Get NVIDIA Jetson AGX Thor Developer Kit at just $3499 with 2070 TFLOPS AI Performance for Edge AI and Robotics

The NVIDIA Jetson AGX Thor Developer Kit is a high-performance AI computing platform built around the Jetson T5000 module with the latest Blackwell GPU architecture. It delivers up to 2070 TFLOPS (FP4), making it powerful enough for generative AI, robotics, autonomous machines, and multimodal AI workloads like LLMs, vision models, and real-time sensor fusion.

The kit features a 2560-core GPU with 96 Tensor Cores, a 14-core Arm Neoverse V3AE CPU, and 128GB LPDDR5X memory, ensuring both fast training and inference at the edge. It supports up to 20+ cameras, high-speed networking (4x 25GbE, 5GbE, QSFP28), 1TB NVMe storage, and rich I/O including HDMI, DisplayPort, USB 3.2/USB-C, PCIe Gen5, CAN, and GPIOs. It also excels in video processing with multi-stream 4K/8K encoding and decoding. With a power range of 40W–130W, a compact design, and an integrated thermal transfer plate, the Jetson AGX Thor ensures efficient cooling and reliable performance for demanding AI workloads.

NVIDIA Jetson Thor Developer Kit with Jetson T5000

Previously, Seeed Studio has introduced some AI PCs, including the J3011B Edge AI PC, the J3011 Edge AI PC, and others. You can check them out if you’re interested in these products.

NVIDIA Jetson AGX Thor Developer Kit Specifications:

• Module: Jetson T5000
• AI Performance: Up to 2070 TFLOPS (FP4, Sparse)
• GPU: 2560-core NVIDIA Blackwell architecture GPU with 96 Tensor Cores, Multi-Instance GPU (MIG) with 10 TPCs, up to 1.57 GHz
• CPU: 14-core Arm Neoverse-V3AE 64-bit CPU @ up to 2.6 GHz
  • 1 MB L2 cache per core
  • 16 MB shared L3 cache
• Memory: 128 GB 256-bit LPDDR5X, 273 GB/s bandwidth
• Storage: 1TB NVMe (M.2 Key M), support for external SSDs (USB 3.2)
• Display: HDMI 2.0b, DisplayPort 1.4a, 4x shared HDMI 2.1
• Vision Accelerator: 1x PVA v3
• Video Encode: up to 6x 4Kp60 (H.265/H.264), 24x 1080p60 (H.265)
• Video Decode: up to 4x 8Kp30, 10x 4Kp60, 92x 1080p30 (H.265/H.264)
• Camera Interfaces:
  • Up to 20 cameras via HSB
  • Up to 6 cameras via 16-lane MIPI CSI-2
  • Up to 32 cameras with Virtual Channels
  • C-PHY 2.1 (10.25 Gbps), D-PHY 2.1 (40 Gbps)
• Networking:
  • 1x 5GbE RJ45
  • 1x QSFP28 (4x 25GbE)
  • 4x 25GbE
• PCIe Expansion:
  • M.2 Key M (x4 PCIe Gen5)
  • M.2 Key E (x1 PCIe Gen5)
  • Additional Gen5 support up to x8 lanes
• USB:
  • 2x USB-A (3.2 Gen2)
  • 2x USB-C (3.1)
  • 3x USB 3.2
  • 4x USB 2.0
• Other I/O:
  • CAN, SPI, I2C, UART, PWM, GPIOs, Audio headers, JTAG
  • Connector: 699-pin B2B
• Power Consumption: Configurable 40W to 130W
• Cooling: Integrated Thermal Transfer Plate (TTP), optional fan or heatsink
• Dimensions: 243.19 x 112.40 x 56.88 mm (board), 100 x 87 mm (module)

NVIDIA Jetson Thor hardware overview

The NVIDIA Jetson AGX Thor Developer Kit runs on the NVIDIA Jetson Linux software stack, which provides the operating system, drivers, libraries, and tools needed to develop and deploy AI applications at the edge. The latest CUDA, cuDNN, and TensorRT are supported, enabling AI inference and training tasks to be accelerated by developers. For robotics and automation, seamless integration with NVIDIA Isaac SDK and Isaac ROS is achieved, with ready-to-use frameworks for perception, navigation, and control being offered. DeepStream SDK for video analytics, Riva for conversational AI, and Omniverse for simulation and digital twins are also supported by the platform. Developers can make use of MIG (Multi-Instance GPU) capabilities to partition the GPU for different workloads, and the built-in Transformer Engine for optimising large language models and multimodal AI tasks.

As of now, the price of the NVIDIA Jetson AGX Thor Developer Kit is $3,499 USD, and it offers volume pricing of $3,429 USD for 10+ units. It is currently out of stock, but notifications for availability can be requested from NVIDIA partners and distributors like Seed Studio, Arrow, and RS Online. You can visit the official product page for more information.

Images used courtesy of NVIDIA and Seeed Studio.