Chief Foresight: In-depth research, technology tracing, landscape analysis, policy interpretation, and trend forecasting
Na's Observation: Sharp comments on industry hotspots, frontline observations, and application interpretationOptoelectronics Focus: Daily news headlines, disseminating highly influential optoelectronic information in the industryOptoelectronics Rendezvous: WeChat Channels live broadcast every Tuesday night (20:00-21:30), sharing trends in the optical communication industry
ICC News - Based on the industry cycle classification by the ICC Xunshi Industry and Research Institute, optical communication has officially entered the Optical Communication 3.0 development stage (2026-2030). Driven by the dual engines of AI computing intelligent interconnection and 6G ubiquitous coverage, it achieves a deep integration of communication, computing, and sensing. In the past, the growth of the optical fiber industry relied on traditional human-internet demands such as FTTH (Fiber to the Home) access and 5G carriage. Currently, with the explosion of the large model Token economy and the full launch of 6G standards, the physical bottlenecks of traditional G.652 solid-core single-mode optical fiber in bandwidth, latency, cabling density, and transmission energy consumption continue to emerge. Three new types of optical fiber media—G.654E ultra-low loss optical fiber, MCF (Multi-Core Fiber) multi-core optical fiber, and HCF (Hollow-Core Fiber) anti-resonant hollow-core optical fiber—have completed technological renewal, becoming the core carriers to resolve the performance contradictions between computing power and 6G scenarios and to build next-generation computing and network infrastructure.
Relying on the special research achievements of optical fiber media innovation from the ICC Xunshi Industry and Research Institute, this article systematically dissects the rigid demands and industrial challenges of AI intelligent computing and 6G carriage for new optical fibers. It comprehensively reviews the underlying principles, standardization processes, engineering implementation, and medium-to-long-term evolution routes of the three innovative optical fibers, and provides practical development suggestions for the entire industry chain, offering industrial reference for optical fiber preform enterprises, optical cable manufacturers, equipment vendors, and operators to evaluate medium-to-long-term track opportunities.
I. Era's Rigid Demand: Dual Engines of AI and 6G Force Comprehensive Renewal and Upgrade of Optical Fiber Media
1.1 AI Intelligent Computing Clusters Bring New Optical Interconnection Constraints
With the rapid popularization of large models and intelligent agents, the scale of Token invocations has advanced from the hundred-billion level to the million-billion level. Computing clusters continue to iterate, evolving from thousand-card clusters to ten-thousand-card and hundred-thousand-card GPU clusters. The industry has formed a three-layer computing power scaling architecture: Scale-Up for chip-level single-machine tight coupling, Scale-Out for horizontal expansion of data centers, and Scale-Across for cross-regional computing power collaborative scheduling. The traditional optical fiber system is difficult to match the stringent indicators of computing clusters, exposing three core shortcomings:
First, the contradiction between cabling density and spatial resources is prominent. The GPU capacity per cabinet has increased to hundreds of cards, and the optical fiber demand scale for computing clusters is expected to reach tens of thousands of cores by 2030. The spatial utilization rate of traditional single-core optical cables is insufficient, making data center cabling congestion and urban pipeline resource saturation common pain points;
Second, transmission latency restricts distributed training efficiency. The elongation of cross-node transmission paths and the increase in forwarding hops bring latency and jitter, directly reducing the convergence speed of parallel training for large models, which is the core shackle for performance improvement in ultra-large-scale intelligent computing clusters;
Third, nonlinear loss in high-speed transmission is difficult to avoid. After the popularization of 400G/800G coherent transmission, the nonlinear effects of traditional single-mode optical fibers are significant, limiting the transmission distance of high-order modulation. Long-distance computing interconnection requires the deployment of a large number of optical amplifiers, pushing up network-wide energy consumption and O&M costs.
According to industry data, the global demand for data center-specific optical fibers increased by 75.9% year-on-year in 2025. The proportion of optical fiber demand in computing interconnection scenarios continues to rise, gradually replacing traditional communication businesses and becoming the primary growth curve of the optical fiber market.
1.2 6G Integrated Sensing, Communication, and Computing Proposes Tenfold Performance Standards for Optical Carriage
In 2026, 3GPP officially launched the first 6G standard version R21, with the standard freeze expected to be completed in 2029. ITU-T will complete the release of the IMT-2030 evaluation in 2029-2030. Compared with 5G, the key performance indicators of 6G networks have increased by 10 to 50 times, covering four core scenarios: immersive communication, massive machine-type communications, native AI, and ultra-reliable low-latency communication. The integrated air-space-ground-sea networking architecture puts forward diversified rigid requirements for the underlying optical transmission media:
Terrestrial 6G macro stations and edge computing nodes require 50G/200G PON high-density access optical fibers; low-altitude UAVs and LEO satellite networking require inter-satellite 100G and satellite-to-ground 400G ultra-low loss transmission media; the ubiquitous integrated sensing, communication, and computing network simultaneously requires optical fibers to combine multiple characteristics such as low attenuation, ultra-large bandwidth, bend resistance, and long-distance unrepeated transmission.
The industry has reached a unified consensus: the vast majority of 6G wireless services are based on optical networks as the core carriage foundation, and the technological renewal of optical fiber media is the most certain prerequisite for the commercial landing of 6G.
1.3 Traditional Solid-Core Silica Optical Fibers Face Physical Limits That Are Difficult to Break Through
Traditional G.652.D optical fibers rely on the total internal reflection of the silica core to transmit optical signals, and the intrinsic properties of the material form a long-term performance limit: the loss is constrained by Rayleigh scattering, with the lowest commercial attenuation being about 0.18dB/km; the core area is fixed, and high-power signals are prone to nonlinear interference; a single fiber has only a single transmission channel, and capacity expansion can only add new optical cables, occupying a large amount of urban pipeline resources; the transmission speed of light in silica media is limited, and the latency compression space for cross-city computing interaction is close to its limit.
Based on the above contradictions, the industry has formed a layered and progressive media technology route: large-scale deployment of G.654E ultra-low loss optical fibers in the short term, pilot implementation of MCF multi-core optical fibers in the medium term, and cutting-edge iteration of HCF anti-resonant hollow-core optical fibers in the long term. The three types of products are complementary in different scenarios, jointly building a complete media matrix for Optical Communication 3.0.
II. Panoramic Analysis of Three Renewed Optical Fiber Media: Principles, Standards, and Engineering Implementation
2.1 G.654E Ultra-Low Loss Optical Fiber — The Optimal Solution for Computing Backbone at the Current Stage
(1) Principles and Core Parameters
G.654E is an optimized land-specific single-mode solid-core optical fiber. By adjusting the doping formula of the core layer and optimizing the drawing deposition process, it simultaneously achieves two core advantages: ultra-large effective area and ultra-low attenuation. The effective area ranges from 110-130μm², an increase of 56% compared to the general G.652.D, significantly suppressing nonlinear effects such as four-wave mixing; the commercial attenuation in the 1550nm band is stable at 0.15-0.17dB/km, with the optimal laboratory attenuation reaching 0.143dB/km. It has outstanding ultra-long-distance unrepeated transmission capability, which can greatly reduce the deployment of optical amplifiers on trunk lines.
(2) Standards and Large-Scale Live Network Applications
In 1988, ITU-T first defined the first-generation G.654 standard, which was initially only adapted for submarine communication; in 2015, ITU-T officially established the independent sub-category standard for G.654.E land optical fibers; in 2018, the domestic industry standard YD/T 3348-2018 was released, and in 2020, China Telecom built the world's first all-G.654E land long-distance trunk line. As of Q3 2025, a total of 123,000 sheath-kilometers of G.654E optical fibers have been deployed nationwide, fully covering the eight national computing hubs of the "East Data and West Computing" project. The 1,970-kilometer trunk line from Shanghai to Guangzhou and the 2,500-kilometer cross-domain computing link from Ulanqab to Guiyang have both been applied on a large scale, reducing the configuration of optical amplifiers by about 20% for the same transmission distance.
(3) Medium and Long-Term Evolution Direction
The spectrum continues to expand to the full S+C+L band, adapting to 1.6T and 3.2T next-generation ultra-high-speed coherent optical modules; integrating 4-core space-division multiplexing technology to double the transmission capacity of a single fiber; application scenarios are sinking from national long-distance backbones to computing metropolitan area networks and data center DCI interconnections, while simultaneously optimizing sea-land dual-use performance to adapt to deep-sea transoceanic submarine cable construction.
2.2 MCF Multi-Core Optical Fiber — The Core Medium-Term Solution to Break the High-Density Bandwidth Bottleneck
(1) Underlying Technical Principles
Multi-core optical fibers integrate 4, 7, or even 38 independent transmission cores within the same silica cladding, relying on spatial dimension multiplexing technology to multiply the transmission capacity. The theoretical single-fiber capacity of 4-core optical fibers is increased by 4 times, accurately solving the capacity expansion pain points such as cabling congestion in computing machine rooms and saturation of urban optical cable pipeline resources. The industry is divided into two technical routes: weakly coupled MCF is the current commercial mainstream, with a core spacing greater than 30μm and channel crosstalk ≤-45dB/100km, requiring no matching MIMO-DSP chips; strongly coupled MCF has a higher arrangement density, requires matching signal processing chips, and is aimed at ultra-large-scale dedicated computing power lines. The entire series has a unified standard cladding outer diameter of 125μm, compatible with the existing full set of engineering equipment.
(2) Standards and Benchmark Projects
In 2026, the multi-core optical fiber MSA interoperability standard was officially released, breaking through the cross-vendor compatibility barriers. Representative implementation projects include: the 7-core MCF super optical network in the Guangdong-Hong Kong-Macao Greater Bay Area, with a total line capacity exceeding 100T; Shandong built the world's first 4/7-core optical fiber hybrid live network in the same cable; 4-core MCF paired with high-density connectors increases the cabling space utilization rate in machine rooms by 4 times, perfectly adapting to ten-thousand-card intelligent computing clusters.
(3) Core Implementation Scenarios
High-density GPU interconnection in the three-layer Scale architecture of AI intelligent computing centers, 800G/1.6T ultra-high-speed capacity expansion in telecom backbones and metropolitan area networks, cross-data center DCI computing interconnection, and 6G massive base station backhaul carriage.
2.3 HCF Anti-Resonant Hollow-Core Optical Fiber — The Disruptive Cutting-Edge Media Breaking the Limits of Silica Materials
(1) Differentiated Advantages of Transmission Principles
Traditional solid-core optical fibers rely on silica to confine the optical field, and loss, latency, and nonlinearity are all constrained by the material; anti-resonant hollow-core optical fibers use air as the core transmission medium, relying on the anti-resonant effect of the microstructured cladding to confine optical signals. They possess four disruptive performances: transmission latency reduced by about 30%, commercial attenuation below 0.1dB/km, nonlinear effect attenuation reduced by nearly a thousand times, and available spectrum broadened by 10 times, which can support long-distance unrepeated transmission of single-wavelength 1.2T ultra-high-speed signals.
(2) Standards and Commercial Pilot Progress
In 2025, two domestic standard technical reports (TR) for hollow-core optical fibers completed review. In July 2026, the ITU-T SG15 Montreal plenary session approved the hollow-core optical fiber technical report TR.hcf, demonstrating the international standard discourse power of domestic enterprises. The three major operators have implemented benchmark commercial projects: China Telecom's Dongguan-Hong Kong 110km financial high-frequency dedicated line with a two-way latency of only 0.93ms; China Mobile's Ningxia 246km DCI commercial optical cable; and Gansu Telecom's Qingyang "East Data and West Computing" node has completed multi-scenario verification including pipeline, aerial, and direct burial.
(3) Evolution Route and Industrialization Bottlenecks
In the short term, focus on small-scale demonstrations for financial and high-end AI computing dedicated lines; in the medium term, penetrate metropolitan DCI after mass production yield improvement and cost reduction; in the long term, comprehensively cover long-distance trunks, transoceanic submarine cables, and satellite laser communications. At this stage, the industrialization bottlenecks are concentrated in several points: high mass production costs, the need for specialized equipment for construction and O&M, insufficient accumulation of long-term batch O&M data, and an imperfect supporting industry chain such as high-power optical amplifiers and high-speed optical modules.
III. Layered Implementation Route: Complementary Symbiosis of Three Renewed Optical Fiber Media
1. Short Term (2026-2028): Comprehensive Popularization of G.654E
Newly built national computing backbones and "East Data and West Computing" trunks are prioritized for selection, and existing old trunk lines are upgraded in batches; this optical fiber has mature technology and a complete industry chain support, which can quickly solve the pain points of long-distance transmission loss and nonlinear interference.
2. Medium Term (2028-2030): Large-Scale Volume Production of MCF Multi-Core Optical Fiber
Adapted to computing metropolitan area networks, high-density interconnection in supercomputing machine rooms, capacity expansion in pipeline-saturated areas, and 6G base station backhaul, it can multiply the transmission bandwidth without building new pipelines, significantly reducing comprehensive capacity expansion costs.
3. Long Term (Post-2030): Deep Cultivation of Extreme Scenarios by HCF Hollow-Core Optical Fiber
Focusing on high-value scenarios such as financial high-frequency trading, ultra-low latency cross-domain computing clusters, and deep-sea submarine cables, it will coexist with G.654E and MCF solid-core optical fibers in the long term with division of labor and matching.
IV. Industry Judgment: Opportunities and Potential Risks in the Renewed Optical Fiber Media Track
4.1 Core Development Opportunities
First, the underlying logic of industry demand has completely switched. Continuous investment in AI computing power, the landing of 6G standards, and the integrated air-space-ground-sea network build a long-term stable demand base. The construction cycle of preform production lines is as long as 12-18 months, and new special optical fibers will continue to maintain a tight balance of supply and demand before 2027.
Second, the domestic industry has mastered the standard dominance. China has established a fully independent industry chain for optical fiber preforms, drawing, and optical cables, deeply participated in the formulation of ITU-T, CCSA, and MSA standards, and possesses the discourse power of global technical rules in the three new optical fiber media tracks.
Third, special optical fibers open up an incremental blue ocean. The growth of the ordinary G.652 optical fiber market is gradually peaking, and the added value of G.654E, multi-core, and hollow-core optical fiber products has significantly increased, driving incremental growth across the entire chain of special preforms, high-density connectors, and supporting optical devices.
4.2 Potential Risks in the Entire Industry Chain
First, structural mismatch of capacity. After 2027, the expanded capacity of general optical fibers will be released in a concentrated manner, and ordinary single-mode fibers may fall into homogenized price competition again; while new special optical fibers have high process barriers and long capacity expansion cycles, they will continue to maintain a tight balance of supply and demand and product premium.
Second, prominent shortcomings in processes and supporting facilities. The mass production yield of hollow-core optical fiber preforms is low, and the supply of special high-purity sleeves is limited; the maturity of the industry chain for supporting MIMO chips and high-density connectors for multi-core optical fibers is insufficient, restricting large-scale live network deployment.
Third, product specifications of manufacturers are not completely unified, and there are obstacles to cross-vendor interoperability, delaying the pace of large-scale industry implementation.
Fourth, there is a risk of phased demand fluctuations. If global AI computing power capital expenditures shrink, the growth rate of data communication optical fiber demand may slow down. However, domestic "East Data and West Computing" and 6G base station construction are medium-to-long-term rigid infrastructure investments, which can hedge short-term overseas computing power cycle fluctuations, and the long-term growth logic of the industry remains unchanged.
V. Initiatives for Industrial Collaborative Development
1. Initiative 1: Relying on Associations and Chain Leaders to Jointly Tackle Common Technologies in Optical Communication 3.0
Linking national industry associations and leading enterprises in the industry chain, integrating preform, optical fiber, optical module, and equipment manufacturers, as well as the three major operators, to conduct collaborative R&D on common problems such as hollow-core optical fiber splicing, multi-core optical fiber crosstalk, and full-band broadband spectrum transmission. Multiple parties share the R&D and verification costs, reduce the trial-and-error pressure on enterprises, accelerate the commercialization of new optical fibers adapted to AI intelligent computing and 6G integrated sensing, communication, and computing scenarios, and build a solid physical foundation for the computing and network.
2. Initiative 2: Gradient Layout of Special Optical Fiber Capacity to Avoid Low-End Homogenized Competition
Combine the media renewal cycle of Optical Communication 3.0 to differentially layout production lines: traditional manufacturers expand the production of G.654E ultra-low loss optical fibers to meet current computing backbone and cross-city DCI demands; enterprises with special processes layout 4/7-core multi-core optical fibers in advance to seize the opportunities of metropolitan computing, 5G-A base station capacity expansion, and 6G base station pilot construction in 2028; leading enterprises tackle the hollow-core optical fiber preform process to improve yield and reduce costs. Strictly control the new capacity of general G.652 optical fibers and promote the transformation of the industry to high-value-added special optical fibers.
3. Initiative 3: Coordinate Domestic and International Standard Construction to Break Optical Fiber Interoperability Barriers
Domestically, accelerate the improvement of multi-core and hollow-core optical fiber standards, unify indicators such as attenuation, crosstalk, splicing, and protection, and support with engineering construction and acceptance specifications to support large-scale live network deployment; internationally, actively submit independent technical solutions, participate in the formulation of ITU-T and MSA standards, promote the globalization of domestic new optical fiber solutions, and enhance the standard discourse power of China's Optical Communication 3.0 industry.
VI. Conclusion
In the era of Optical Communication 3.0, the optical fiber industry has completed the paradigm upgrade from FTTH broadband transmission pipelines and 5G carriage links to an intelligent hub of computing and network integration. The million-billion-level Token economy of AI and the ubiquitous integrated sensing, communication, and computing construction of 6G have forced the systematic renewal of the underlying optical transmission media. In the short term, G.654E optical fibers build the long-distance arteries for computing power; in the medium term, MCF multi-core optical fibers solve the problem of high-density bandwidth capacity expansion; in the long term, HCF hollow-core optical fibers break through the physical transmission limits of silica materials.
2026-2027 is the window period for key technology iteration and industrial layout in the new optical fiber media track. The entire industry chain relies on industry associations, industry chain leaders, and standard organizations for collaborative innovation and layered gradient implementation, so as to seize the ten-year industrial dividends driven by the dual engines of AI and 6G, and steadily promote China's transition from a global optical fiber manufacturing giant to a strong country in the optical fiber media industry with complete technical discourse power.
[References]
1. "Optical Fiber Media Innovation from the Perspective of Optical Communication 3.0: Building the Physical Foundation of Computing and Network Integrating AI and 6G" by Wu Jun, 2026.07
2. ITU-T, CCSA, MSA standard documents related to new optical fibers
3. MIIT Communication Industry Statistical Bulletin, Operator Computing Trunk Engineering White Paper
[Special Note] This article is an original research achievement of the ICC Xunshi Industry and Research Institute. The copyright of all viewpoints, data, and industry judgments in the article belongs to the ICC Xunshi Industry and Research Institute. If you need to reprint, forward, excerpt, or cite the content of this article, you must obtain the prior consent of the ICC Xunshi Industry and Research Institute, and fully mark the original title, original organization, and publication time. Unauthorized private tampering, excerpting for commercial use is strictly prohibited. The analysis and judgments on market trends, technology iteration, track opportunities, etc., made in the article based on the current status of industry development are only for industrial research reference and do not constitute investment advice for any stocks, investment and financing of optical fiber industry chain projects, or industrial production decisions. Relevant entities should independently and prudently evaluate their commercial layout and investment behavior based on their actual situations.