The Hidden Glass Highways: How AI Is Reshaping the Fibre Optics Industry

When most people think about the AI boom, they think about chips, software models, or flashy consumer applications. But one of the most important parts of this entire build-out is far more physical.

Beneath the ground, across oceans, inside data centers, and between buildings, the world runs on optical fibre. These ultra-thin strands of glass carry data as pulses of light. For years, fibre was seen as basic digital plumbing — essential, but not particularly exciting.

That may now be changing, and changing fast.

My key takeaway after studying this space is simple:

AI is not just increasing demand for computing. It is changing the economics, product mix, bottlenecks, and profit pools of the optical fibre industry.

This is no longer just a telecom cable story.

It is increasingly becoming a data center, AI infrastructure, and supply-chain control story.

  1. The industry has moved from slump to scarcity very quickly

What makes this cycle interesting is that optical fibre was not in a boom until very recently.

Through late 2024 and much of 2025, the industry was actually weak. Telecom operators had built excess inventory during earlier supply-chain disruptions and then slowed fresh purchases while consuming that stock. As a result, factories were underutilized, orders were soft, and sentiment remained dull.

Then came 2026.

The market flipped from oversupply to shortage in a short span of time.

The reason appears to be clear: large-scale AI infrastructure builders have entered an arms race. They are no longer spending only to expand cloud capacity. They are now building what can effectively be called AI factories, and the speed at which this physical backbone gets built may determine who gains the edge.

That one shift has materially changed the demand equation for fibre.

  1. AI data centers consume far more fibre than traditional data centers

This is the heart of the thesis.

Traditional data centers mostly handled traffic going in and out. A user sends a request, the server responds, and the workload is relatively straightforward. The machines inside do not need to communicate with each other at extreme intensity all the time.

AI changes that completely.

In AI workloads, thousands of chips need to communicate continuously. Instead of simple traffic flowing only inward and outward, there is now enormous traffic moving within the data center — between racks, between chips, and increasingly between multiple buildings within the same campus.

That is why this part of the market matters so much.

An AI-focused data center can require many times more optical fibre than a traditional data center, with fibre density also materially higher than older norms.

That single shift explains why this is becoming a powerful segment-level tailwind.

This is not normal internet growth.

This is a very different level of infrastructure intensity.

  1. The tailwind is moving from generic telecom fibre to high-density AI architecture

This is where the opportunity starts becoming more specific.

Earlier, standard fibre cables with relatively lower fibre counts were sufficient for many telecom applications.

That is no longer enough in modern AI environments.

Demand is now moving toward very high fibre-count cables, because operators want to pack more fibres into the same ducts and conduits instead of spending heavily on digging new routes.

So the tailwind is not just about more fibre.

It is increasingly about denser, more complex, higher-value fibre products.

That distinction matters a lot.

Because when a cycle moves from simple volume growth to product-mix upgrade, value creation does not get distributed evenly. Some parts of the value chain benefit much more than others.

  1. This is not just a cable story — it is becoming an architecture story

Another important shift is that fibre is no longer being viewed merely as cable sold by the kilometre.

The market is moving toward solutions that help customers deploy networks faster, fit more connections into limited space, and reduce installation complexity.

In other words, the profit pool may gradually move away from plain commodity supply toward:

- high-density designs,

- faster deployment solutions,

- pre-connected systems,

- and products that reduce time, labor, and space usage.

That changes the industry structure.

The winners may not be those who simply sell the most fibre.

They may be those who solve the biggest bottlenecks in real-world deployment.

  1. The real bottleneck is not cable capacity — it is preform capacity

This may be the single most important supply-side insight in the entire sector.

Many people assume fibre can be made simply by heating glass and drawing it into thin strands. In reality, the most critical input is the glass preform.

A preform is the engineered glass rod from which fibre is drawn. It is highly complex to produce, requires extreme precision, and forms the economic core of the value chain.

This matters because preform contributes a very large share of the total cost of optical fibre.

So if preform supply becomes tight, the entire industry gets constrained.

And unlike simpler capacity additions, preform plants cannot be built quickly. These are specialized facilities that often take long periods to build, stabilize, and scale properly.

This is what makes the current setup powerful.

It is not just a story of rising demand.

It is a story of rising demand colliding with a hard physical supply bottleneck.

  1. A critical raw material has become a geopolitical choke point

Preform is not the only issue.

Another hidden bottleneck in the value chain is a critical mineral used to modify the glass so that light can travel correctly through the fibre. Without steady access to this input, producing high-quality telecom-grade fibre becomes much more difficult.

And here the issue becomes geopolitical.

Global supply of this material is concentrated, and export controls have already shown how fragile the value chain can become when too much dependence sits in one geography.

The result is straightforward:

- raw material prices can surge,

- supply security becomes strategic,

- and manufacturers without sourcing control become far more exposed.

So this segment is now being shaped not only by technology and capex, but also by critical mineral geopolitics.

That changes how investors should think about the sector.

This is no longer just an industrial product cycle.

It is increasingly a strategic supply-chain cycle.

  1. Pricing power has started shifting toward suppliers

When strong demand collides with preform shortages, raw material stress, and long expansion timelines, prices move.

That appears to be what this cycle is now showing.

Prices of basic fibre in some markets have risen sharply from late-2025 levels. Finished cable realizations have also improved.

This is important because it signals a clear market structure shift:

the industry has moved from a buyer’s market to a seller’s market.

That means the tailwind is not only in volumes, but also potentially in:

- better realizations,

- better product mix,

- and stronger bargaining power for capable suppliers.

Of course, this benefit will not be uniform across all players.

Positioning matters a lot.

  1. Large customers are beginning to treat fibre as strategic infrastructure

One of the clearest signs of the new market structure is that major customers are no longer comfortable depending entirely on spot market availability.

Instead, they are increasingly willing to secure long-term supply arrangements and lock in manufacturing capacity well in advance.

That is a very important signal.

It suggests that optical fibre is no longer being treated as just another low-attention input. It is being treated as strategic infrastructure.

And that changes customer behaviour in a big way:

- capacity may get pre-booked,

- supplier expansion may get supported,

- and reliability may start mattering more than just the lowest price.

That can reshape industry economics over time.

  1. So where exactly are the tailwinds moving?

This, in my view, is the most important investor takeaway.

The tailwind is not moving evenly across the entire fibre industry.

It is moving in layers:

First, from traditional telecom stocking cycles toward AI- and hyperscaler-led infrastructure buildout.

Second, from standard fibre-count cables toward high-fibre-count, high-density architectures.

Third, from plain cable selling toward more value-added, deployment-friendly connectivity solutions.

Fourth, from simple manufacturing scale toward control over preform and raw material sourcing.

Fifth, from commodity volume stories toward strategic supply-chain stories.

Sixth, from domestic telecom dependence toward global data center and export opportunities.

In short:

The real winners may not be the companies that merely sell more kilometres of cable.

They may be the ones that control glass, solve density, reduce installation time, and survive supply-chain shocks.

Final view

Optical fibre is quietly becoming one of the most important physical enablers of the AI era.

As AI data centers scale up, the world needs more glass, denser glass, better-connected glass, and more secure access to the materials required to make that glass.

That is why this segment deserves to be tracked not as a boring telecom utility, but as a strategic infrastructure layer.

For investors, this space may need to be studied through four key lenses:

1. Demand intensity

How much AI and data center capex is pulling through fibre demand.

2. Product mix

Whether the company is exposed to commodity cable or higher-density, higher-value solutions.

3. Backward integration

Whether it controls preform or remains dependent on outside suppliers.

4. Geopolitical positioning

Whether it can benefit from supply-chain de-risking and avoid raw-material vulnerability.

The market may still talk about GPUs first.

But underneath that narrative, a quieter race is underway —

a race to build the glass highways of AI.

And that race may end up creating a very different set of winners than the market currently assumes.

4 Likes

Not sure if there are Indian players here, but globally all and sundry related to Fiber Optics seems relatively discovered and has been run up pretty hard. Personally feel sector is “already priced” in at the moment.

Thanks for your write-up. Its well written, concise and identifies all the key drivers for someone new to the Optical Fibre industry.

Would be also quite helpful for the community if you may please use your sector expertise in evaluating relevant Indian co’s in various layers of the value chain (e.g. Sterlite Tech, HFCL etc.) using the 4 lenses you have suggested in article. Thanks.

2 Likes

In India, Sterlite Technologies looks like one of the biggest potential beneficiaries of the emerging AI-led optical infrastructure cycle.

The first reason is capability. Sterlite has steadily moved into the elite bucket of optical fiber manufacturers that are not just making standard OFC, but are also building capabilities in advanced products such as high-density IBR cables and hollow-core fiber. These are the kinds of products that become increasingly relevant as AI infrastructure, hyperscale data centers, and low-latency networks demand far higher fiber density, better performance, and faster deployment. Importantly, only a limited number of companies globally have meaningful capabilities in these areas.

The second reason is integration. As demand for advanced optical products rises, upstream raw materials such as preforms are also becoming more critical. Preform availability has tightened, and pricing has risen as capacity constraints have started showing up across the ecosystem. This creates a strong advantage for players that are backward integrated. Sterlite belongs to that smaller global set of companies that have this integration, which gives it a double edge: it is positioned not only to participate in the value-added product opportunity, but also to protect itself better from raw material tightness.

This becomes even more important when one looks at the broader industry context. Tightness in preforms, including supply-side stress in China, already pushing up bare optical fiber prices and, in turn, raising the cost base for higher-value cable products. In such an environment, companies that depend on external preform supply remain more exposed.

That is where the contrast with HFCL becomes interesting. HFCL has clearly built strong capabilities in high-fiber-count cable products and is pushing aggressively into the data center opportunity. However, it still lacks the same level of backward integration today. While HFCL has announced capex for preform manufacturing, that capacity is still a few years away from becoming operational and contributing meaningfully.

So, at least at this stage of the cycle, Sterlite appears to be better placed not just because of what it can sell, but also because of what it can secure internally.

That said, the market is usually quick to discount visible advantages. So yes, there is a high probability that a good part of this narrative is already getting priced into the stock.

3 Likes