FutureMain Bolsters Global Industrial AI Competitiveness with NPU-Based Time-Series AI Project

SEOUL, South Korea, Jan. 27, 2026 /PRNewswire/ -- FutureMain Co., Ltd., a specialized industrial AI company focused on equipment diagnostics and predictive maintenance, announced that it is playing a pivotal role in a next-generation intelligent semiconductor R&D project centered on the development of integrated artificial intelligence software for processing time-series big data on NPU-based environments.

The project is part of the Next-Generation Intelligent Semiconductor Technology Development (Design) Program, within the Intelligent Semiconductor Software category. Titled "Development of an Integrated AI Processing Software Package for NPU-Based Time-Series Big Data," the initiative spans three years and nine months, from April 2023 through December 2026.

The consortium is led by Mobilint Co., Ltd., with FutureMain, Raonroad Co., Ltd., and KAIST (Korea Advanced Institute of Science and Technology) participating as co-research institutions.

The project focuses on developing an integrated software package designed to process large-scale time-series data generated in smart factory and smart city environments using CNN- and RNN-based AI models, optimized for both edge and server NPUs. The software package integrates time-series data management, security, and visualization functions, along with NPU SDK components including compilers, quantizers, and runtime libraries.

To validate real-world applicability, two types of NPU prototype boards—an Edge NPU board and a Server NPU board—are being developed and tested. The consortium aims to implement and verify demonstration applications in both smart factory and smart city domains, with active development and validation currently underway.

In the smart factory domain, the project includes the development of software for equipment and process data–based condition monitoring, anomaly detection, automated diagnostics, and visualization. In the smart city domain, efforts focus on traffic time-series data collection, average section speed prediction, application of traffic digital twin technologies, and pilot demonstrations in collaboration with local governments.

As a co-research institution responsible for the smart factory domain, FutureMain is leading research and development related to AI and security technologies based on equipment and process data. The company is responsible for defining and classifying equipment and process data, designing data collection methodologies, and developing technologies across the time-series data lifecycle. FutureMain is also overseeing training module requirement analysis, algorithm management architecture design, data anonymization and secure storage modules applicable to both edge and server environments, and the deployment of server-based AI algorithms for anomaly detection in industrial equipment and processes.

A FutureMain representative stated, "Through this project, we are advancing AI and security technologies that enable efficient processing of smart factory time-series data in NPU-based environments. We will continue our R&D efforts to ensure that anomaly detection technologies based on equipment and process data can be effectively applied in real industrial settings."

Building on the NPU-based time-series big data AI processing technologies secured through this project, FutureMain plans to further enhance technologies applicable across a wide range of industrial fields, including smart factories. In particular, the company will continue to expand the real-world applicability of AI-based anomaly detection and security technologies for equipment and process data progressively.

FutureMain is also exploring opportunities for global market expansion. With growing interest in NPU-based AI processing technologies—particularly in the United States and other global markets—the company aims to strengthen the competitiveness of its industrial AI solutions by leveraging the technological capabilities secured through this project.

SOURCE FutureMain Co., Ltd.