Specifications
CPU
10th Generation Intel Xeon / Core i9 / i7 / i5 / i3 processor, LGA1200 socket
Chipset
Intel W480E
GPU Support
250W GPU or dual NVIDIA Tesla T4 graphics cards
Memory
2 x 260-pin DDR4 SO-DIMM, 2933/2666/2400MHz, up to 64GB, ECC or non-ECC
Graphics
Integrated Intel HD Graphics; 1 x HDMI and 1 x DB-15 analog RGB
LAN
4 x RJ-45 ports for GbE / 2.5GbE (1 x Intel i225LM 2.5GbE, 1 x Intel i219LM with iAMT 14.0, 2 x Intel i211AT)
Digital I/O
16 x DI, 16 x DO with 1.5KV isolation, DO supports 24V 200mA
Serial Ports
2 x RS-232/422/485 configurable (default RS-232), 2 x RS-232
USB Ports
2 x USB 3.2 Gen2 10Gbps, 4 x USB 3.2 Gen1 5Gbps, 1 x USB 2.0, internal USB dongle port
Storage
2 x 2.5in drive bays externally accessible; 1 x M.2 B-Key 2242/3052/2280 (PCIe x2 + USB 3.0 + SATA) shared with SIM for 5G/LTE or storage
Expansion
1 x mini-PCI Express full-size with SIM for WiFi / BT / 3G / LTE / GPS; 1 x M.2 E-Key 2230 (PCIe x1 + USB 2.0)
Audio
Mic-in / Line-out
Watchdog
1~255 levels reset
Power Input
Wide-range DC 12 ~ 36V
Operating Temperature
80W TDP CPU: -20 ~ 50°C; 65W TDP CPU: -20 ~ 55°C; 35W TDP CPU: -20 ~ 70°C; with NVIDIA Tesla T4: -20 ~ 50°C
Storage Temperature
-40 ~ 85°C
Humidity
10 ~ 95% @ 50°C non-condensing
Vibration
MIL-STD-810G Method 514.6 Category 4 with SSD
Shock
MIL-STD-810G Method 516.6 Procedure I Table 516.6-II, 20G
Certification
CE, FCC Class A
OS Support
Windows 10 IoT Enterprise 2019, Linux Kernel 4.9
Overview
The ARBOR FPC-9107-L2U4-G2 is a ruggedized edge AI computing platform designed for GPU inference in harsh or mobile environments. It pairs a 10th Generation Intel Xeon or Core processor on the W480E chipset with support for either a 250W discrete GPU or two NVIDIA Tesla T4 data-centre accelerators.
Dual T4 support is the key differentiator: two 70W T4 modules offer a combined 130 TOPS of INT8 inference and 512 tensor cores while remaining within the platform's thermal envelope, which suits multi-stream video analytics where a single large GPU would exceed the available cooling and power budget.
Industrial integration is extensive. Sixteen isolated digital inputs and sixteen digital outputs (1.5KV isolation, 24V 200mA output capability) are available directly, alongside four RJ-45 GbE/2.5GbE ports, four serial ports, and USB 3.2 Gen2 connectivity. Two externally accessible 2.5in drive bays and an M.2 B-Key slot shared between 5G/LTE modems and storage complete the picture.
The system accepts wide-range DC 12-36V input, complies with MIL-STD-810G vibration and 20G shock requirements, and operates from -20°C to 70°C depending on CPU TDP selection. It is intended for vehicle-mounted, roadside and factory-floor AI inference.
Key Benefits
Supports dual NVIDIA Tesla T4 accelerators or a single 250W GPU for high-density video inference; up to 160 isolated digital I/O points on-board; four GbE/2.5GbE ports with Intel AMT; wide-range DC 12-36V input for vehicle and battery power; MIL-STD-810G vibration and 20G shock qualification; externally accessible storage bays and 5G/LTE expansion; -20 to 70°C operating range depending on CPU TDP.
Applications
Vehicle-mounted video analytics; roadside and tolling AI systems; factory floor GPU inference; multi-channel video management servers; defence and public safety edge nodes; rail and mass transit surveillance.
Request a Quote — ARBOR FPC-9107-L2U4-G2 — RUGGED EDGE AI PLATFORM WITH 250W GPU OR DUAL NVIDIA T4
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →