Key Findings
A consortium comprising Nautilus Technologies, Internet Initiative Japan (IIJ), TEPCO Power Grid, Chubu Electric Power Grid, Fujitsu Limited, and 1FINITY has jointly launched a demonstration project for next-generation distributed AI processing infrastructure. This project aims to build an “All-Photonics Network (APN)” by combining existing optical fiber networks owned by electric power companies with advanced communication equipment from Fujitsu and 1FINITY, thereby validating the effectiveness of geographically distributed AI processing.
Technical Details and Demonstration Content
- All-Photonics Network (APN): An APN transmits and processes data entirely as optical signals, minimizing electrical conversions to achieve ultra-low latency and high bandwidth. This project utilizes the extensive optical fiber network of electric power companies to construct the APN.
- 1FINITY Optical Network Interface Card (NIC): This NIC, currently under development by 1FINITY, directly connects servers to the network via optical signals, enabling direct server-to-server communication over distances of 100km or more. This capability allows remote data centers or AI resources to operate as if they were co-located.
- Distributed AI Processing: By distributing AI processing across multiple data centers and edge devices, the project aims for efficient utilization of computing resources, reduced latency, and enhanced data security. The demonstration specifically validates how large-scale AI model training and inference tasks are efficiently executed over the APN.
- Collaboration with “KYOJO CUP”: Data from the “KYOJO CUP” women’s motor race will be utilized for building the demonstration environment and acquiring data. This enables the evaluation of distributed AI processing performance in real-time environments, such as motorsport.
Background and Industry Context
With the advancement of AI technology, the demands for computational resources have exponentially increased. The emergence of AI models too large for single data centers and edge AI applications requiring low latency has driven the need for geographically distributed AI processing. The optical fiber networks owned by electric power companies represent a vast infrastructure spanning across Japan, holding significant potential for building new ultra-high-speed, low-latency communication foundations. This initiative also contributes to the concretization of Japan’s proposed IOWN concept.
Strategic Significance and Outlook
The success of this demonstration project marks a crucial step towards realizing distributed AI processing infrastructure. Direct optical server-to-server connections over distances exceeding 100km will revolutionize inter-data center collaboration and significantly accelerate the development of next-generation technologies like AI, IoT, and 5G. In the future, infrastructure combining such APNs and optical NICs is expected to foster new service creation in areas such as autonomous driving, smart cities, digital twins, and advanced medical care, contributing to Japan’s enhanced digital competitiveness. Investors should watch for new market opportunities arising from the convergence of power infrastructure and communication technology.
Source: https://jp.ibtimes.com/nautilus-fujitsu-launch-kyojo-cup-linked-distributed-ai-network-trial-103070
Get our weekly technology intelligence — free
Receive an infographic that lets you judge at a glance whether each field’s analysis report is worth reading.
Subscribe Free — Weekly Tech Intelligence
By subscribing, you’ll receive Troy-Technical’s weekly technology intelligence newsletter.
- Your email and selected fields are used only to deliver the newsletter.
- We never share your information with third parties.
- You can unsubscribe anytime via the link in each email.
See our Privacy Policy for details.
Takes about a minute · Unsubscribe anytime
Comments