MIC-742-AT: AI Inference System Accelerated by NVIDIA Jetson Thor
MIC-742-ATStarting at $10,000.00
Accelerated by the NVIDIA Jetson Thor module, the Advantech MIC-742-AT delivers up to 2,070 FP4 TFLOPS AI performance with 128GB high-bandwidth LPDDR5X memory. Designed for industrial-grade reliability, it offers a total system power consumption of just 150W, wide operating temperature support from -10°C to 60°C, and a compact form factor for space-constrained deployments.
- Embedded with NVIDIA Jetson T5000 up to 2070 TFLOPS(FP4)
- Supports 1 x QSFP28 (4 x 25GbE)
- Supports 1 x 5GbE, 4 x USB 3.2 Gen 2, 1 x M.2 AEkey(WiFi), 1 x M.2 Bkey(LTE)
- Support 4 x CANbus, 1 x I2C
- Support Total 8-ch GMSL2.0 with mini FAKRA connectors
Frequently Asked Questions
The MIC-742-AT provides the following I/O:
- 1 x QSFP28 port (4 x 25GbE on the Jetson T5000 SKU, 3 x 25GbE on the Jetson T4000 SKU)
- 1 x RJ45 10/100/1000/2500/5000 Mbps Ethernet
- 4 x USB 3.2 Gen 2
- 1 x Micro USB OTG
- 1 x Micro USB console (UART to USB) for debug
- 1 x HDMI display output (max resolution 3840x2160 @ 60Hz)
- 1 x audio speaker out and onboard MIC in
- 4 x CAN FD (T5000 only)
- 1 x I2C
- 1 x Nano SIM slot on the side of the board
- Support for 8-channel GMSL2.0 via mini FAKRA connectors (optional, 4 x MIPI 4-lane)
The unit accepts a power input of 19-36V DC through a 2-pin terminal block connector.
The MIC-742-AT operates from -10 to 60 °C with 0.7 m/s airflow. Operating humidity is rated at 95% at 40 °C (non-condensing), and vibration tolerance is 3 Grms at 5-500 Hz, random, 1 hour/axis. It is certified CE / FCC / CCC / KCC / BSMI / UL. Note: it does not carry RED certification.
The system includes:
- 1 x M.2 2280 M-key NVMe (PCIe x4) slot — a 1TB SSD is pre-installed on the A1 SKUs
- 2 x SATA connectors
- 1 x M.2 2230 E-key slot (PCIe x1 + USB2.0) for Wi-Fi
- 1 x M.2 3052/3042 B-key slot (USB3.0 + USB2.0) for LTE/5G
- 1 x onboard TPM
Memory is fixed to the module: 128GB LPDDR5X on the Jetson T5000 SKU or 64GB LPDDR5X on the Jetson T4000 SKU.
The MIC-742-AT runs NVIDIA Jetpack 7.0.
The unit measures 217 x 76.4 x 184.5 mm (W x D x H) and weighs 3 kg. It supports wall mount or DIN rail installation.
The MIC-742-AT has an M.2 A/E-key slot for an optional Wi-Fi 6 + BT module and an M.2 B-key slot for an optional 5G Sub6 + GNSS module. A Nano SIM slot is provided on the side of the board for cellular connectivity. These wireless modules are optional add-ons, not included by default. Note that the product does not carry RED certification.
The MIC-742-AT is offered with two NVIDIA modules:
- Jetson T5000: 14-core Arm Neoverse V3AE CPU, 2560-core Blackwell GPU with 96 fifth-gen Tensor Cores, 128GB LPDDR5X memory, 4 x 25GbE QSFP28, and 4 x CAN FD support.
- Jetson T4000: 12-core Arm Neoverse V3AE CPU, 1536-core Blackwell GPU with 64 fifth-gen Tensor Cores, 64GB LPDDR5X memory, and 3 x 25GbE QSFP28 (no CAN FD support).
Both are available as developer kit SKUs (A1, with 1TB NVMe pre-installed and JetPack 7.0 loaded) or as bare module SKUs (A2, no eMMC, no JetPack, no NVMe pre-installed).
How-to Guides
- Confirm the unit dimensions (217 x 76.4 x 184.5 mm) and weight (3 kg) fit your installation location.
- Choose wall mount or DIN rail installation, both of which are supported.
- For DIN rail installation, use the optional DIN-Rail bracket kit (part 1960120622N000) and DIN rail assembly (part 1950016395T108) with the supplied M3x6L screws.
- For general mounting, the optional mounting bracket kit (part 1960116156N001) with screws (part 1939000410) can be used.
- Identify the onboard M.2 2280 M-key slot, which supports NVMe over PCIe x4.
- Note that A1 SKUs ship with a 1TB NVMe SSD pre-installed; A2 SKUs ship without a pre-installed SSD.
- If upgrading, an optional 2TB M.2 2280 NVMe SSD (SSSTC 96FD80-P2TB-TS1, rated -25°C to 85°C) is available.
- Two SATA connectors are also available on the board for additional storage.
- Confirm your DC power source supplies 19-36V.
- Connect the power leads to the 2-pin terminal block plug (part 1652002202, included in the packing list).
- Insert the terminal block plug into the power input connector on the unit.
- Optionally, use the available 230W 24V power adapter with terminal block output (part XMIC-FSP230-AAAN3) and a region-appropriate power cord.
- Locate the M.2 A/E-key slot (2230, PCIe x1 + USB2.0) for the optional Wi-Fi 6 + BT module (part EWM-W179M201E).
- Locate the M.2 B-key slot (3052/3042, USB3.0 + USB2.0) for the optional 5G Sub6 + GNSS module (part AIW-356-DQ).
- Connect the appropriate antenna cables: SMA/MHF4L cables for Wi-Fi (part 1751000620-01) and for 5G (part 1751000623-01), with corresponding dipole/SMA antennas.
- Insert a Nano SIM card into the slot on the side of the board if using cellular connectivity.
- Apply the optional thermal pad for the 5G module (part 1990040296N010) and secure with the supplied screws (part 1930005673-11).
Accelerated by the NVIDIA Jetson Thor module, the Advantech MIC-742-AT delivers up to 2,070 FP4 TFLOPS AI performance with 128GB high-bandwidth LPDDR5X memory. Designed for industrial-grade reliability, it offers a total system power consumption of just 150W, wide operating temperature support from -10°C to 60°C, and a compact form factor for space-constrained deployments. The platform also features 8-channel GMSL 2.0 support, enabling high-speed, low-latency sensor connectivity for advanced perception systems. These capabilities make it ideal for humanoid robots, autonomous mobile robots (AMRs), surgical robotics, and other high-performance edge AI applications.
Aligning with the industry trend toward sensor standardization and Ethernet packet-based transmission, Advantech is adopting the NVIDIA Holoscan platform with NVIDIA GPUDirect RDMA, allowing sensor data to stream directly into GPU memory—bypassing CPU bottlenecks and reducing latency—to accelerate AI inference and real-time decision-making. For developers requiring real-time multimodal sensor synchronization, Advantech offers optional MIC-FG-HSBA1, powered by the Holoscan Sensor Bridge, and EKI-2712X-SPE switch accessories. These enable low-latency, high-bandwidth, and sub-millisecond alignment of data from multiple sensors—such as stereo cameras, LiDAR, radar, ultrasound, and IMUs—ensuring AI models operate on precisely timed, multimodal data for more reliable perception and decision-making in robotics.
| Manufacturer | Advantech |
|---|---|
| Short Description | Accelerated by the NVIDIA Jetson Thor module, the Advantech MIC-742-AT delivers up to 2,070 FP4 TFLOPS AI performance with 128GB high-bandwidth LPDDR5X memory. Designed for industrial-grade reliability, it offers a total system power consumption of just 150W, wide operating temperature support from -10°C to 60°C, and a compact form factor for space-constrained deployments. |
| Processor | 14 core Arm Neoverse V3AE (64-bit) SMP CPU architecture L1 Cache (I, D) per core: 64KB + 64KB, L2 1MB, L3 16MB |
| Graphics | 2560 NVIDIA CUDA Cores 96 5th Gen Tensor cores MAXN: 1.57 GHz |
| Memory | 128GB LPDDR5X |
| Video Port | HDMI (Max. resolution 3840x2160 @ 60Hz) |
| Audio | 1 x audio speaker out, MIC in on board |
| M.2 | 1 x 2280 M-key NVMe (PCIex4) |
| SATA | 2 x SATA connector |
| DC Input | 19-36V |
| Power Type | Terminal Block 2 Pin |
| Fan | 3 x SMART FAN |
| Dimension | 217 x 76.4 x 184.5 mm |
| Weight | 2kg |
| Operating Temperature | -10 ~ 60 °C with 0.7 m/s air flow |
| Humidity | 95% @ 40 °C (non-condensing) |
| Vibration | 3 Grms @ 5 ~ 500 Hz, random, 1 hr/axis |
| Max Resolution | 3840x2160 @ 60Hz |