Advantech MIB-742-AT AI Inference Systems

Advantech MIB-742-AT AI Inference Systems are high-performance platforms for Edge AI and computer vision applications. Advantech offers configurations powered by NVIDIA® Jetson Thor™ T5000 or T4000 modules, providing up to 2070 FP4 TFLOPS of AI performance. The system combines NVIDIA Blackwell architecture graphics, fifth-generation Tensor Cores, high-speed networking, and large-capacity LPDDR5X memory for demanding inference workloads.

The Advantech MIB-742-AT AI Inference Systems support 5GbE, up to 4x 25GbE, USB 3.2, M.2 expansion, and optional eight-channel GMSL2 camera connectivity. Additional interfaces include CAN FD, I2C, HDMI, audio, and SATA connections. NVIDIA JetPack™ 7.0 support provides a comprehensive software environment for developing and deploying accelerated Edge AI applications.

Features

  • NVIDIA Jetson Thor T5000 and T4000 module support
  • NVIDIA Blackwell architecture GPU with fifth-generation Tensor Cores
  • High-performance Edge AI inference capability
  • QSFP28 interface for high-bandwidth Ethernet connectivity
  • Optional eight-channel GMSL2 connectivity through mini-FAKRA connectors
  • M.2 expansion for wireless connectivity and NVMe storage
  • Multiple USB, CAN FD, I2C, display, audio, and storage interfaces
  • Integrated TPM for platform security
  • NVIDIA JetPack 7.0 software support
  • Wall-mount installation for embedded deployments

Applications

  • Edge AI inference
  • Computer vision
  • Multi-camera vision systems
  • Industrial AI
  • Intelligent video analytics
  • Robotics
  • High-bandwidth sensor processing
  • AI model development and deployment

Specifications

  • NVIDIA Jetson Thor module options
    • Jetson T5000
      • 14-core Arm® Neoverse™ V3AE 64-bit CPU
      • 2560-core NVIDIA Blackwell architecture GPU
      • 96 fifth-generation Tensor Cores
      • 128GB LPDDR5X memory
      • AI performance up to 2070 FP4 TFLOPS
    • Jetson T4000
      • 12-core Arm Neoverse V3AE 64-bit CPU
      • 1536-core NVIDIA Blackwell architecture GPU
      • 64 fifth-generation Tensor Cores
      • 64GB LPDDR5X memory
  • Ethernet connectivity
    • 1x RJ45 interface supporting 10Mbps, 100Mbps, 1000Mbps, 2500Mbps, and 5000Mbps
    • 1x QSFP28 interface
    • Up to 4x 25GbE with Jetson T5000
    • Up to 3x 25GbE with Jetson T4000
  • Display and camera connectivity
    • HDMI output with maximum resolution of 3840 x 2160 at 60Hz
    • 4x four-lane MIPI interfaces with optional eight-channel GMSL2 support
    • Mini-FAKRA connectors for GMSL2 cameras
  • USB and communication interfaces
    • 4x USB 3.2 Gen 2
    • 1x Micro-USB OTG
    • 1x Micro-USB console for UART-to-USB debugging
    • 4x CAN FD interfaces with Jetson T5000
    • 1x I2C interface
  • Audio
    • Speaker output
    • Onboard microphone input
  • Expansion
    • 1x M.2 2230 E-key with PCIe x1 and USB 2.0
    • 1x M.2 3052/3042 B-key with USB 3.0 and USB 2.0
    • 1x M.2 2280 M-key with PCIe x4 NVMe support
    • 2x SATA connectors
    • 1x Nano SIM slot
  • Power
    • Input voltage range of 19V to 36V
    • 2-pin terminal block
    • 3x smart fan headers
  • Environmental
    • Operating temperature range of -10°C to +60°C with 0.7m/s airflow
    • Operating humidity of 95% at 40°C, non-condensing
    • Vibration rating of 3Grms from 5Hz to 500Hz for one hour per axis
  • Mechanical
    • Dimensions of 180mm x 180mm
    • Weight of 1kg
    • Wall-mount installation
  • NVIDIA JetPack 7.0 operating system
  • Certifications
    • CE
    • FCC

Kit Contents

  • MIB-742-AT AI Inference SBC with NVIDIA Jetson Thor module
  • Micro-USB cable for system recovery
  • Terminal plug

Layout

Advantech MIB-742-AT AI Inference Systems
Veröffentlichungsdatum: 2026-09-03 | Aktualisiert: 2026-09-03