Shin Dong-ju, CEO of Mobilint, gives a presentation at ISSCC 2026./Courtesy of Mobilint

Mobilint said on the 19th that it presented research papers on its AI Semiconductor chips "ARIES" and "REGULUS" at the International Solid-State Circuits Conference (ISSCC) 2026. It also held a live demonstration to show that its AI Semiconductor enables high-performance inference even in on-device environments. The academic conference hosted by ISSCC, the world's most prestigious society in semiconductor design, is called the "semiconductor Olympics."

Mobilint was selected as a presenter in the Highlighted Chip Releases for AI session at the event. Nvidia, Microsoft, and STMicroelectronics took part in this session.

In this presentation, Mobilint introduced that ARIES and REGULUS have a platform architecture that can scale from on-device to on-premises environments. The company said, "For Multimodal AI inference, we applied a Neural Processing Unit (NPU) architecture based on co-design of hardware and software," and noted, "We secured stable performance and power efficiency across diverse AI models through mixed-precision computation and memory-efficiency optimization." It also presented directions for handling multimodal workloads, including vision models and large language models (LLM).

Shin Dong-ju, CEO of Mobilint, said, "Releasing AI chip technologies alongside global corporations in the ISSCC Highlight session is a meaningful achievement that demonstrates the competitiveness of the Mobilint AI Semiconductor architecture," and added, "In the Multimodal AI era, we will step up our push into global markets through an AI Semiconductor platform that can scale from on-device to on-premises."

※ This article has been translated by AI. Share your feedback here.