us for pricing, volume orders and deployment support.
