AAEON unveils HPC-ARHm COM-HPC mini with Intel Core Ultra for high-performance edge Artificial Intelligence

AAEON has launched the HPC-ARHm, a COM-HPC R1.2 Mini module that brings Intel Core Ultra Series 1 and Series 2 processors to compact edge systems. The module targets low-latency, high-performance artificial intelligence workloads with integrated NPU and Arc GPU capabilities.

AAEON announced the HPC-ARHm, a COM-HPC R1.2 Mini module that supports both Series 1 and Series 2 Intel Core Ultra processors formerly known as Meteor Lake and Arrow Lake. The 95 mm x 70 mm, 80 gram module accepts 28 W CPUs across both series and offers up to 64 GB of soldered LPDDR5. The design lets users leverage the integrated Intel AI Boost NPU and the desktop-class Intel Arc GPU capabilities on each Intel Core Ultra platform, delivering up to 96 TOPs of artificial intelligence performance depending on the selected SKU.

The HPC-ARHm includes up to 256 GB of onboard NVMe storage and a range of expansion options for data-intensive, low-latency applications. Expansion features comprise one PCIe Gen 4 x8 slot and one PCIe Gen 3 x4 slot, with two lanes co-layed to provide two SATA interfaces configurable via BIOS and BOM. The module also provides four PCIe Gen 3 x1 slots intended for less-demanding peripherals. These I O and storage options are suited to workloads such as healthcare imaging, IC testing equipment, and artificial intelligence-driven robotics that require fast local processing of large datasets.

By combining a small COM-HPC Mini form factor with support for Intel Core Ultra Series 1 and Series 2 processors, the HPC-ARHm aims to deliver desktop-class graphics and on-chip neural processing at the edge. The module balances power and performance with 28 W CPU support and soldered LPDDR5 memory to sustain high throughput for inference and graphics tasks. With onboard NVMe and flexible PCIe expansion, AAEON positions the HPC-ARHm for embedded and edge computing systems that need compact hardware capable of handling demanding artificial intelligence workloads and low-latency data processing.

52

Impact Score

Samsung winds down chip lines before 18-day strike

Samsung is moving its semiconductor factories into emergency management mode ahead of an 18-day worker strike. The slowdown could disrupt global DRAM and NAND Flash supply and add pressure to an already tight memory market.

Musk and Altman clash over credibility in final trial week

The final week of the Musk v. Altman trial centered on whether Elon Musk or Sam Altman is more credible, and whether OpenAI abandoned its nonprofit mission. Jurors are now weighing competing claims over control, restructuring, and Artificial Intelligence safety.

Artificial Intelligence model learns to say it does not know

South Korean researchers developed a training method that helps Artificial Intelligence models recognize when they lack knowledge instead of responding with misplaced confidence. The approach aims to reduce hallucinations and improve reliability in areas such as autonomous driving and medicine.

Artificial Intelligence reshapes the UK jobs market

Artificial Intelligence is changing how UK businesses hire, train and structure work, with growing adoption among SMEs and rising concern over entry-level roles. The shift is increasing demand for digital skills while deepening worries about youth unemployment and long-term skills shortages.

State media shapes large language model outputs

Research in Nature finds that government control of media can influence large language model behavior through training data. The effect appears especially visible across languages, with models producing more favorable answers about China when prompted in Chinese.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.