South Korean fabless chipmaker Mobilint showcases live demonstration of large language models (LLMs) in network-free edge environments, garnering global attention

Mobilint, a leading South Korean fabless edge AI chip startup, showcased a live demonstration of a large language model (LLM) inferencing in an edge environment at CES 2025. While some AI chips that emulate cloud environments have previously demonstrated LLM capabilities, Mobilint has achieved a groundbreaking milestone as the first Korean company to showcase a fully functioning LLM use case at the edge.

At the core of this demonstration is ARIES, the fabless company’s flagship edge AI accelerator chip recognized as one of the highest-performing edge neural processing units (NPUs) globally. Designed for high energy efficiency, cost-effectiveness, and user convenience, ARIES supports a wide range of AI applications, including edge servers, smart factories, smart cities, robotics, and intelligent CCTV systems. The chip is compatible with various neural network architectures from CNNs and RNNs to LSTMs and Transformer-based models, which means it can seamlessly process LLMs and multimodal AI models.

AI Tech Insights: XAgent AI Revolutionizes Procurement with Intelligent Bidding Technology

Mobilint’s CES 2025 demonstration highlighted ARIES’s ability to reliably handle complex language model inferences with low power consumption and low latency. In addition to LLM demos, the company showcased ARIES’s wide range of application scenarios through demos of vision-language models, super-resolution applications, and object detection tasks, leaving a strong impression on attendees and industry professionals.

In tandem, Mobilint unveiled ‘REGULUS,’ an on-device AI system-on-chip (SoC), which received a CES Innovation Award in the Artificial Intelligence category. REGULUS delivers over 10 TOPS of AI performance with power consumption under 3W, making it ideal for power- and space-constrained AI applications such as drones, robots, AI-enabled CCTVs, and AI-powered IoT devices.

AI Tech Insights: Top Challenges and Risks of Large Language Models (LLMs) and How Modern Cybersecurity Technology Platforms Mitigate Them

“Our ARIES and REGULUS chips demonstrate world-class AI performance across diverse environments, from edge servers to on-device AI,” said Shin Dong-joo, CEO of Mobilint. “2025 marks a pivotal year for Mobilint as we expand our presence in the global AI chip market. We are committed to meeting the growing demand for efficient AI processing worldwide.”

The successful live demos of ARIES and REGULUS at CES 2025 underscore Mobilint’s leadership in AI chip technology and its global competitiveness. Mobilint anticipates significant revenue growth as it scales up mass production for key markets, including the U.S., Europe, Japan, and Taiwan. The company is poised to solidify its position as a global leader in AI chips.

AI Tech Insights: Global AI Launches R&D Lab, Adds 14 Senior Experts for Next-Gen Solutions

Source –  PR Newswire

To share your insights, please write to us at news@intentamplify.com