Fiber Optic Shortage Enters a New Phase: Is the G.657.A2 Buying Window Closing?

The latest Fiber Optic Shortage is beginning to look different from the industry’s familiar boom-and-bust cycles.

In previous shortages, a wave of telecom orders would tighten supply, manufacturers would add capacity, and the market would eventually swing back toward oversupply. This time, demand no longer has a single center of gravity. AI data centers are increasing fiber density at extraordinary speed. Polarization-maintaining fiber is moving into a major expansion cycle. Fiber-optic-controlled Drone systems are creating recurring material consumption. Traditional broadband, industrial communication and infrastructure projects have not stopped buying.

The issue, then, is no longer simply how much fiber the world will need. The more immediate question is how scarce manufacturing capacity will be allocated.

When high-margin products and the world’s largest technology buyers begin reserving output years in advance, how much G.657.A2 capacity will still be available to ordinary buyers?

AI Giants Are Buying More Than Fiber — They Are Buying Future Capacity

In January 2026, Meta entered a multiyear agreement worth up to $6 billion for optical fiber, cable and connectivity products. A few months later, NVIDIA established another long-term manufacturing partnership with the same global supplier.

According to Corning’s January 2026 announcement and NVIDIA’s partnership announcement, the expansion plan includes a tenfold increase in U.S. optical-connectivity manufacturing capacity, more than a 50% increase in domestic fiber output, three new facilities in North Carolina and Texas, and more than 3,000 additional jobs.

At first glance, those figures appear to promise supply relief. Read differently, however, they send a more cautionary signal. If a mature global manufacturer must expand connectivity capacity tenfold merely to meet the next phase of AI demand, existing output is clearly insufficient for what major customers expect to consume. The structure of the deals matters just as much as their size: through multiyear contracts, advance commitments and strategic investment, large technology companies are securing priority access to future output.

For smaller buyers, the central question may not be whether the industry is expanding. It may be whether much of the expansion has already been spoken for.

Fiber Optic Shortage Is Shifting From Demand Growth to Capacity Allocation

REBIO GROUP’s industry review, drawing on CRU data, shows that global optical-fiber shipments reached 662 million fiber-kilometers in 2025, up 15.3% year on year. China shipped about 372 million fiber-kilometers, or 56.3% of the global total.

Demand linked to AI data centers rose 75.9% in 2025 and could increase from roughly 5% of world demand in 2024 to around 30% by 2027.

Those numbers tell two stories.

The first is straightforward: total demand is growing.

The second is more important for procurement teams: the fastest-growing customers are also the best financed, the least price-sensitive and the most capable of signing multiyear supply agreements.

That changes factory economics. Manufacturers must now allocate silica preforms, drawing towers, coating equipment, testing capacity and experienced engineers among high-fiber-count AI cables, G.657.A2, ultra-low-loss fiber, polarization-maintaining fiber and other specialty products.

This Fiber Optic Shortage may therefore not appear as a simultaneous stockout across every specification. It is more likely to develop as a structural shortage:

• Higher-margin products receive production priority.

• Large customers reserve months or years of future allocation.

• Smaller buyers can still obtain quotations but struggle to secure firm delivery dates.

• Spot material remains available, yet its price moves far above the level implied by long-term contracts.

Polarization-Maintaining Fiber Will Not Replace G.657.A2 — but It May Squeeze Its Supply

Yicai reported in July 2026 that traditional FTTx demand could grow at a compound annual rate of about 2.47% through 2030, while fiber demand associated with AI computing centers could expand at roughly 20.74%.

More strikingly, demand for polarization-maintaining fiber may rise tenfold or even twentyfold within one to two years, driven by high-performance processors, silicon photonics and co-packaged optics.

Does that mean polarization-maintaining fiber will replace G.657.A2? Probably not.

The two products solve different engineering problems. Polarization-maintaining fiber preserves a stable polarization state and is commonly used in silicon-photonics components, CPO systems, lasers and precision optical equipment. G.657.A2 is valued for low bend loss, making it suitable for compact routing, high-density installation, access networks and specialized platforms.

But the absence of direct substitution does not mean there is no supply-side impact.

Preform capacity, drawing towers, coating lines, metrology equipment, specialist engineers and capital budgets are finite. Because polarization-maintaining fiber carries substantially higher unit value and margins, producers have strong incentives to direct new equipment, research teams and quality-control resources toward it.

A tenfold increase in polarization-maintaining-fiber demand may not remove one kilometer of G.657.A2 demand, but it could slow the pace at which new G.657.A2 capacity reaches the market.

G.657.A2 Is Being Pulled by Both Data Centers and Defense Demand

CRU’s March 2026 market update specifically identified rising demand for G.657.A2 from data centers and fiber-optic-controlled Drone applications.

Because G.657.A2 offers better margins than conventional G.652.D fiber, some producers have already redirected drawing capacity toward the bend-insensitive product.

That shift is significant.

For years, the market mainly associated G.657.A2 with FTTH and high-density indoor cabling. It is now entering another fast-growing market with a recurring-consumption profile.

Reuters has documented fiber-optic-controlled Drone systems deploying lines of roughly 20 kilometers in operational environments. Unlike telecom fiber, which can remain in service for years after installation, the fiber carried by a Drone is typically unspooled during a single mission and left across the terrain.

The result is continuing material consumption rather than merely a one-time infrastructure build.

This does not mean every such system uses the same G.657.A2 specification, nor does it mean defense demand has overtaken the communications market. It does mean that the demand model is changing. Buyers must now consider conventional network construction, AI data centers and recurring defense consumption at the same time.

For projects involving UGV, UAV, Drone and critical-infrastructure communications, stable access to industrial fiber and cable is becoming a project-continuity issue rather than a routine purchasing matter. The same supply-chain logic increasingly affects CBRN monitoring networks, solar and photovoltaic facilities, automated factories and IVD localization programs.

Price Signals Have Already Appeared — and They Are Spreading Globally

The market has not reached a universal stockout across all fiber categories, but the pricing signal is already difficult to ignore.

CRU’s separate pricing analysis found that Chinese G.652.D bare-fiber prices rose by more than 80% between November 2025 and January 2026.

Preform availability became the critical bottleneck, with effective capacity at many Chinese suppliers close to its limit. Dormant capacity cannot be restored in a matter of weeks; it requires capital, technical preparation, qualified staff and time.

By March 2026, the CRU global optical-fiber price index had climbed to 263, one of the sharpest monthly increases since the index began in 2017. Higher G.657.A2 scheduling, material constraints in India, Europe’s reliance on imports, and limited U.S. preform and drawing capacity were all contributing to the global move.

An 80% rise in G.652.D does not mean G.657.A2 must rise by exactly the same percentage.

It does, however, reveal tightening across the preform-and-drawing system that both products depend on. In that environment, a more technically demanding product with faster demand growth and stronger margins is unlikely to remain insulated from the broader price trend.

Lead times are sending the same message.

CRU’s OFC 2026 review reported that some high-fiber-count ribbon cables were carrying lead times of more than a year. Certain U.S.-compliant fiber capacity was largely sold out for 2026, with delivery periods exceeding 50 weeks.

Those products are not identical to standard G.657.A2, but they expose an increasingly clear order of priority: high-margin AI programs and large multiyear customers are receiving raw materials, equipment time and production slots first.

A 300,000-Kilometer Inquiry Shows the Risk Buyers Most Often Underestimate

REBIO GROUP’s published review recently described a representative inquiry.

A customer requested an urgent supply of 300,000 kilometers of G.657.A2 fiber. At the time, REBIO GROUP could coordinate approximately 30,000 kilometers of immediate material, but the customer considered the price too high and declined the available lot. This customer's momentary hesitation regarding the price allowed another customer to snatch up the entire batch of goods. By the time the first customer finally decided to accept the price he had initially refused, the limited production capacity had already slipped through his fingers.

The point is not to judge whether the customer made the right decision.

The episode reveals a familiar mismatch in a tightening market: buyers continue to evaluate quotations against the historical prices of an oversupplied cycle, while suppliers are already pricing confirmed preforms, production allocation and delivery windows.

In a normal market, rejecting a high price and waiting for another quotation can be a rational procurement strategy. In a capacity-constrained market, waiting can produce three different outcomes: a higher replacement price, a longer lead time, or the loss of the intended production slot.

Three Questions Buyers Should Confirm

• Does the quotation have confirmed preform and drawing capacity behind it?

• Can the delivery schedule be written into the contract rather than left as a verbal estimate?

• Can a framework order secure monthly allocation and staged deliveries for the coming six to twelve months?

A low quotation without production support may ultimately be less valuable than a higher quotation backed by a firm monthly delivery schedule.

REBIO GROUP’s role in this market is not limited to locating a one-time spot lot. It is to connect overseas buyers with Chinese manufacturing resources, quality-control processes and cross-border delivery capacity.

For customers with identifiable demand over the next six to twelve months, framework orders, monthly allocation and phased shipment may now matter more than continuing to search for the lowest immediate quotation.

Fiber Optic Shortage May Not Have Peaked — but Procurement Logic Has Already Changed

It would be premature to claim that G.657.A2 prices will rise by a fixed percentage indefinitely.

AI projects can be delayed. New capacity will gradually come online. Defense demand can shift with policy and operating conditions. Optical fiber remains a cyclical industry, and any argument that prices can only rise would be too simple.

But the evidence currently points in one direction.

Technology companies are signing multibillion-dollar supply agreements. A leading manufacturer is expanding U.S. connectivity capacity tenfold. Polarization-maintaining-fiber demand could rise tenfold to twentyfold within one or two years. G.657.A2 is being pulled by data-center and fiber-optic-controlled Drone demand. Preform utilization is high, while lead times for some optical products are already measured in nearly a year.

The most important risk in this Fiber Optic Shortage is not that the market will suddenly announce there is no fiber left. It is that high-quality capacity will be progressively reserved through long-term contracts and priority accounts before many buyers return to the market for a quotation.

For buyers with genuine G.657.A2 requirements, the decisive question may no longer be:

“Can the price be a little lower?”

It may be:

“When the project starts, will we still have a place in the production queue?”