{"id":6659,"date":"2025-07-31T13:11:26","date_gmt":"2025-07-31T13:11:26","guid":{"rendered":"https:\/\/openzeka.com\/en\/?page_id=6659"},"modified":"2025-07-31T13:11:26","modified_gmt":"2025-07-31T13:11:26","slug":"nvidia-jetson-thor","status":"publish","type":"page","link":"https:\/\/openzeka.com\/en\/nvidia-jetson-thor\/","title":{"rendered":"NVIDIA Jetson Thor"},"content":{"rendered":"

NVIDIA Jetson Thor<\/h1><\/div>

The ultimate platform for physical AI and humanoid robotics.<\/p>\n<\/div>

Coming Soon<\/h4><\/div>

Get notified when NVIDIA\u00ae Jetson Thor\u2122 is available.<\/p>\n<\/div>

Notify Me<\/span><\/a><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>

NVIDIA Jetson Thor<\/h1><\/div>

The ultimate platform for physical AI and humanoid robotics.<\/p>\n<\/div>

Coming Soon<\/h4><\/div>

Get notified when NVIDIA\u00ae Jetson Thor\u2122 is available.<\/p>\n<\/div>

Notify Me<\/span><\/a><\/div><\/div><\/div><\/div>
<\/span><\/div><\/div><\/div><\/div><\/div>

A Compact Powerhouse for Advanced AI and Robotics<\/strong><\/h2><\/div>

NVIDIA Jetson Thor series modules give you the ultimate platform for physical AI and robotics, delivering up to 2070 FP4 TFLOPS of AI compute and 128 GB of memory with power configurable between 40 W and 130 W. They deliver over 7.5x higher AI compute than\u00a0NVIDIA AGX Orin\u2122, with 3.5x better energy efficiency.<\/p>\n<\/div><\/div><\/div><\/div><\/div>

Highlights<\/h2><\/div>

Industry-Leading Performance for Humanoid Robots<\/h2><\/div>

AI Performance<\/strong><\/p>\n<\/div>

2070<\/span> \u00a0TFLOPS<\/p>\n<\/div><\/div><\/div>

Memory Bandwidth<\/strong><\/p>\n<\/div>

273<\/span> \u00a0GB\/s<\/p>\n<\/div><\/div><\/div>

CPU<\/strong><\/p>\n<\/div>

14<\/span> \u00a0cores<\/p>\n<\/div><\/div><\/div><\/div>

<\/span><\/div><\/div><\/div><\/div><\/div>

Benefits<\/h2><\/div>

Blackwell GPU, Sensor Processing, and Robotic AI Software Stack<\/strong><\/h2><\/div><\/div><\/div>
<\/span><\/div>

Supercomputer for Humanoids<\/strong><\/h3><\/div>

Accelerate generative AI and large transformer models at the edge with the 2070 FP4 TFLOPS Blackwell GPU.<\/p>\n<\/div><\/div><\/div>

<\/span><\/div>

High-Speed Sensor Processing<\/strong><\/h3><\/div>

Ingest high-speed sensor data for real-time performance with 4x 25 GbE networking, a camera offload engine, and a Holoscan Sensor Bridge.<\/p>\n<\/div><\/div><\/div>

<\/span><\/div>

Robotic AI Software<\/strong><\/h3><\/div>

Discover a solution designed for humanoid robotics and physical AI applications, powered by the NVIDIA Isaac\u2122 platform and GR00T foundational models.<\/p>\n<\/div><\/div><\/div>

<\/span><\/div>

Robust Security<\/strong><\/h3><\/div>

Deliver end-to-end safety and security across the compute platform, AI models, and the entire edge-to-cloud pipeline.<\/p>\n<\/div><\/div><\/div><\/div><\/div>

Specifications<\/h2><\/div>

NVIDIA Jetson Thor Series<\/strong><\/h2><\/div>\n
\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
#<\/strong><\/th>\nJetson AGX Thor Developer Kit<\/th>\nJetson T5000<\/th>\nJetson T4000<\/th>\n<\/tr>\n<\/thead>\n
AI Performance<\/strong><\/td>\n2070 TFLOPS (FP4\u2014Sparse)<\/td>\n2070 TFLOPS (FP4\u2014Sparse)<\/td>\n1200 TFLOPS (FP4\u2014Sparse)<\/td>\n<\/tr>\n
GPU<\/strong><\/td>\n2560-core NVIDIA Blackwell architecture GPU with 96 fifth-gen Tensor Cores
\nMulti-Instance GPU (MIG) with 10 TPCs<\/td>\n		2560-core NVIDIA Blackwell architecture GPU with 96 fifth-gen Tensor Cores
\nMulti-Instance GPU (MIG) with 10 TPCs<\/td>\n		1536-core NVIDIA Blackwell architecture GPU with 64 fifth-gen Tensor Cores
\nMulti-Instance GPU (MIG) with 6 TPCs<\/td>\n<\/tr>\n
CPU<\/strong><\/td>\n14-core Arm\u00ae Neoverse\u00ae-V3AE 64-bit CPU
\n1 MB L2 cache per core
\n16 MB shared system L3 cache<\/td>\n		14-core Arm\u00ae Neoverse\u00ae-V3AE 64-bit CPU
\n1 MB L2 cache per core
\n16 MB shared system L3 cache<\/td>\n		12-core Arm\u00ae Neoverse\u00ae-V3AE 64-bit CPU
\n1 MB L2 cache per core
\n16 MB shared system L3 cache<\/td>\n<\/tr>\n
CPU Max Frequency<\/strong><\/td>\n2.6 GHz<\/td>\n2.6 GHz<\/td>\n2.6 GHz<\/td>\n<\/tr>\n
Vision Accelerator<\/strong><\/td>\n1x PVA v3<\/td>\n1x PVA v3<\/td>\n1x PVA v3<\/td>\n<\/tr>\n
Memory<\/strong><\/td>\n128 GB 256-bit LPDDR5X
\n273 GB\/s<\/td>\n		128 GB 256-bit LPDDR5X
\n273 GB\/s<\/td>\n		64 GB 256-bit LPDDR5X
\n273 GB\/s<\/td>\n<\/tr>\n
Storage<\/strong><\/td>\n1 TB NVMe M.2 Key M Slot<\/td>\nSupports NVMe through PCIe
\nSupports SSD through USB3.2<\/td>\n		Supports NVMe through PCIe
\nSupports SSD through USB3.2<\/td>\n<\/tr>\n
Video Encode<\/strong><\/td>\n6x 4Kp60 (H.265)
\n12x 4Kp30 (H.265)
\n24x 1080p60 (H.265)
\n50x 1080p30 (H.265)
\n48x 1080p30 (H.264)
\n6x 4Kp60 (H.264)<\/td>\n		6x 4Kp60 (H.265)
\n12x 4Kp30 (H.265)
\n24x 1080p60 (H.265)
\n50x 1080p30 (H.265)
\n48x 1080p30 (H.264)
\n6x 4Kp60 (H.264)<\/td>\n		6x 4Kp60 (H.265)
\n12x 4Kp30 (H.265)
\n24x 1080p60 (H.265)
\n50x 1080p30 (H.265)
\n48x 1080p30 (H.264)
\n6x 4Kp60 (H.264)<\/td>\n<\/tr>\n
Video Decode<\/strong><\/td>\n4x 8Kp30 (H.265)
\n10x 4Kp60 (H.265)
\n22x 4Kp30 (H.265)
\n46x 1080p60 (H.265)
\n92x 1080p30 (H.265)
\n82x 1080p30 (H.264)
\n4x 4Kp60 (H.264)<\/td>\n		4x 8Kp30 (H.265)
\n10x 4Kp60 (H.265)
\n22x 4Kp30 (H.265)
\n46x 1080p60 (H.265)
\n92x 1080p30 (H.265)
\n82x 1080p30 (H.264)
\n4x 4Kp60 (H.264)<\/td>\n		4x 8Kp30 (H.265)
\n10x 4Kp60 (H.265)
\n22x 4Kp30 (H.265)
\n46x 1080p60 (H.265)
\n92x 1080p30 (H.265)
\n82x 1080p30 (H.264)
\n4x 4Kp60 (H.264)<\/td>\n<\/tr>\n
Camera<\/strong><\/td>\nHSB camera via QSFP slot
\nUSB camera<\/td>\n		Up to 20 cameras via HSB
\nUp to 6 cameras through 16x lanes MIPI CSI-2
\nUp to 32 cameras using Virtual Channels
\nC-PHY 2.1 (10.25 Gbps)
\nD-PHY 2.1 (40 Gbps)<\/td>\n		Up to 20 cameras via HSB
\nUp to 6 cameras through 16x lanes MIPI CSI-2
\nUp to 32 cameras using Virtual Channels
\nC-PHY 2.1 (10.25 Gbps)
\nD-PHY 2.1 (40 Gbps)<\/td>\n<\/tr>\n
PCIe<\/strong><\/td>\nM.2 Key M slot with x4 PCIe Gen5
\nM.2 Key E slot with x1 PCIe Gen5<\/td>\n		Up to Gen5 (x8 lanes)
\nRoot port only\u2014C1 (x1) and C3 (x2)
\nRoot Point or Endpoint\u2014C2 (x1), C4 (x8), and C5 (x4)<\/td>\n		Up to Gen5 (x8 lanes)
\nRoot port only\u2014C1 (x1) and C3 (x2)
\nRoot Point or Endpoint\u2014C2 (x1), C4 (x8), and C5 (x4)<\/td>\n<\/tr>\n
USB<\/strong><\/td>\n2x USB-A (3.2 Gen2)
\n2x USB-C (3.1)<\/td>\n		xHCI host controller with integrated PHY (up to)
\n3x USB 3.2
\n4x USB 2.0<\/td>\n		xHCI host controller with integrated PHY (up to)
\n3x USB 3.2
\n4x USB 2.0<\/td>\n<\/tr>\n
Networking<\/strong><\/td>\n1x 5GBe RJ45 connector
\n1x QSFP28 (4x 25 GbE)<\/td>\n		4x 25 GbE<\/td>\n3x 25 Gbe<\/td>\n<\/tr>\n
Display<\/strong><\/td>\n1x HDMI 2.0b
\n1x DisplayPort 1.4a<\/td>\n		4x shared HDMI2.1
\nVESA DisplayPort 1.4a\u2014HBR2, MST<\/td>\n		4x shared HDMI2.1
\nVESA DisplayPort 1.4a\u2014HBR2, MST<\/td>\n<\/tr>\n
Other I\/O<\/strong><\/td>\nQSFP connector
\nM.2 Key E expansion slot (WLAN\/BT, x1 PCIe, USB2.0, UART, I2C, I2S)
\nM.2 Key M connector (NVMe for storage)
\nPCIe x4 lane, I2C, PCIe x2 lane
\n2x 13-pin CAN header
\n2x 6-pin automation header
\nLED
\nJTAG connector (2x 5-pin header)
\n1x fan connector \u201412V, PWM, and Tach
\nAudio panel header (2x 5-pin)
\nMicrofit power jack
\nRTC backup battery connector 2-pin<\/td>\n		5x I2S\/2x audio hub (AHUB), 2x DMIS, 4x UART, 4x CAN, 3x SPI, 13x I2C, 6x PWM outputs<\/td>\n5x I2S\/2x audio hub (AHUB), 2x DMIS, 4x UART, 3x SPI, 13x I2C, 6x PWM outputs**<\/td>\n<\/tr>\n
Power<\/strong><\/td>\n40 W\u2013130 W<\/td>\n40 W\u2013130 W<\/td>\n40 W\u201375 W<\/td>\n<\/tr>\n
Mechanical<\/strong><\/td>\n243.19 mm x 112.40 mm x 56.88 mm
\nThermal Transfer Plate (TTP) and optional fan or heat sink<\/td>\n		100 mm x 87 mm
\n699-pin B2B connector
\nIntegrated Thermal Transfer Plate (TTP) with heatpipe<\/td>\n		100 mm x 87 mm
\n699-pin B2B connector
\nIntegrated Thermal Transfer Plate (TTP) with heatpipe<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<\/div><\/div><\/div><\/div>

Coming Soon<\/strong><\/h2><\/div>

Get notified when Jetson Thor is available.<\/h2><\/div>
Notify Me<\/span><\/a><\/div>