New Ultra-Low-Latency Fiber for AI

Hollow-core fiber (HCF) transmits light through an air-filled or vacuum channel instead of a solid glass core. This design allows light to travel at nearly its speed in a vacuum (about 300,000 km/s), a significant increase from the approximately 200,000 km/s in conventional solid-core fiber. The result is a latency reduction of about 30-33%, or 1.54 microseconds for every kilometer the data travels. This reduction in signal delay is critical for high-frequency trading (HFT), where even microseconds can translate into a significant financial edge. In HFT, algorithms execute trades based on real-time market data; lower latency means receiving data and sending orders faster than competitors, which can directly impact profitability and reduce slippage. Major players are already deploying HCF for financial applications. euNetworks, for instance, installed a hollow-core cable to connect to the London Stock Exchange. Similarly, Comcast has deployed a 40-km hybrid cable with both hollow-core and solid-core fibers in North America. These deployments are often focused on the "last mile" connections between network access points and data centers where trading occurs. Beyond finance, the technology is crucial for scaling AI infrastructure. Distributed AI training and inference across multiple data centers are highly sensitive to latency. By reducing this delay, HCF allows for larger, more geographically dispersed AI clusters, giving operators more flexibility in data center location and design. Yangtze Optical Fibre and Cable (YOFC) has been a key developer, achieving record-low signal loss (attenuation) of 0.05dB/km and producing single fiber lengths over 20 kilometers. In June 2024, YOFC partnered with China Mobile to launch the world's first 800G hollow-core fiber transmission test network, demonstrating its application in high-capacity, real-world conditions. The focus on AI and intelligent infrastructure is a central theme for MWC Barcelona 2026. As AI workloads become more complex and distributed, the underlying network's performance becomes a critical bottleneck. Technologies like HCF are moving from experimental to essential for building the next generation of AI-powered systems. While HCF offers significant advantages, its manufacturing is more complex and costly than traditional fiber. However, ongoing advancements are improving scalability and performance, positioning it as a key enabler for future low-latency communication networks.