アドバンテック List of Products and Services

【Compact & Fanless Design】 With a compact size (190 x 230 x 192 mm) and a fanless design for the entire system, it can be installed in harsh environments with limited space. 【Achieving AI Inference and Computational Acceleration with MXM GPU】 In collaboration with NVIDIA, we provide an embedded GPU solution with a long-term supply of over 5 years. It is compatible with NVIDIA Quadro MXM 3.1 Type A/B form factor GPUs with a maximum of 80W, supporting up to 2,560 CUDA cores and 8.25 TFLOPS.

The "MIC-770 V3 (sold separately)" and the expansion module: "MIC-75GF10" are compatible with MXM GPU cards and provide an exceptional industrial edge AI solution.

The "AIR-150" is an industrial system equipped with a 13th generation Intel Core mobile processor and the AI engine "Hailo-8." Despite its very compact size of 156 mm in width, 112 mm in depth, and 60 mm in height, it is a high-performance model capable of executing AI processing calculations at 26 TOPS (26 trillion operations per second). To be implemented in various industry applications, this product not only offers compactness but also features a rich array of I/O ports, including COM, USB, LAN, and M.2 E-Key, ensuring seamless expansion and connectivity. Additionally, to provide peace of mind in harsh environments, it supports a wide operating temperature range of -20 to 60 degrees Celsius and complies with IEC (International Electrotechnical Commission) standards, making it a robust model. Furthermore, this product equipped with "Hailo-8" is a powerful AI system with low power consumption, making it ideal for AI-enabled applications in expanding markets such as autonomous mobile robots, baggage inspection, defect inspection, and labor safety monitoring.

We have announced the compact edge AI inference system "AIR-310," which supports MXM GPU card expansion. The "AIR-310" is designed for various industries where vision AI applications are implemented, such as smart cities, factory lines, and medical settings. It is a highly expandable system equipped with a high-performance CPU, GPU, and abundant I/O. Additionally, it offers unique development support tools to comprehensively assist customers in their AI application development.

The "AIR-520" is an edge AI server compatible with Phison's patented technology, "aiDAPTIV+", which efficiently processes large amounts of data at the edge. In addition to high-performance hardware such as AMD EPYC 7003 series processors, SQ ai100 AI SSDs, and NVIDIA GPU cards, it enables rapid deployment by providing development support tools like "AIR-520 for Gen AI Applications" and "NVIDIA AI Enterprise." The "AIR-520" is available in four models: "L13B," "L33B," "L70B," and "L70B Plus," tailored to system scale and applications.

Advantech has announced the latest GPU card "EAI-3300" equipped with the Hailo-8 AI engine. The "EAI-3300" is a PCIe-connected AI acceleration card that features two Hailo-8 VPUs. It provides 52 TOPS of AI processing capability with a maximum power consumption of 15 watts. This low-profile design card ensures easy integration into various chassis configurations, and the optimized cooling solution enables reliable operation in a temperature range of -40 to 70°C. It is compatible with Advantech's industrial PCs and embedded boards equipped with PCIe expansion slots, making it ideal for demanding AI applications such as medical image analysis, traffic monitoring, and defect inspection.

Advantech, as an NVIDIA Elite Partner, provides industrial GPU solutions specialized for NVIDIA professional graphics cards aimed at enterprises. This allows us to offer our customers early access to the latest generation of GPU cards and dedicated technical support from NVIDIA. NVIDIA Professional RTX GPU cards accelerate AI workflows in industrial settings. They are designed to provide an unprecedented edge computing experience. These cards are highly regarded as the best choice for accelerating image processing and edge AI in the manufacturing, transportation, and healthcare industries, making them the most preferred products.

- Equipped with AMD Ryzen 8040 processor, maximum 54W and AI performance of 39TOPS (CPU + iGPU + iNPU) - DDR5 5600MT/s dual-channel memory, up to 96GB - PCIe x16 slot, supports 550W GPU, built-in 850W power supply - Edge AI SDK development tools provided, supported OS: Windows 11 LTSC & Ubuntu 24.04 LTS *For more details, please download the PDF (English version) or feel free to contact us.

- Equipped with Intel Atom x7433RE, Core i3-N305, and Processor N97 - Achieves up to 26 TOPS with Hailo-8 M.2 AI module - Rich I/O: 1 (HDMI), 1 (DP), 2 (LAN), 4 (USB), 2 (COM), etc. - M.2: B key 2280, E key 2230, B key 3052 - Wide operating temperature range of -20 to 60°C and wide voltage range of 12 to 24V - Compatible with development kit for Hailo *For more details, please download the PDF (English version) or feel free to contact us.

- AMD EPYC Embedded 8004 processor, Socket SP6 (LGA 4844) - 6x DDR5 4800 MHz ECC RDIMM up to 576GB - 4x PCIe Gen5 x16 slots (CXL support on slots 4/6) - Intel X710-AT2 Dual 10GbE & I226 GbE ports - 4x MCIO interfaces + 5x USB 3.2 (Gen1) ports - 2x M.2 2280 PCIe Gen4 x4 - Built-in TPM 2.0 - Operating temperature range: 0 to 60℃ *For more details, please download the PDF (English version) or feel free to contact us.

- Compatible with LGA 6096 (Socket SP5), supports AMD EPYC Embedded 9004 processors - Equipped with DDR5 4800 MHz RDIMM - 5x PCIe x16 (Gen5), 2x PCIe x8 (Gen5), supports 4 double-deck cards - Intel X710-AT2 dual 10GbE ports - 9x SATA3 ports, 7x USB 3.2 (Gen1) ports - 2x M.2 2280/22110 (SATA / PCIe compatible) - Operating temperature range (ambient temperature): 0 to 60℃ (32 to 140℉) *For more details, please download the PDF (English version) or feel free to contact us.

- Equipped with AI engine Hailo-8 (26 TOPS) - M.2 2242 *Expandable to 2280 as an option - Operating temperature: -20 to 65°C *When using optional heatsink - Low power consumption: Standard 2.5W - Provides a rich AI development toolkit for runtime SDK integration, model building, model conversion, and pre-trained AI applications *For more details, please download the PDF (English version) or feel free to contact us.

- Equipped with NVIDIA Jetson AGX Orin, up to 275 TOPS - Compact fanless design - Compatible with NVIDIA Isaac Robot Operating System (ROS2) - Supports a total of 8 channels of GMSL 3.0/2.0, equipped with FAKRA connectors *For more details, please download the PDF (English version) or feel free to contact us.

- Equipped with a 12-core NVIDIA Arm Cortex-A78AE CPU - Ampere-based GPU with 2048 CUDA cores and 64 Tensor cores - 64GB of LPDDR5 memory and 64GB of eMMC storage - Optimized for robotics and mobility AI *For more details, please refer to the related links or feel free to contact us.*

- Equipped with NVIDIA Jetson T5000, up to 2070 TFLOPS (FP4) - Supports 1x QSFP28 (4x 25 GbE) - Supports 1 x 5GbE, 4 x USB 3.2 (Gen 2), 1 x M.2 AE key (Wi-Fi), 1 x M.2 B key (LTE) - Supports 4 x CANbus, 1 x I2C *For more details, please download the PDF (English version) or feel free to contact us.

- Equipped with NVIDIA Jetson AGX Thor, up to 2070 TFLOPS (FP4) - 1xQSFP28 (4x25GbE) - Supports 1x5GbE, 4xUSB 3.2, 1xM.2 AE key (Wi-Fi), 1xM.2 B key (LTE) - Utilization of large-scale AI models to meet a wide range of generative AI needs *For more details, please download the PDF (English version) or feel free to contact us.