LLMs and Optical Networks: A Symbiotic Relationship

2026-06-29

AI Summary

This work addresses the dual challenges posed by geographically distributed large language model (LLM) training to optical network transmission capacity and the urgent need for intelligent operations in optical networks. It presents the first systematic framework enabling bidirectional synergy between LLMs and optical networks: on one hand, it enhances optical infrastructure through a wide-area-network-aware collective communication library (WAN-aware CCL), ZR+ pluggable optical modules, and hollow-core fiber to efficiently support distributed LLM training; on the other, it introduces LLM-based autonomous management techniques to realize intelligent optical network operations. This integrated approach significantly improves communication efficiency in distributed training and fosters co-evolution of AI infrastructure and communication networks.

Abstract

This paper explores the emerging symbiosis between LLMs and optical networks. Massive LLMs require geo-distributed training, which demands advanced optical transport capabilities that require new key technical enablers, as WAN-aware CCL algorithms, ZR+ pluggables, and Hollow Core Fibers. Conversely, LLMs also enable new forms of autonomous network management.

Problem

• LLMs
• optical networks
• geo-distributed training
• autonomous network management
• WAN-aware communication

Innovation

• LLMs
• optical networks
• WAN-aware CCL
• ZR+ pluggables
• Hollow Core Fibers