Advantech integrates Intel Core Ultra Series 3 for scalable Edge AI

Advantech and Intel are collaborating to integrate new edge processing architectures into industrial-grade motherboards and edge computing hardware. This joint development focuses on executing high-density computational workloads directly on factory floors, targeting sectors such as industrial automation, manufacturing, and medical diagnostics.

System Integration and Hardware Roles
The cooperation pairs processor technology from Intel with hardware engineering from Advantech. The deployment addresses the industrial challenge of running localized, high-speed data analysis without excessive reliance on remote servers. Advantech integrates Intel Core Ultra Series 3 processors into specific edge computers and motherboards, including the MIO-5381, AIMB-234, SOM-5886 module, and ARK-2252 systems. Intel provides the underlying semiconductor architecture built on its 18A process node, featuring integrated acceleration that delivers up to 180 platform tera operations per second (TOPS). This division of responsibilities ensures that standard industrial hardware can handle demanding computational tasks locally, reducing latency and mitigating cloud data transfer requirements.

Graphics and Software Architecture
To process high-resolution visual data without external graphics cards, the hardware utilizes integrated graphics processing units featuring up to 12 compute cores. This architectural configuration increases graphics processing speed by 50 percent compared to preceding hardware iterations, eliminating integration complexity associated with discrete GPUs. On the software side, Advantech provides an orchestration platform and a software development kit. This software layer includes tools for custom large language model development and embedded inference optimization, allowing system integrators to manage fleet-wide updates across decentralized digital infrastructure.

Industrial Applications and Deployment
The combined hardware and software systems are deployed in environments requiring continuous, low-latency automated decision-making. Specific industrial applications include real-time defect inspection in manufacturing, automated optical inspection using deep learning, and predictive maintenance monitoring. In medical environments, the motherboards leverage the integrated graphics and neural processing units to analyze high-resolution endoscopic video streams in real time. By processing data directly at the edge rather than transmitting it to centralized servers, the resulting systems reduce cloud bandwidth consumption, lower data transfer expenses, and improve operational stability for critical industrial processes.

Edited by Natania Lyngdoh, Induportals editor, assisted by AI.

www.advantech.com