NVIDIA Jetson AGX Orin development kit
4320,00 € INCL. VAT
- Up to 275 TOPS for robotics, multi-camera vision and generative AI
- AGX Orin 64 GB module: 2048-core Ampere GPU, 12-core A78AE CPU
- DisplayPort, 10 Gbit/s Ethernet, NVMe PCIe Gen4, Wi-Fi and power supply included
Business, school or public body? business account Education offers
- Description
Description
Jetson AGX Orin developer kit: up to 275 TOPS for robotics and embedded AI
The NVIDIA Jetson AGX Orin developer kit combines, in a 110 x 110 mm enclosure, an AGX Orin 64 GB module, its reference carrier board, a Wi-Fi module and the USB-C power supply. It comes out of the box with Ubuntu preinstalled on the eMMC: connect a DisplayPort monitor and a keyboard, and start developing.
It is the reference platform for autonomous robots, multi-camera video analytics and generative AI models running locally. The kit can also be reflashed to emulate the Jetson Orin NX and Orin Nano modules, so you can validate a future product on the right target.
Highlights
- Ampere GPU with 2048 CUDA cores and 64 Tensor cores : up to 275 TOPS of inference for neural networks, vision and language processing.
- 64 GB of LPDDR5 at 204.8 GB/s : enough to load large models and several simultaneous video streams.
- 12-core Arm Cortex-A78AE CPU up to 2.2 GHz for the system, robot control and preprocessing.
- Ethernet up to 10 Gbit/s : a single 10GBASE-T RJ45 port for IP cameras and dataset transfers.
- PCIe Gen4 expansion : M.2 Key M (x4) for an NVMe SSD, M.2 Key E (x1) and a PCIe x16 slot wired as x8.
- Ready to use : Ubuntu on the 64 GB eMMC, then update to the latest version of JetPack.
Technical specifications
| Features | Jetson AGX Orin (64 GB module) |
|---|---|
| IA | 275 TOPS (INT8) |
| GPU | NVIDIA Ampere, 2048 CUDA cores, 64 Tensor cores, 1.3 GHz |
| CPU | 12-core Arm Cortex-A78AE v8.2 64-bit, 2.2 GHz |
| Memory | 64 GB LPDDR5 256-bit, 204.8 GB/s |
| Storage | 64 GB eMMC 5.1 + M.2 Key M NVMe + microSD UHS-I |
| Display | 1 DisplayPort 1.4a (MST), up to 8K 30 Hz or 4K 120 Hz |
| USB | 2 USB-A 3.2 Gen2, 2 USB-A 3.2 Gen1, 2 USB-C 3.2 Gen2, 1 micro-USB (console) |
| Network | 1 RJ45 up to 10GBASE-T, Wi-Fi module supplied on M.2 Key E |
| PCIe | x16 slot (x8 Gen4), M.2 Key M (x4 Gen4), M.2 Key E (x1 Gen4) |
| Cameras | 16-lane MIPI CSI-2 connector |
| Other I/O | 40-pin GPIO (UART, SPI, I2S, I2C, CAN, PWM), 12-pin automation header, audio, JTAG, 4-pin fan, RTC battery |
| Power supply | USB-C power supply included; 5.5 x 2.5 mm DC jack, centre positive |
| Kit dimensions | 110 x 110 x 71.65 mm |
Which one should you choose?
| Platform | AI acceleration | Who is it for? |
|---|---|---|
| Jetson AGX Orin (this product) | Ampere GPU, up to 275 TOPS | robotics, multi-camera, large models, industrial prototyping |
| reComputer AI R2130-12 | Raspberry Pi 5 + Hailo-8, 26 TOPS | compact, low-power industrial vision |
| Ollama local AI kit Raspberry Pi 5 16 GB | Pi 5 processor, no dedicated accelerator | small, low-cost language model server |
Our advice: as soon as the project relies on CUDA, several cameras or models with several billion parameters, the AGX Orin is the obvious choice. For a single camera or a lightweight local assistant, a Raspberry Pi 5 solution is enough. See our comparison Raspberry Pi or NVIDIA Jetson.
Compatibility and pitfalls to avoid
- No HDMI : DisplayPort is the only video output, and USB-C does not carry video. For an HDMI monitor, use a DisplayPort to HDMI adapter.
- A single Ethernet port, 10 Gbit/s capable, and no M.2 Key B slot for a 4G/5G modem.
- First boot : for a headless setup via USB, do not connect anything to the DisplayPort.
- Updating : the eMMC may contain an older version of JetPack; NVIDIA recommends the Jetson ISO method from a USB flash drive.
- GPIO : the 40-pin connector does not guarantee compatibility with Raspberry Pi HATs.
What's in the box
- NVIDIA Jetson AGX Orin module mounted on the reference carrier board, with cooling
- Wi-Fi module installed
- USB-C power supply
- USB-C to USB-A cable
Recommended accessories
- RJ45 CAT6 Ethernet cable: to make full use of the high-speed network port and IP cameras.
- Official Raspberry Pi USB 3.0 flash drive: a handy medium for the Jetson ISO installation (check the image size).
- Official Raspberry Pi AZERTY keyboard: USB keyboard with built-in hub for the initial setup.
Uses and projects
- Autonomous robot : perception, mapping and planning on the same machine.
- Multi-stream video analytics : counting, anomaly detection or quality control on a production line.
- Local generative AI : language and vision models without sending data to the cloud. Read: deploying a local AI server in an SME.
- Product prototype : develop on the AGX Orin, then emulate an Orin NX or Orin Nano for production.
Frequently asked questions
Does the Jetson AGX Orin have an HDMI output?
No. The only video output is a DisplayPort 1.4a. Use a DisplayPort monitor or a DisplayPort to HDMI adapter.
How much AI performance does this kit offer?
The AGX Orin 64 GB module reaches 275 TOPS in INT8, with 2048 CUDA cores and 64 Tensor cores.
Does the kit come with a power supply?
Yes. NVIDIA supplies a USB-C power supply, to be connected to the USB-C port located above the DC jack.
Can I add an NVMe SSD?
Yes, in the M.2 Key M slot wired as PCIe Gen4 x4. The system can then be installed on it with the NVIDIA tools.
Which operating system is installed?
Ubuntu with the NVIDIA BSP (L4T) is preinstalled on the 64 GB eMMC. Then update to the current version of JetPack.
Can this kit be used to develop for an Orin Nano?
Yes. NVIDIA provides for reflashing the kit to emulate the Orin NX and Orin Nano modules, in both performance and power consumption.
Accelerate your embedded AI projects with the Jetson AGX Orin. Shipped within 24/48 h of payment from our stock in France, official and tested products, returns possible within 14 days with return authorization, free pickup point delivery on orders from €200 and payment by Visa, Mastercard, PayPal, Apple Pay or Google Pay. Businesses and laboratories: open a business account and discover the NVIDIA range.
