NVIDIA 900-13834-0080-001 | Jetson AGX Thor T5000 Module, 2070 TOPS, 128GB, Blackwell

Q: What type of applications is the Jetson AGX Thor T5000 designed for?
A: The T5000 module is designed for physical AI applications including humanoid robotics, autonomous machines, industrial automation, advanced medical systems, edge AI inference, multi-sensor fusion, and real-time generative AI deployments requiring high-performance computing in compact form factors.

Q: What is the performance improvement over previous Jetson AGX Orin modules?
A: The Jetson AGX Thor T5000 delivers 7.5x higher AI compute performance compared to AGX Orin, with 3.5x better energy efficiency. It provides 2070 FP4 TFLOPS versus Orin's capabilities, along with double the memory capacity at 128GB.

Q: What software frameworks and AI models does the T5000 support?
A: The module supports NVIDIA's full AI software stack including Isaac for robotics, Metropolis for visual agentic AI, and Holoscan for sensor processing. It can run generative AI models including Vision Language Action (VLA) models, large language models (LLMs), and vision-language models (VLMs) such as NVIDIA Isaac GR00T and Cosmos Reason.

Q: Does the T5000 module require a carrier board for operation?
A: Yes, the Jetson AGX Thor T5000 is a system-on-module that requires a compatible carrier board to provide power, I/O expansion, and physical connectivity. NVIDIA offers a reference developer kit, and ecosystem partners provide custom carrier boards for various deployment scenarios.

Q: What are the networking capabilities of the Jetson AGX Thor T5000?
A: The module features 4x 25 Gigabit Ethernet (GbE) ports for high-speed networking, enabling real-time multi-sensor data ingestion, low-latency communication for robotic systems, and integration into industrial or autonomous vehicle networks.