Quick Answer: HCF Is Deployable, but the Interface Must Be Qualified
Hollow core fiber has demonstrated lower propagation latency and, in selected research fibers, attenuation below the best reported solid-core silica result. Those results do not make conventional connector, splice and test assumptions transferable to every HCF link.
For a pilot, separate three questions: how HCF sections are spliced to each other, how HCF transitions to standard single-mode fiber, and how the finished assembly is connectorized and tested. Require measured insertion loss, return loss, end-face criteria, power at each interface and bidirectional link-test evidence for the exact fiber design being purchased.
1. What Current HCF Evidence Proves-and What It Does Not
In June 2026, China Telecom, YOFC and Dekoli reported a field trial carrying 1.2 Tb/s per wavelength and 51.3 Tb/s aggregate capacity over a 206.5 km unrepeatered HCF span. The system used EDFA amplification without intermediate signal regeneration or remote-pumped amplification. These are participant-reported results from a specific link and system design, not a generic specification for all HCF cable or passive assemblies.
| Published result | What it supports | What it does not establish |
|---|---|---|
| 51.3 Tb/s over 206.5 km; 1.2 Tb/s per wavelength | High-capacity, long unrepeatered field transmission is feasible on the tested HCF link. | Universal connector loss, splice yield, service life or multi-vendor interoperability. |
| 33.5 dBm maximum amplifier output | The trial used a high-power amplification architecture. | That every connector in a pilot will receive 33.5 dBm. Power must be calculated at each interface. |
| Approximately 30% lower propagation latency | For a fiber-dominated latency budget, the same propagation delay can support roughly 1.45–1.5× the route length. | A guaranteed application-level latency improvement; switching, routing and processing remain in the path. |
Field-trial source: CNII coverage of the China Telecom/YOFC/Dekoli announcement.
A separate 2025 experiment reported 1.001 Tb/s over an effective distance of 10,714.28 km. It used a 146.77 km HCF recirculating loop repeated 73 times; it was not a 10,714 km installed cable. The distinction matters when laboratory transmission evidence is compared with field deployment.
2. HCF vs. Solid Core: Decision Data with Boundaries
| Dimension | Solid-core SMF | Hollow-core fiber | Procurement implication |
|---|---|---|---|
| Propagation latency | About 4.9 µs/km | About 3.3–3.4 µs/km for air-guiding designs | Value depends on how much of the end-to-end delay is fiber propagation. |
| Attenuation | Commercial cable performance is mature; the best reported silica-fiber result is about 0.14 dB/km. | A 15 km research fiber measured 0.091 dB/km at 1550 nm and below 0.1 dB/km across an 18 THz window. | Do not substitute a laboratory record for the guaranteed maximum of the purchased cable. |
| Nonlinearity | Constrained by interaction between light and glass at high power. | Research reports three to four orders of magnitude higher power tolerance and practically negligible optical nonlinearity in the tested design. | System benefit depends on amplifier, transceiver, residual SMF and splice design. |
| Spectrum | Mature components and amplification across established telecom bands. | Research fibers show broad low-loss windows, but gas absorption and design-specific transmission bands remain relevant. | Specify the guaranteed window and attenuation ripple, not a general bandwidth claim. |
| Interoperability | Mature ITU-T and IEC ecosystem. | No G.652-equivalent HCF recommendation or universal HCF interface specification as of August 2026. | Acceptance criteria must be contract-specific and based on measured data. |
| Cost | High-volume and widely quoted. | Public prices vary by fiber count, cable construction, route, termination, testing and supplier. | Compare project quotations on the same unit basis; avoid a universal cost multiplier. |
Low-loss research source: Nature Photonics, "Broadband optical fibre with an attenuation lower than 0.1 decibel per kilometre". Solid-core specifications: ITU-T G.652.
3. The Three Passive Interfaces to Specify Separately
HCF-to-HCF splice
This joint extends the same HCF design. The splice recipe must preserve the microstructure, control higher-order-mode coupling and deliver adequate mechanical strength. A loss value from one photonic-bandgap or antiresonant design is not automatically transferable to another.
HCF-to-SMF transition
This interface must manage mode-field mismatch and the glass-to-air boundary. Relevant evidence includes insertion loss in both directions, return loss, wavelength dependence, higher-order-mode content, environmental sealing and mechanical robustness. GRIN-based mode-field adapters, thermally expanded-core fibers, bridge fibers and factory-packaged transition assemblies are different architectures rather than interchangeable names.
Connectorized assembly
LC, SC and other names describe a connector format; they do not by themselves solve HCF-to-SMF mode conversion. A supplier may terminate the SMF pigtail of a qualified transition assembly with a familiar connector, but the optical transition remains a separate engineered element.
One current vendor example illustrates the need for this separation. YOFC lists a telecom HCF mode-field diameter of 21 ±2 µm at 1550 nm, an HCF-to-SMF adapter insertion-loss limit of ≤0.5 dB with return loss ≥40 dB, and HCF-to-HCF splicing "as low as 0.05 dB." These are YOFC product claims for defined products, not industry-wide limits. Independent research also shows a trade-off between coupling loss and back-reflection in HCF-to-SMF fusion splicing.
Sources: YOFC HCF and adapter specifications; ACS Photonics research on HCF-to-SMF splicing.
4. How to Assess Your Existing Passive Plant
Interface inventory
List every HCF-to-HCF splice, HCF-to-SMF transition and mated connector pair. Record the exact fiber design and supplier at each point; "LC" or "SC" alone is not an optical compatibility statement.
Splicing capability
Confirm that the proposed splicer, holders, cleaver and arc recipe have been validated for the purchased HCF. Require sample results and mechanical-strength evidence rather than relying on a generic SMF process.
Test capability
HCF backscatter can vary along the link, so conventional one-way OTDR assumptions may produce misleading event loss. Confirm bidirectional acquisition, event matching and HCF-specific post-processing.
Inspection and power safety
Calculate optical power at each accessible interface. Define inspection, cleaning, power-off and record-keeping procedures before installation, especially where local power exceeds the organization's established high-power threshold.
5. Splicing, Testing and Acceptance
Use a distribution, not one universal splice-loss number
A frequently repeated 0.16 dB value comes from a 2013 HCF photonic-bandgap-fiber-to-itself experiment. It should not be presented as a current field average for all HCF. In the 2025 10,714 km recirculating-loop experiment, the authors reported typical splice-induced losses of 0.05–0.15 dB for the selected DNANF spans. Both examples are useful evidence, but neither replaces project-specific qualification.
Sources: 2013 OFC HCF splice paper; 2025 1-Tb/s recirculating-loop paper.
Bidirectional testing is part of the link design
VIAVI identifies variable and low backscatter, reflective transitions and short-distance event separation as reasons that standard SMF OTDR analysis cannot simply be reused. Its HCF solution combines bidirectional OTDR with HCF-specific analysis; the required range and modules still depend on the link.
Testing source: VIAVI HCF testing guidance.
Minimum acceptance record
| Record | What to capture |
|---|---|
| Assembly identity | Fiber design, supplier, reel or lot, interface type and assembly serial number. |
| Optical conditions | Wavelength or band, launch condition, direction, test leads and power at the interface. |
| Loss and reflection | Bidirectional insertion/event loss where applicable, return loss and the specified statistical limit. |
| End-face evidence | Inspection image, cleaning status, acceptance zones and operator/time stamp. |
| Mechanical and environmental evidence | Splice proof or tensile method, sealing approach, bend conditions and supplier qualification report. |
| Link baseline | Final attenuation profile, event map and raw files retained for later comparison. |
IEC 61300-3-35 can inform conventional connector inspection, but an HCF project should add supplier-approved criteria for the hollow structure and transition assembly. High-power safety requirements should be based on measured or calculated power at the actual interface, following the operator's safety program and current system guidance.
6. Where HCF May Justify Its Premium-and Where It May Not
Higher-probability early applications
- Latency-sensitive metro DCI: propagation delay is a material part of the service objective, and route length or site location has measurable business value.
- Financial and timing-sensitive routes: the customer can quantify the value of microsecond-level latency improvement and can support a specialist operating model.
- Selected long unrepeatered trials: the system design can use low nonlinearity and high launch power while explicitly managing spectral ripple, gas absorption and passive-interface risk.
Cases where solid-core fiber remains the safer baseline
- Cost-sensitive access and FTTH: the mature G.652/G.657 ecosystem, field skills and multi-vendor supply chain generally outweigh HCF's latency benefit.
- Short links: switching and processing may dominate end-to-end latency, leaving too little fiber-delay saving to justify a specialist cable and test process.
- Projects requiring standardized multi-vendor acceptance: HCF standardization work is active, but no G.652-equivalent HCF recommendation is available as of August 2026.
7. A Passive-Component Supplier's View
Our view is based on a limited set of HCF-adjacent inquiries and reviews of published interface specifications; it should not be read as a market-share survey. In one recent DCI-adjacent RFQ, the buyer asked about end-face geometry, inspection evidence and high-power qualification before requesting price. That is an anecdotal signal, but it identifies the evidence a passive-component supplier should be prepared to provide.
Before describing a product as "HCF-ready," a supplier should identify the fiber and transition architecture, publish the test conditions behind loss and reflection figures, state whether each figure is typical or guaranteed, and document sealing, mechanical and power-handling limits. Conventional adapters, patch cords, panels and closures can support the surrounding SMF plant, but they should not be presented as HCF transition products without qualification data.
Glory's relevant role is to review the conventional passive baseline and evaluate custom assemblies against an agreed test plan. Current company capabilities and quality-system information are available on our OEM/ODM page and company profile. Project-specific HCF claims should be supported by a sample qualification report before production.
8. Ten-Point Checklist for a 2026–2027 HCF Pilot
| # | Requirement |
|---|---|
| 1 | Map every HCF-to-HCF splice, HCF-to-SMF transition and mated connector pair separately. |
| 2 | Record the HCF design, mode-field diameter, transmission band and supplier for every link section. |
| 3 | Set insertion-loss and return-loss limits for each interface type; do not use one universal HCF connector figure. |
| 4 | Require the splice recipe, equipment configuration, sample distribution and mechanical-strength evidence. |
| 5 | Calculate power at every accessible interface and define power-off, inspection and cleaning procedures. |
| 6 | Define end-face acceptance criteria with the HCF and transition-assembly supplier. |
| 7 | Use bidirectional link testing and retain raw files, event maps and inspection images. |
| 8 | Specify sealing and contamination controls for transport, installation, re-entry and storage. |
| 9 | Compare quotations on the same basis: fiber count, cable construction, terminations, testing, training, spares and support. |
| 10 | Write acceptance around measured project data and schedule a review when HCF-specific standards or supplier revisions change. |
Procurement FAQ
Q: How much lower is HCF propagation latency?
A: Representative air-guiding designs are approximately 30% lower in propagation latency than solid-core SMF-about 3.3–3.4 µs/km versus about 4.9 µs/km. Application-level improvement will be smaller when switching, routing and processing dominate the path.
Q: Is hollow core fiber standardized?
A: HCF is under active ITU-T discussion, but there is no mature G.652-equivalent HCF recommendation or universal HCF passive-interface specification as of August 2026. Use project-specific, measured acceptance criteria.
Q: Can an HCF link use LC or SC connectors?
A: The SMF side of a qualified transition assembly may use a familiar connector format. LC or SC identifies the connector format; it does not replace the HCF-to-SMF mode-field and glass-to-air transition.
Q: What splice loss should be budgeted?
A: There is no universal value. Published results vary by HCF design and joint type. Ask for the supplier's measured distribution for the exact fiber, recipe, wavelength and environment, then use an agreed maximum-not a cross-industry average-in the link budget.
Q: Can a standard OTDR test HCF?
A: Some OTDR hardware may be usable, but standard SMF settings and one-way analysis are not sufficient for many HCF links. Confirm dynamic range at the required pulse width, bidirectional acquisition and HCF-specific event analysis.
Q: What does an HCF link cost?
A: Public comparisons are not consistent enough for a universal multiplier. Obtain like-for-like quotations covering fiber count, cable construction, HCF-to-SMF transitions, splicing, test equipment or services, training, spare assemblies and support.
Q: Is HCF ready for FTTH/PON?
A: Not as a general replacement for G.652/G.657 access fiber. FTTH prioritizes cost, field familiarity, bend performance, standardized components and multi-vendor availability; HCF's latency benefit is usually less valuable in this use case.
Q: Is long-term HCF reliability proven?
A: Operational deployments exist, but HCF does not yet have the same decades-long field record or standardized multi-vendor qualification framework as conventional SMF. Review the exact cable, splice and transition design, environmental test evidence, sealing method and restoration plan.
Primary and Technical References
- CNII - China Telecom/YOFC/Dekoli 206.5 km field-trial coverage (June 2026)
- Nature Photonics - Broadband optical fibre with attenuation below 0.1 dB/km
- Ge et al. - 1-Tb/s/λ transmission over a 10,714 km effective recirculating distance
- Microsoft Azure - Deployment of HCF in Azure's network
- YOFC - Hollow core fibre, adapter and splicing product specifications
- ACS Photonics - HCF-to-SMF splicing loss and back-reflection research
- OFC 2013 - Robust low-loss HCF photonic-bandgap-fiber splicing
- VIAVI - Hollow core fiber testing guidance
- ITU-T SG15 - 2026 HCF standardization status
- ITU-T G.652 - Characteristics of a single-mode optical fibre and cable
- ITU-T G Supplement 39 - Optical-system engineering and high-power safety considerations
Results remain design-, wavelength- and test-condition-specific. Confirm current document editions and the purchased supplier's instructions before issuing a final specification.
Planning an HCF Pilot?
Send the proposed fiber design, interface map, wavelength range, power budget and acceptance requirements. Glory can review the conventional passive baseline and define the evidence required before a custom assembly is described as HCF-ready.
Request an Engineering Review Contact the Technical Team
About the author: Glory Optical Communication is a Ningbo-based manufacturer of passive fiber-optic components serving FTTH/ODN, data-center cabling and OEM/ODM programs. External performance figures in this article are attributed to their published sources; project specifications should be based on the exact fiber and assembly being purchased.
