Specifications
SoC
Intel Core Ultra 5 325 or Core Ultra 7 366H (Panther Lake)
GPU Expansion
MXM Type A GPU module slot
AI Scalability
Up to 400 TOPS hybrid CPU / GPU / NPU / MXM configuration
Architecture
Hybrid AI — allocates work between iGPU, NPU and MXM GPU by power budget
Memory
Up to 128 GB DDR5
Display
2× HDMI, USB Type-C with ALT mode
Ethernet
4× GbE
USB
4× USB 3.x, 2× USB 2.0, 1× USB-C
Serial / Fieldbus
6× COM, 2× CAN, GPIO
Expansion
M.2 3052 B-key, M.2 2230 E-key, M.2 2280 M-key, Advantech iDoor
Audio
Yes
Power Input
12–24 V DC
Operating Temperature
-20 °C to 60 °C
Dimensions
260 × 60 × 145 mm
Weight
2.5 kg
Overview
The Advantech AIR-355 is a compact edge AI inference system built on Intel Core Ultra Series 3 Panther Lake processors — Core Ultra 5 325 or Core Ultra 7 366H — with an MXM Type A slot for a discrete GPU module. Support for up to 128 GB of DDR5 and an MXM GPU takes the platform's AI capability to roughly 400 TOPS in a chassis measuring 260 × 60 × 145 mm.
Advantech's design point is a 'hybrid AI architecture': rather than running every workload on the heaviest available engine, the system allocates inference between the integrated GPU, the NPU and the MXM GPU according to the power and latency budget of each task. That keeps always-on low-rate workloads on the NPU and reserves the discrete GPU for burst demand, which cuts average power draw and total cost of ownership in always-on deployments.
I/O is aimed squarely at industrial vision and control: four Gigabit Ethernet ports, six COM channels, two CAN buses and GPIO, plus HDMI and USB-C display outputs. The 12–24 V DC input and -20 to 60 °C range let the unit be mounted in cabinets and vehicles rather than a climate-controlled rack.
Key Benefits
MXM Type A GPU slot allows the accelerator to be chosen and upgraded per application. Up to 400 TOPS from a hybrid CPU / iGPU / NPU / MXM configuration. 128 GB DDR5 supports large models locally. 4× GbE, 6× COM and 2× CAN cover camera, machine and vehicle networks. Compact 60 mm profile fits shallow cabinets.
Applications
Smart-city and traffic analytics, automated optical inspection, machine-vision cells, medical imaging consoles, and industrial AI inference where a discrete GPU must be fitted into a compact, wide-temperature chassis.
Request a Quote — ADVANTECH AIR-355 — INTEL CORE ULTRA SERIES 3 PANTHER LAKE EDGE AI SYSTEM WITH MXM GPU
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →