• Black circuit board of nvidia igx t5000 ai module with central green to yellow gradient sensor and metallic black components
1 1

NVIDIA

NVIDIA 900-23834-0080-000 | IGX Thor T5000 AI Module, 128GB, 070 FP4 TFLOPS

900-23834-0080-000

Stock Status: Enquire

$10,689.25 inc. GST
$9,717.50 ex. GST
Sale Sold out
Shipping calculated at checkout.
20-min response

Need a quote or bulk pricing? Answered within 20 minutes, Sydney business hours.

Description

The IGX Thor T5000 is an industrial-grade edge AI system-on-module from NVIDIA, built for medical, industrial and robotics designs that need server-class Blackwell performance in a compact form. It suits medical imaging, surgical and humanoid robotics, industrial AI automation and other safety-critical edge applications, and is integrated onto custom carrier boards.

The module pairs a 2,560-core Blackwell GPU with fifth-generation Tensor Cores and Multi-Instance GPU support with a 14-core Arm Neoverse-V3AE CPU running up to 2.6 GHz. It delivers 2,070 FP4 TFLOPS (sparse) and carries 128GB of 256-bit LPDDR5X at 273 GB/s. This is the module only variant with an integrated GPU and no discrete GPU. Connectivity includes 4x 25GbE, up to 8 PCIe Gen5 lanes, USB 3.2 and HDMI 2.1, with a configurable power envelope of 40 W to 130 W. A Functional Safety Island on the SoC supports safety-critical designs.

The T5000 connects to its carrier board through a 699-pin board-to-board connector and includes an integrated thermal transfer plate with heat pipe. It runs the NVIDIA IGX software stack and supports NVIDIA AI Enterprise, NIM, Isaac, Metropolis and Holoscan, and it can run generative AI workloads including vision-language-action models and LLMs at the edge.

Enquire with XS Network Tech for new or used pricing and availability.

Features

- Server-class Blackwell GPU performance at the edge with Transformer Engine and Multi-Instance GPU for workload isolation
- Functional Safety Island inside the SoC for safety functions through the IGX software stack
- Industrial-grade hardware designed for a long lifecycle
- Runs generative AI models including VLA models such as Isaac GR00T N and LLMs and VLMs such as Cosmos Reason
- Supports NVIDIA AI Enterprise, NIM, Isaac, Metropolis and Holoscan
- Ecosystem carrier boards help accelerate medical (IEC 60601, 62304) and industrial functional safety (ISO 26262, IEC 61508) certifications
- Configurable 40 W to 130 W power envelope to balance performance and thermals

Warranty

All products sold by XS Network Tech include a 12-month warranty on both new and used items. Our in-house technical team thoroughly tests used hardware prior to sale to ensure enterprise-grade reliability.

All technical data should be verified on the manufacturer data sheets.

View full details

specs-tabs

Technical Specifications

FAQs

Technical Specifications

- Module: NVIDIA IGX T5000
- AI performance: 2,070 TFLOPS (FP4-Sparse)
- GPU: 2,560-core NVIDIA Blackwell with fifth-generation Tensor Cores
- Multi-Instance GPU: Supported, 10 TPCs
- GPU max frequency: 1.57 GHz
- CPU: 14-core Arm Neoverse-V3AE 64-bit, up to 2.6 GHz
- CPU cache: 64 KB I-cache, 64 KB D-cache, 1 MB L2 per core, 16 MB shared L3
- Memory: 128 GB 256-bit LPDDR5X, 273 GB/s
- Storage support: NVMe over PCIe, SSD over USB 3.2
- Vision accelerator: 1x PVA v3
- Video encode: 2x NVEncode
- Video decode: 2x NVDecode
- Networking: 4x 25GbE
- PCIe: Up to 8 lanes Gen5
- USB: Up to 3x USB 3.2 and 4x USB 2.0
- Display: 1x shared HDMI 2.1
- Discrete GPU: None (integrated GPU only)
- Safety: Functional Safety Island on SoC
- Power: 40 W to 130 W, configurable
- Dimensions: 100 x 87 x 15.29 mm
- Connector: 699-pin board-to-board
- Thermal: Integrated thermal transfer plate with heat pipe

FAQs

Q: Does this module include a discrete GPU?
A: No. This is the module only variant using the integrated Blackwell GPU. A discrete GPU is an option on the IGX T7000 board kit, not on the T5000 module.

Q: How does the module connect to a system?
A: It connects to a custom carrier board through a 699-pin board-to-board connector. The carrier board supplies power and exposes the module I/O.

Q: What power does it need?
A: The module has a configurable power envelope of 40 W to 130 W, supplied through the carrier board via the board-to-board connector.

Q: Does it need an NVIDIA AI Enterprise licence?
A: NVIDIA AI Enterprise support is listed for the T5000. No licence is bundled with this SKU, so confirm your software requirements before deployment.

Q: What storage can it use?
A: It supports NVMe storage over PCIe and SSD storage over USB 3.2, depending on the carrier board design.

Related Products

View all Enterprise GPUs & Graphics Cards

Recently Viewed

  • Request a Quote

    Looking for competitive pricing? Submit a request, and our team will provide a tailored quote that fits your needs.

  • Contact Us Directly

    Have a question or need immediate assistance? Call us for expert advice and real-time support.

    Call us Now  

  • Contact Us Directly

    Have a question or need immediate assistance? Call us for expert advice and real-time support.

    Contact us