From left, Sung Je-hyun, head of KT Enterprise's Greater Seoul North Corporate Customer Division (executive director); Kim Nok-won, DeepX CEO; and Lee Tae-hyeon, Sesol CEO, pose for a commemorative photo after signing a business agreement at KT's Greater Seoul North Corporate Customer Division in Seoul on the 29th./Courtesy of DeepX

Artificial intelligence (AI) semiconductor corporations DeepX said on the 31st that it will jointly develop a Neural Processing Unit (NPU)-based "AI edge box" with KT and Sesol. The goal is to provide real-time AI video analysis services at electric vehicle charging stations, mobile closed-circuit (CC)TV, and buses and commercial vehicles. Sesol is a traffic and commercial vehicle terminal company.

The three companies signed a "MOU for cooperation on NPU-based on-device artificial intelligence Internet of Things (AIoT) business" on the 29th. The AI edge box is a compact computing device equipped with an AI Semiconductor that analyzes and processes data in real time on site. It applies an On-device AI method that performs computation on the device itself without sending data to the cloud.

The three companies will cooperate on: ▲ joint development of the NPU-based AI edge box ▲ promotion of the on-device AIoT business ▲ expansion of the AI video control market for transportation infrastructure such as electric vehicle charging stations, buses, and commercial vehicles ▲ creation of a domestic AI edge ecosystem.

KT will handle service commercialization and business development by leveraging its network and control platform. DeepX will provide its self-developed low-power NPU "DX-M1" and software development kit (SDK), and Sesol will be responsible for development and manufacturing of the AI edge box and field validation.

The DX-M1 is a semiconductor designed to perform AI inference on the device without cloud connection. It can be applied to industrial cameras and robots, Smart City, and transportation infrastructure. The three companies plan to use the DX-M1 to develop products capable of real-time video analysis while reducing power consumption and heat compared with graphics processing unit (GPU)-based equipment. Industrial sites with constraints on installation space and power supply are also key targets. KT plans to expand the on-device AIoT business focusing on public institutions, local governments, and traffic safety, and to identify additional services that combine AI Semiconductor and networks.

DeepX CEO Kim Nok-won said, "By combining KT's network and service capabilities, Sesol's terminal development and manufacturing technology, and DeepX's low-power AI Semiconductor technology, we will implement an AI edge solution usable on industrial sites."

Sung Je-hyun, head of KT Enterprise Division's Northern Seoul Corporate Customer Headquarters (senior vice president), said, "By combining AI Semiconductor and edge technology with KT's network and AI service capabilities, we will provide related services to industrial sites that need real-time video analysis."

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