Fiber optics, could it be the next storage?

As the market focuses on the high prosperity of the AI computing power industry chain, some seemingly “traditional” industries are quietly reaching a turning point due to supply being squeezed by AI. Optical fiber and cables are among the most typical examples.

The latest development is that signals of shortages and price increases in the optical fiber industry are becoming clearer. Guotai Haitong’s report on February 5 states that domestic mainstream G.652D loose fiber has a structural contradiction of “high prices and difficulty in obtaining goods with cash,” with longer lead times. Prices are expected to rebound in the second half of 2025, further rise in January 2026, and the proximity to operator centralized procurement and pre-holiday stocking may drive continued price increases.

The core incremental demand on the supply side comes from internal interconnection within AI data centers and DCI (Data Center Interconnect). Guotai Haitong cites CRU, stating that global data center fiber optic cable demand will increase by 75.9% year-over-year in 2025, and it is expected that the proportion of fiber demand driven by AI scenarios could rise from less than 5% in 2024 to 35% in 2027. Overseas “volume locking” is also intensifying tensions, with Corning and Meta signing long-term supply agreements worth up to $6 billion and expanding production to support their AI data center construction.

For investors, fiber optics resemble “storage that is not perfectly replicated but has ‘small storage’.” Shenwan Hongyuan Research on February 11 positioned fiber optics as “a partially restored traditional sector + short-term upward options based on AI business progress,” noting that since 2023, supply has continued to decline, confirming a cycle bottom. However, the sector’s PB ratio is at a historical high, and returns are more likely to come from fundamental realization and structural upgrades rather than purely valuation recovery.

Template of Storage: How High Prosperity “Squeezes Out Supply” and Spillover into Traditional Categories

Shenwan Hongyuan’s review states that the core of the 2025 storage market is driven by changes in AI training demand structure, prioritizing capacity allocation to high-margin products like HBM, squeezing out traditional DRAM and NAND capacity. Meanwhile, AI inference has elevated broad storage demand, leading to dramatic supply-demand shifts that double prices and valuations.

This mechanism is summarized as “four rules of high prosperity spillover”:

1. Performance growth comes from the inflection point after supply is squeezed out, with valuation growth driven by expectations of recovery from previous lows or median levels in history;

2. Spillover markets have a bottom line, requiring clear cycle bottoms, manageable demand decline risks, and sufficient supply clearance;

3. Success rates are not inherently high, needing alpha factors such as bargaining power, capacity expansion bottlenecks, and potential entry into high prosperity industries;

4. Valuation of spillover structures is difficult to break through the historical median in the long term; after cost-effectiveness declines, it is more prone to high-level oscillations.

Supply Side: Long Expansion Cycles for Optical Rods, Capacity Restructuring Compresses Traditional Supply

The “rigidity” of fiber optic supply is the foundation of the current price increase narrative. Guotai Haitong states that the core bottleneck in the fiber industry chain is the preparation of optical fiber preforms (fiber rods), with expansion cycles lasting 2 to 3 years. Since the last industry downturn, long-tail capacity of fiber rods has been largely cleared, and manufacturers’ expansion decisions are generally cautious, resulting in slow global fiber rod capacity growth and overall supply rigidity.

Meanwhile, capacity is structurally tilted toward high-demand, high-margin products. Shenwan Hongyuan notes that conventional capacity is shifting toward high-profit products for AI data centers and drones, squeezing traditional fiber capacity. Upstream fiber rod expansion cycles are long, and capacity adjustments face bottlenecks.

Guotai Haitong also states that leading manufacturers, while ensuring contractual fulfillment, continue to shift conventional G.652D capacity toward high-margin products such as G.654.E ultra-low loss, G.657A2, multi-core, and hollow-core fibers, creating a structural imbalance of “insufficient high-end capacity and squeezed traditional capacity.”

Price increases are happening: Shortage of G.652D raises spot prices and lead times

The direct carriers of price increases are the prices of loose fiber and delivery lead times. Guotai Haitong notes that the market for loose fiber is dominated by G.652D. In the first half of 2025, demand was tepid, putting pressure on prices, but they bottomed out and rebounded after June 2025. In the second half, demand for overseas G.657A2 and other products surged, prompting capacity switching. Supply of G.652D decreased, lead times lengthened, and in Q4 2025, mainstream G.652D general optical fiber saw significant price hikes and supply tightness, with some major manufacturers unable to meet orders and turning to external procurement.

In terms of prices, Guotai Haitong states that by the end of 2025, G.652D loose fiber prices had risen to 24.0–25.0 yuan per core-kilometer, with cumulative price increases exceeding 50% for the year. As of January 2026, the latest loose fiber prices exceeded 30 yuan per core-kilometer, approaching 40 yuan, more than doubling the fiber prices at the time of operator centralized procurement in 2025. They believe that as telecom and mobile operators’ procurement approaches and pre-holiday industry stocking intensifies, prices may continue to rise, and domestic large clients are likely to accept the price hikes, supported by sector volatility.

Demand Side: Computing Power Competition Shifts Fiber from “Telecom Cycle” to “Data Center Cycle”

While the cycle attributes tied to traditional telecom investments have not disappeared, AI is reconstructing marginal demand. Guotai Haitong states that the rigid demand for ultra-high bandwidth and low latency transmission in intelligent computing centers makes them the core driver of fiber demand growth. A typical GPU cluster with thousands of cards requires tens of thousands of core-kilometers of fiber for internal interconnection alone, and a single intelligent computing center’s fiber demand can be several times or even ten times that of traditional data centers.

Data also reinforces this trend. Guotai Haitong cites CRU, stating that global data center fiber optic cable demand will increase by 75.9% YoY in 2025, with North America experiencing the fastest growth driven by data centers. Guolian Minsheng’s weekly report on February 2 notes that top-tier companies face tight spot supply of quality fiber products, with some order scheduling extending to several months. The combined effects of AI data center construction and other factors have led to concentrated demand for various fibers, and the two-year expansion cycle of fiber preforms further exacerbates supply-demand mismatches.

Where are the “upward options”: Going overseas, locking in giants, and progress in hollow-core fiber industry

Fiber is not only rising in price but also shifting from “quantity” to “quality.” Guotai Haitong states that demand for specialty and multimode fibers continues to grow, and hollow-core fibers are viewed as an important direction within data centers due to their low latency and high capacity. Several manufacturers’ progress is disclosed: Longfei’s hollow-core fiber has a minimum loss of 0.05dB/km, and Hengtong Optoelectronics won a Guangdong Unicom hollow-core fiber procurement project with a total of 82 core-kilometers, with unit prices not exceeding 33,000 yuan per core-kilometer; Zhongtian Technology participated in a pilot project for hollow-core fiber with UAE operator e&, with test results showing about 30% lower transmission delay compared to traditional fibers.

Overseas, Guotai Haitong states that Microsoft and AWS are both deploying hollow-core fibers. AWS Vice President of Core Network Matt Rehder said that in the future, hollow-core fibers may replace traditional fibers as the standard choice for long-haul backbone networks, but current supply is insufficient. Demand certainty is also reflected in orders and capital expenditure: Guotai Haitong and Guolian Minsheng both mention Corning and Meta signing supply agreements worth up to $6 billion, and Guolian Minsheng citing Tencent Technology that Meta disclosed a capital expenditure of $135 billion in 2026.

Going overseas provides another prosperity cue for domestic manufacturers. Guotai Haitong states that in December 2025, exports of optical cables, fibers, and fiber rods totaled 41,000 tons, up 6.5% YoY, with export value of $450 million, up 41.8%; for the full year 2025, exports reached 454,000 tons, up 11.1%, with total export value of $4.475 billion, up 44.1%. They believe going abroad has become an important profit growth point for industry companies.

Can it become “the next storage”? Short-term success probability rises, but valuation and sustainability are more challenging

From a “mechanism” perspective, fiber optics have elements similar to storage: explosive demand from AI, bottlenecks in fiber rod expansion, capacity restructuring squeezing traditional models, price increases reflected in prices and lead times, and additional upward options from high-end products and overseas markets. Shenwan Hongyuan also believes that fiber optics have a “high short-term success rate” and views the expectation of high-end products entering North America as one of the upward options.

However, unlike storage, the valuation starting point for fiber sectors is not low. Shenwan Hongyuan explicitly points out that the PB ratio of fiber sectors is at a historical high, mainly due to previous market pricing of some companies’ optical connectivity businesses. This means the market’s outlook relies more on the continuation of price increases and fundamental visibility. If price increases spread or demand visibility fluctuates, the trend may shift from upward to high-level oscillation.

Risks include slower-than-expected AI technological progress, excessive supply expansion, demand decline beyond expectations; Guotai Haitong also warns of risks from technological updates, increased competition, raw material price fluctuations, and market demand volatility.

For investors, the key variables determining whether fiber can break out of a “storage-like” trend are whether price increases can surpass centralized procurement pricing, whether supply bottlenecks continue to ease, and whether high-end products and overseas expansion can extend prosperity from “shortage” to “structural upgrade.”

Risk Disclaimer and Legal Notice

Market risks exist; investment should be cautious. This article does not constitute personal investment advice and does not consider individual users’ specific investment goals, financial situations, or needs. Users should consider whether any opinions, views, or conclusions herein are suitable for their particular circumstances. Invest at your own risk.