COM Express module targets high-performance edge AI
ADLINK Technology Inc. introduces the Express-PTL COM Express module based on Intel Core Ultra Series 3 processors to support demanding edge AI and embedded applications.
www.adlinktech.com
ADLINK Technology Inc. has launched its first COM Express module built on Intel Core Ultra Series 3 processors, code-named Panther Lake-H. The new Express-PTL module is designed for edge AI and general embedded systems requiring high compute density, advanced AI acceleration, and industrial-grade reliability in compact form factors.
Processor architecture and AI acceleration
The Express-PTL module integrates Intel’s hybrid CPU architecture, combining performance, efficiency, and low-power efficiency cores to balance computational throughput with power efficiency. The configuration includes up to 16 CPU cores, enabling parallel processing for complex workloads.
AI acceleration is delivered through an integrated NPU 5.0 providing up to 50 TOPS of dedicated neural processing performance. In parallel, the next-generation Intel Xe3 integrated GPU delivers up to 120 TOPS of AI compute capability while supporting up to 12 Xe3 engines. This architecture is intended to handle compute-intensive inference, vision processing, and graphics workloads without the need for external accelerators.
Memory, I/O, and industrial features
The module supports up to 128 GB of DDR5 SO-DIMM memory with In-Band ECC (IBECC), enabling low-latency and reliable memory access for mission-critical systems. High-bandwidth PCIe interfaces are included to support low-latency communication with peripherals and accelerators, which is relevant for robotics and real-time industrial systems.
Industrial-temperature SKUs are specified for operation from –40°C to 85°C, addressing deployment in harsh environments. Additional industrial features include Time Coordinated Computing (TCC), Time-Sensitive Networking (TSN), extended temperature support, and functional safety-related design considerations, supporting deterministic behavior and long-term stability in industrial automation and infrastructure applications.
Application scope
The Express-PTL module is designed for a range of edge AI use cases, including medical imaging, industrial automation, robotics, and infotainment systems. The integrated Xe3 GPU simplifies system design for graphics-intensive applications by reducing the need for discrete GPUs, while the NPU enables efficient AI inference for tasks such as object recognition, navigation, and real-time decision-making in autonomous systems.
The ruggedized design and extended temperature support make the module suitable for outdoor systems and industrial edge deployments exposed to vibration and thermal stress.
Roadmap and form factor expansion
Following the Express-PTL release, ADLINK plans to extend the platform with the COM-HPC-mPTL module in a COM-HPC Mini form factor measuring 95 mm × 70 mm. This variant integrates CPU, GPU, and NPU with soldered memory to improve reliability under shock and vibration. Availability is planned for the second quarter of 2026.
Development kits for both Express-PTL and COM-HPC-mPTL, including reference carrier boards and full I/O support, are scheduled for release in Q2 2026 to support rapid prototyping and system integration.
www.adlinktech.com
