Product profile
Jetson Thor
Jetson Thor is NVIDIA edge-AI computer platform for physical robotics and humanoids. It provides high-performance onboard compute for running vision-langua...
Facts
- Company
- NVIDIA
- Product type
- component
- Status
- commercial
- Website
- https://www.nvidia.com/en-us/autonomous-machines/embedded-systems/jetson-thor/
About Jetson Thor
Jetson Thor is an embedded-compute family built around NVIDIA’s Blackwell GPU and Arm Neoverse CPU architecture. The T5000 module provides up to 2,070 FP4 teraflops and 128 gigabytes of memory within a configurable 40–130-watt power range, while the T4000 provides a lower-power 64-gigabyte configuration.
High-speed Ethernet, camera interfaces, programmable vision acceleration and Multi-Instance GPU support allow perception, generative models and lower-level robot workloads to run concurrently with isolated resources. The developer kit provides a carrier board and storage for prototyping, while production modules are integrated into customer machines. Jetson software connects Thor to Isaac, GR00T, Holoscan and wider CUDA-based robotics workflows.
Market context
compute_module, official_or_source_visual, e92778e6ed37475c5f07a5d868eb52a0c8dc56d098c5b76616dfbe4b8c0743d6, https://www.nvidia.com/content/nvidiaGDC/us/en_US/autonomous-machines/embedded-systems/jetson-thor/_jcr_content/root/responsivegrid/nv_container_copy/nv_container_1143095/nv_container_852351204/nv_image.coreimg.jpeg/1773669305914/jetson-thor-highlights-ari-1920x1080.jpeg, supply_chain, clean_photo_or_studio_fallback, edge_ai_compute, https://zlsbkmaixmmnjrgpkcyp.supabase.co/storage/v1/object/public/robotics_assets/products/nvidia/nvidia-jetson-thor-official-20260523.jpg, Existing product image is already clean and suitable; generation skipped., 2026-06-02, 2026-05-23, clean_existing_product_media
Summary
Type: AI computer
Core: Edge AI compute
Function: Inference
Audience: Robot OEMs
Specifications
- ai_performance
- Up to 2070 TFLOPS FP4 sparse
- architecture
- NVIDIA Blackwell GPU-based edge AI compute platform
- camera_io
- HSB camera via QSFP slot, USB camera, up to 20 cameras via HSB, up to 6 cameras through 16 lanes MIPI CSI-2, up to 32 cameras using virtual channels
- cpu
- 14-core Arm Neoverse-V3AE 64-bit CPU
- cpu_max_frequency
- 2.6 GHz
- deployment
- Onboard physical-AI and humanoid robotics compute
- display
- Developer kit: HDMI 2.0b and DisplayPort 1.4a
- gpu
- 2560-core NVIDIA Blackwell architecture GPU with fifth-generation Tensor Cores and MIG
- mechanical
- Developer kit: 243.19 x 112.40 x 56.88 mm; module: 100 x 87 mm with 699-pin B2B connector
- memory
- 128 GB 256-bit LPDDR5X
- memory_bandwidth
- 273 GB/s
- msrp
- Jetson AGX Thor Developer Kit $3,499.00; NVIDIA Marketplace page marked out of stock USD
- networking
- Developer kit: 1x 5GbE RJ45 and 1x QSFP28 (4x 25GbE); T5000 module: 4x 25GbE
- pcie
- Developer kit: M.2 Key M x4 PCIe Gen5 and M.2 Key E x1 PCIe Gen5
- power_range
- 40 W to 130 W
- software_stack
- NVIDIA Isaac platform, GR00T workflows, Holoscan sensor processing and Jetson Linux ecosystem
- storage
- 1 TB NVMe M.2 Key M slot on developer kit; module supports NVMe through PCIe and SSD through USB 3.2
- usb
- Developer kit: 2x USB-A 3.2 Gen2 and 2x USB-C 3.1
Operations
- Product detail
- compute_module, official_or_source_visual, e92778e6ed37475c5f07a5d868eb52a0c8dc56d098c5b76616dfbe4b8c0743d6, https://www.nvidia.com/content/nvidiaGDC/us/en_US/autonomous-machines/embedded-systems/jetson-thor/_jcr_content/root/responsivegrid/nv_container_copy/nv_container_1143095/nv_container_852351204/nv_image.coreimg.jpeg/1773669305914/jetson-thor-highlights-ari-1920x1080.jpeg, supply_chain, clean_photo_or_studio_fallback, edge_ai_compute, https://zlsbkmaixmmnjrgpkcyp.supabase.co/storage/v1/...
Research, Articles and recent verified events
- AEye Apollo lidar validated on NVIDIA DRIVE AGX Thor — 2026-08-05
- NVIDIA releases Alpamayo 2 Super for commercial autonomous-driving use — 2026-08-04
- Innoviz and top-10 automaker begin InnovizTwo highway-perception programme — 2026-08-04
- NVIDIA adds agent-ready PhysicsNeMo and CUDA-X libraries to Agent Toolkit — 2026-07-26
- Fujitsu begins physical-AI business study with FANUC, Yaskawa and Kawasaki — 2026-07-16
- NVIDIA and Noetra announce 140MW Vera Rubin AI factory in Japan — 2026-07-16
- Toyota expands NVIDIA physical-AI stack across factories, vehicles and cities — 2026-07-15
- NVIDIA introduces Cosmos 3 Edge and Jetson Thor T2000/T3000 — 2026-07-15