[Edaily Marketin YunJi Kim Reporter] iA, Inc. is partnering with FuriosaAI to build small-scale data centers equipped with domestically produced artificial intelligence (AI) semiconductors. The plan is to develop on-site AI infrastructure that can be installed more quickly than large-scale data centers and scaled as needed, and to apply it to sectors such as manufacturing, defense, and healthcare.

iA, Inc. and IA Cloud announced on the 15th that they have signed a business agreement with FuriosaAI to integrate AI semiconductors, data centers, and cloud operation technologies.

This collaboration centers on “NeoCube,” a compact AI data center platform currently under development by iA, Inc. NeoCube is a data center that integrates AI computing chips, servers, storage devices, cloud software, and power and cooling equipment into a single standardized module. It can be deployed in container form in the required locations, with capacity scalable according to demand.

The three companies plan to equip NeoCube with hardware based on FuriosaAI’s “RNGD” AI inference chip for data centers and verify its performance, power efficiency, and operational stability. They will also pursue the integration of liquid cooling technology—designed to reduce heat generated by high-density NPUs—with iA, Inc.’s cloud operations and virtualization software.

The pilot projects will take place in Hwaseong, Gyeonggi Province, and Gwangju. The Hwaseong pilot site is scheduled to be established in August, while the Gwangju site is slated for October. Each facility will be built with a power supply capacity of 150 kW and a power consumption capacity of 100 kW for IT equipment such as servers.

At the demonstration sites, the company plans to test domestically produced AI semiconductors, servers, software, and power and cooling equipment in real-world operating environments. Subsequently, the technology will be applied to services requiring real-time data processing, such as predictive maintenance for defense equipment and smart factories, as well as the bio and healthcare sectors.

This MOU was signed amid growing interest in small and edge data centers, driven by the recent rise in AI services—such as autonomous driving, robotics, and smart factories—that require immediate on-site data processing. Compared to large data centers, small and edge data centers offer the advantages of shorter construction times and the ability to expand in phases according to power and site conditions.

Meanwhile, iA, Inc. aims to increase the localization rate of the equipment and technology used in NeoCube to up to 95%. The company plans to expand its collaboration with domestic firms not only in the field of domestically produced AI semiconductors but also in servers, networks, storage devices, cloud software, and power and cooling systems.

An iA, Inc. official stated, “Through our collaboration with FuriosaAI, we will validate domestically produced NPU-based AI infrastructure in real-world environments and expand its application to sectors such as manufacturing, defense, and digital healthcare.”