NTT Demonstrates 30% Energy Reduction in AI Training via Remote Distributed Deployment Leveraging IOWN Photonic Network

June 28, 2026

Techstrong.ai Japan

Overview

NTT has successfully completed a proof of concept as part of the IOWN Global Forum, demonstrating that AI training can be shifted to remote sites powered by renewable energy with minimal performance impact, cutting overall energy use by up to 30%. This achievement is realized by leveraging low-latency, all-photonic networking, marking a significant step towards sustainable AI and freeing it from geographical constraints. The results have the potential to reduce AI’s energy footprint and stimulate regional development.

In Depth

Key Findings

NTT, as part of the IOWN (Innovative Optical and Wireless Network) Global Forum, has successfully demonstrated a proof of concept (PoC) showing that AI training workloads can be distributed to remote sites utilizing renewable energy sources, resulting in up to a 30% reduction in overall energy consumption with negligible impact on AI performance. This innovative approach was made possible through the deployment of low-latency, all-photonic networks.

Technical and Market Details

The PoC aimed to shift the enormous power demand of AI training to regions with lower energy costs and abundant renewable energy. NTT’s core all-photonic network (APN) transmits optical signals end-to-end, offering ultra-low latency (light-speed transmission) and significant energy savings compared to conventional electrical signal transmission. This capability minimizes data transfer latency between remote sites and central data centers, even with geographically distributed AI training computation, thus maintaining AI computational efficiency and performance.

In the PoC, specific phases of AI workloads (e.g., data pre-processing, model training, result aggregation) were orchestrated to execute at optimal locations. This allowed for leveling out power consumption peaks and maximizing the utilization of renewable energy. The 30% energy reduction represents a concrete metric for operational cost savings in AI data centers and a significant reduction in carbon emissions, demonstrating the practical viability of the approach.

Background and Industry Context

The rapid advancement of generative AI has dramatically increased data center power consumption, raising concerns about global energy issues and environmental impact. The power footprint of AI systems continues to grow with increasing computational power, necessitating fundamental measures for sustainable AI. NTT’s IOWN concept aims to provide a comprehensive solution to this challenge, with APN as its core technology. Optimizing the geographical placement of AI resources, considering data generation/consumption locations and renewable energy supply points, is becoming a crucial design principle for next-generation data centers.

Strategic Significance and Outlook

The success of NTT’s PoC is highly significant for enhancing AI’s sustainability and geographical flexibility. Widespread adoption of this technology would enable AI data centers to maximize renewable energy utilization, substantially reducing energy costs and environmental impact. This would also facilitate AI development and service deployment in regions previously constrained by power supply or land availability, thereby contributing to regional economic revitalization. In the future, IOWN technologies are expected to be widely adopted across global AI infrastructure, accelerating the realization of a greener and more efficient digital society.

Source: https://techstrong.ai/techstrong-council/rethinking-the-geography-of-ai-to-build-a-more-sustainable-future/

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.