Vecow TGS-2000F-368H Intel Core Ultra X7 368H Edge AI Computer with Arc B390 GPU, 50 TOPS NPU and Wide Temperature

Medical imaging and machine vision often lean on GPU compute as much as on the CPU or NPU, but not every job justifies a discrete card. The Vecow TGS-2000F-368H answers that within the TGS-2000 series: it is the one model with an Intel Core Ultra X7 368H and its integrated Arc B390 GPU.

The Core Ultra X7 368H is a Panther Lake-H processor with 16 cores and 16 threads and an Intel AI Boost NPU rated at 50 TOPS. Its Arc B390 GPU runs on DDR5 7200 memory, and two SO-DIMM sockets are listed for up to 128 GB. That combination lets the TGS-2000F-368H split work between CPU, GPU and NPU, which can help image pre-processing and model inference share one edge AI computer.

Networking is two Intel I226 2.5GbE ports, with Intel AMT on LAN1 for remote management. Displays run from two HDMI 2.1 outputs at up to 4096 x 2304 at 60 Hz and one DisplayPort 1.4 through the USB-C port. USB covers two USB 3.2 Gen 2 Type-A, two USB 2.0 and one USB 3.2 Gen 2×2 Type-C, with mic-in and line-out audio. Inside, an M.2 2280 M key socket on PCIe 4.0 x4 takes NVMe storage up to 4 TB, an M.2 E key socket takes an optional Wi-Fi and Bluetooth module, and an M.2 B key socket with a SIM slot takes an optional 4G or 5G module. Two docking connectors take optional TGS-101 to TGS-107 stack modules for GPIO, isolated DIO, isolated COM, USB 3, extra Gigabit LAN or 4G.

Power is 12 to 24 V DC on a three-pin terminal block. The fan-cooled TGS-2000F-368H is rated from -25 °C to 55 °C with airflow, with shock and vibration tested to IEC 61373. It carries CE, FCC and ICES marks and is designed to EN 50155 and EN 50121-3-2 requirements. We supply it configured and tested, with memory, NVMe storage and Windows 11 or Linux fitted. The TGS-2000F-386H offers a Core Ultra 9 instead, and the TGS-2500F-386H adds an optional MXM GPU layer.