The fundamental physics of conventional optical fibre is approaching its limits. Light travels through glass at approximately 2/3 the speed of light in vacuum. Rayleigh scattering in the silica glass core sets a theoretical minimum attenuation of approximately 0.145 dB/km — and the best commercial fibres are already within 10% of that limit. Adding more capacity requires more wavelength channels, more complex modulation formats, and more amplifier nodes. The era of incremental improvements to solid-core silica fibre is nearing its ceiling.
Hollow-core photonic bandgap fibre (HC-PBF) and hollow-core nested antiresonant nodeless fibre (HC-NANF) fundamentally change the physics. By guiding light through air instead of glass, they achieve 99.7% of the vacuum speed of light — 47% faster propagation than SMF-28. The hollow air core eliminates Rayleigh scattering as the attenuation mechanism, achieving demonstrated attenuation of 0.174 dB/km, with theoretical limits well below 0.10 dB/km. And the near-zero Kerr nonlinearity of air (1000× lower than glass) eliminates the primary impairment in high-capacity DWDM transmission.
Hollow-Core Fibre vs. Conventional Optical Fibre
| Parameter | Conventional SMF-28 (G.652.D) | Ultra-Low-Loss SMF (G.654.E) | HC-NANF (Lumenisity/AIRGUIDE) |
|---|---|---|---|
| Propagation speed | ~200,000 km/s (66.7% c) | ~200,000 km/s (66.7% c) | ~299,000 km/s (99.7% c) |
| Attenuation @1550nm | 0.18–0.20 dB/km | 0.155–0.162 dB/km | 0.174 dB/km (demonstrated) |
| Kerr nonlinearity | High (glass: n₂ ≈2.3×10⁻²⁰ m²/W) | Same as SMF-28 | ~1000× lower (air core) |
| Latency advantage vs. SMF | Baseline | Baseline | −47% latency |
| Minimum bend radius | 7.5mm (G.657.A2) | 7.5mm | 30–50mm (more restrictive) |
| Commercial availability | Full commercial global | Full commercial global | Limited (Lumenisity pilot, 2025) |
| Connector ecosystem | Mature (LC, SC, MPO) | Mature (same as SMF) | Specialised HC connectors required |
Applications Driving Hollow-Core Fibre Adoption
- High-frequency trading: Chicago–New York microwave relay achieves ~4ms latency; SMF-28 fibre achieves ~6ms. HC-NANF would match microwave latency (~4.03ms) with far greater reliability (all-weather, no tower maintenance) and terabit-class bandwidth
- AI/GPU cluster interconnect: Distributed gradient synchronisation in large-scale AI training clusters is latency-sensitive. HC-NANF backbone between data hall buildings reduces all-reduce collective communication time for multi-node training
- Quantum key distribution (QKD): HC-NANF's near-zero nonlinearity enables co-propagation of quantum and classical optical signals in the same fibre — eliminating the need for dedicated dark fibre for QKD under India's National Quantum Mission
- 5G/6G fronthaul: Over a 20km 5G NR fronthaul span, HC-NANF reduces propagation latency from 97μs to 67μs — saving 30μs and enabling reliable compliance with the 100μs eCPRI fronthaul latency budget
- Long-haul DWDM: Lower nonlinearity allows higher signal launch power and longer amplifier spacing, increasing DWDM system capacity×distance product by 15–30% versus SMF-28
The All-Hollow-Core Network: When Air Beats Glass Everywhere
By 2035, hollow-core NANF is projected to reach production costs of 2–3× conventional OS2 — within economic reach for latency-critical infrastructure such as exchange interconnects, AI supercluster backbones, and quantum communication metropolitan rings. The remaining advantage of microwave relay (pure latency for sub-100km routes) will be eroded as HC-NANF closes the speed gap. The fibre plant installed today — with its 25–40 year physical lifetime — should be designed to accommodate HC-NANF migration: 50mm conduit bend radii, 40% fill ratios, and IEC 61300-3-35 Grade B end-face cleaning discipline now will make the hollow-core upgrade a fibre replacement exercise rather than a pathway reconstruction project.