Description

The Waveshare NVIDIA Jetson Thor T5000 is a high-performance Jetson AGX Thor developer kit for physical AI, humanoid robotics, autonomous machines, sensor fusion, and generative AI at the edge. Waveshare-32815 is based on the Jetson Thor T5000 platform with NVIDIA Blackwell architecture, 128GB LPDDR5X memory, and up to 2070 FP4 TFLOPS of AI computing performance.

Compared with Jetson AGX Orin, the Jetson Thor T5000 platform provides up to 7.5x higher AI compute and up to 3.5x better energy efficiency. It combines a 2560-core NVIDIA Blackwell GPU with 96 fifth-generation Tensor Cores, Multi-Instance GPU support, and a 14-core Arm Neoverse-V3AE CPU for low-latency real-time AI applications.

The Jetson AGX Thor developer kit also provides advanced accelerators and high-speed I/O for robotics and physical AI workloads, including PVA v3, dual encoders, decoders, optical flow accelerator, QSFP networking, multi-GbE RJ45, USB, HDMI, DisplayPort, CAN headers, M.2 expansion, and 1TB NVMe storage. It is designed for teams building robotics AI, vision-language-action models, multi-sensor perception, and edge AI agent workflows.

Features

• Jetson Thor T5000 platform: designed for physical AI, humanoid robotics, autonomous machines, and generative AI at the edge.
• 2070 FP4 TFLOPS: delivers high AI computing performance for large generative AI models, VLMs, LLMs, and VLA workloads.
• NVIDIA Blackwell GPU: 2560-core GPU with 96 fifth-generation Tensor Cores and Multi-Instance GPU support.
• 14-core Arm Neoverse CPU: Arm Neoverse-V3AE 64-bit CPU for low-latency real-time applications.
• 128GB LPDDR5X memory: 256-bit memory with 273GB/s bandwidth for large edge AI and robotics workloads.
• High-speed networking: QSFP28 with 4x 25GbE and multi-GbE RJ45 connectivity for sensor fusion and data streaming.
• Advanced accelerators: includes PVA v3, dual encoders and decoders, and optical flow acceleration.
• NVIDIA AI software stack: supports NVIDIA JetPack SDK, Jetson Linux, CUDA, TensorRT, Isaac, Metropolis, Holoscan, and edge AI agent workflows.

Specifications

Applications

• Humanoid robotics, physical AI research, and autonomous machine development.
• Vision-language-action model deployment for robotics and real-world AI control.
• Large language models, vision language models, and generative AI inference at the edge.
• Multi-sensor fusion, high-speed perception, visual agentic AI, and video analytics.
• NVIDIA Isaac robotics, NVIDIA Metropolis visual AI, and NVIDIA Holoscan sensor-processing workflows.

Product List

Other Content

T5000 note: Waveshare-32815 is the Jetson Thor T5000 configuration with 128GB memory and up to 2070 FP4 TFLOPS. Do not confuse it with Jetson Thor T4000 configurations, which use lower AI compute and memory capacity.

Physical AI note: Jetson Thor T5000 is built for robotics and physical AI workflows that combine perception, reasoning, control, and multi-sensor real-time processing.

Software note: Confirm NVIDIA JetPack SDK version, Jetson Linux support, AI framework requirements, storage image, power budget, cooling, camera, networking, and regional compliance requirements before deployment.

Selection note: Choose this Jetson AGX Thor developer kit when a project needs Blackwell GPU acceleration, 128GB memory, high-speed networking, advanced sensor processing, and NVIDIA robotics AI software support.

Resources

• Waveshare Jetson AGX Thor Developer Kit Wiki
• NVIDIA Jetson Thor Overview
• NVIDIA JetPack SDK
• NVIDIA Jetson Linux
• NVIDIA Jetson Download Center

FAQ

What is Waveshare-32815?

Waveshare-32815 is the Waveshare NVIDIA Jetson Thor T5000 developer kit for physical AI and robotics, based on the Jetson AGX Thor Developer Kit platform.

How much AI performance does it provide?

The Jetson Thor T5000 configuration provides up to 2070 FP4 TFLOPS of AI computing performance.

How much memory does it include?

It includes 128GB 256-bit LPDDR5X memory with 273GB/s bandwidth.

What software stack does it support?

It supports NVIDIA JetPack SDK, Jetson Linux, CUDA, TensorRT, and NVIDIA physical AI workflows including Isaac, Metropolis, and Holoscan.

Is this suitable for humanoid robotics?

Yes. Jetson Thor T5000 is designed for physical AI, humanoid robotics, autonomous machines, multi-sensor fusion, and low-latency real-time AI workloads.