[JR Central] Conclusion of Basic Agreement on Joint Study for AI Edge Data Centers Utilizing the Tokaido Shinkansen Network

Central Japan Railway Company (hereinafter "JR Central"), Preferred Networks, Inc. (hereinafter "PFN"), and Actia Co., Ltd. (hereinafter "Actia") have concluded a basic agreement to conduct a joint study toward the construction of a distributed AI infrastructure by installing AI edge data centers (*1) (hereinafter "AI Edge DCs") within JR Central's business area.

In recent years, the use of AI has expanded rapidly across diverse fields. To process big data in real-time, the importance of AI Edge DCs, which offer lower latency compared to general cloud services, is increasing.

In this initiative, the companies will study the construction of a domestic distributed AI infrastructure by combining the idle land and networks along the Tokaido Shinkansen owned by the JR Central Group, the domestic AI semiconductors "MN-Core series (*2)" and AI-related technologies developed by PFN, and the know-how regarding service planning, development, and operation possessed by Actia.

Through this, the companies aim to contribute to increasing the value of regions along the railway line, such as through autonomous driving and smart city initiatives around stations, and the realization of smart factories (*4) utilizing physical AI (*3) in industrial zones.

1. Date of Conclusion March 30, 2026

2. Scope of Study ① Study on the installation feasibility and operational models of AI Edge DCs along JR Central's business area. ② Study on the design and construction of AI Edge DCs (container-type/modular-type *5). ③ Study on business models for social implementation and organization of related systems and regulations.

3. Comments from Each Company ■ Akihiko Nakamura, Representative Director and Executive Vice President, Central Japan Railway Company AI technology has become a vital foundation supporting the sophistication of social infrastructure and industry. To exert its power safely and efficiently, the development of AI infrastructure with low latency and excellent security performance is essential. Leveraging the infrastructure operation capabilities and regional cooperation cultivated through railway operations and regional development, the JR Central Group aims to contribute to increasing the value of regions along our lines by studying the construction of a safe and reliable distributed AI infrastructure together with our partner companies.

Daisuke Okanohara, President and CEO, Preferred Networks, Inc. As generative AI and physical AI enter a period of full-scale adoption, AI Edge DCs will become increasingly important as low-latency and robust distributed AI foundations. We believe that studying the construction of AI Edge DCs in close proximity to AI usage sites along the Tokaido Shinkansen—where industrial zones and cities that support Japan's economy and industry are concentrated—will contribute to strengthening the competitiveness of Japan's economy and industry, as well as increasing the value of the regions along the line. We hope this initiative will provide a foundation that makes the practical and rapid implementation of AI in industrial settings a reality.

Yukio Kitano, President and CEO, Actia Co., Ltd. In physical AI that operates autonomously in the real world, real-time responsiveness is the key to balancing high levels of safety and productivity. While the Tokaido area is a hub for Japan's leading industrial and urban functions, it is a "blank zone" for inference infrastructure far from large-scale data centers. In this initiative, we aim to contribute to increasing the value of regions along the line and strengthening industrial competitiveness by combining flexible AI Edge DCs with our company's know-how to implement infrastructure close to the point of use.

(Reference) Glossary *1: Edge Data Center A small-scale distributed data center placed near users or equipment to process data at high speed on-site. Compared to general cloud services, it achieves AI processing, IoT services, and real-time control with high efficiency by reducing communication latency. *2: MN-Core Series Domestic AI semiconductors developed by PFN. The first generation achieved the world's No. 1 power efficiency for supercomputers three times. The MN-Core L1000 for generative AI is scheduled for release in 2027. *3: Physical AI Next-generation technology that integrates AI and robotics to autonomously perceive, judge, and act in the real world. It realizes advanced automation of on-site operations. *4: Smart Factory An advanced automated factory that connects equipment and production processes within the factory using digital technology, utilizing AI, IoT, and robotics to achieve improved productivity, quality, and energy savings. *5: Container-type Data Center / Modular-type Data Center A data center where servers, power supplies, etc., are housed in containers or small units (modules) such as buildings. Compared to general data centers, they can be built in a shorter period at a lower cost, and are easy to expand or relocate.