Snapshot

Optical fiber is hair-thin glass that carries data as pulses of light. Cable manufacturers draw ultra-pure glass preforms (cylindrical blanks, each yielding thousands of kilometers of fiber) into strands, bundle them into cables with protective jackets, and sell to telcos, hyperscalers, and enterprises. The unit of demand is fiber-km — one kilometer of a single fiber strand. A cable containing 1,728 fibers and running 10 km represents 17,280 fiber-km of demand. Global production runs roughly 550–650 million fiber-km per year est., dominated by Corning (US), Prysmian (Italy), YOFC (China), Sumitomo Electric (Japan), and Furukawa Electric (Japan). These five hold over 40% of the market by revenue (GM Insights, 2024). The total fiber optic cable market was $13B in 2024, projected to reach $34.5B by 2034 at a 10.4% CAGR (GM Insights).

Ticker mismatch warning. The four tickers assigned to this group — COHR, WIRE, CLS, OFS — are a poor fit for "optical fiber & cable" as a product. COHR (Coherent) makes optical transceivers and laser components, not fiber cable. WIRE (Encore Wire) was a copper wire maker acquired by Prysmian in July 2024 and is no longer publicly traded. CLS (Celestica) is a contract electronics manufacturer of servers and switches. OFS (OFS Capital) is a $46M business development company (lending fund) with no connection to fiber optics — the fiber maker "OFS Fitel" is a private subsidiary of Furukawa Electric. The dominant publicly traded optical fiber & cable company is Corning (GLW), which is not in this ticker list. Every section below flags where the tickers diverge from the actual product.

The product & how money is made

The physical product is optical fiber cable: glass strands bundled into protective sheaths. Manufacturing has two stages. First, a preform is made — a cylinder of ultra-pure silica, roughly 1 meter long and 20 cm in diameter, that encodes the refractive-index profile (the "recipe" for how light travels through the glass). One preform yields roughly 8,000 km of fiber est.. Second, the preform is heated and drawn into fiber at a draw tower — a 30-meter-tall furnace pulling glass at 20+ meters per second. Fibers are then coated, bundled (loose-tube or ribbon), and jacketed into cable. Cable designs range from 2-fiber drop cables (for a single home) to ultra-high-fiber-count (UHFC) ribbon cables holding 1,728–6,912 fibers in a single sheath (AFL Hyperscale).

How the money flows

• Cash in: fiber cable sold by the kilometer at a price per fiber-km. Prices vary by type — commodity G.652.D single-mode for telco runs in the low single-digit dollars per fiber-km; specialty bend-insensitive G.657.A2 fiber (increasingly demanded for data centers and drone-deployed cable) commands a premium. In China, G.652.D bare fiber surpassed 40 RMB (~$5.50) per fiber-km in January 2026, up 50%+ year-over-year (KAD, Jan 2026). Connectivity products (pre-terminated assemblies, cassettes, patch panels) carry higher margins than bare cable.
• Cash out: raw materials (silica, germanium, coating chemicals), energy-intensive draw-tower operation, capital for preform and draw-tower capacity. A draw tower costs $50–100M est. and takes 12–18 months to build. Preform manufacturing is even more capital-intensive and has fewer vendors. Labor, protective jacketing materials, and distribution logistics are additional costs.
• Margin structure: Corning's Optical Communications segment (fiber + cable + connectivity) earned $1.048B net income on $6.274B revenue in FY2025 — a 16.7% net margin (Corning FY2025 10-K). This is a capital-intensive, scale-driven business: the largest producers have lower per-fiber-km costs, which creates a barrier to entry.

Demand

Contracted demand

• Meta–Corning: multiyear agreement for up to $6 billion in optical fiber, cable, and connectivity solutions through 2030. Announced January 2026. Meta is the anchor customer for Corning's Hickory, NC manufacturing expansion. Meta has 26 data centers planned or under construction, including a 5 GW campus in Louisiana (Datacenter Knowledge, Jan 2026). contracted
• NVIDIA–Corning: $500M pre-funded equity purchase plus warrants worth up to $2.7B (total potential value ~$3.2B). Multi-year supply + technology partnership for passive optical interconnects inside AI "factories" — rack-level and cluster-scale. Corning expanding optical production by 10x and fiber capacity by 50%. Three new manufacturing plants in North Carolina and Texas, creating 3,000 jobs. Announced May 2026 (Corning/NVIDIA press release). contracted
• Two undisclosed hyperscalers: Corning CEO stated on Q1 2026 earnings call that two additional hyperscaler agreements were finalized, "similar in size and duration" to the Meta $6B deal — implying ~$6B each est.. Customer names not disclosed (finsee.ai, Corning Q1 2026 review). contracted
• Lumen Technologies–Corning: supply agreement for next-generation fiber cable, announced August 2024. Dollar value not disclosed (GM Insights). contracted

Forecast demand

• AI data centers (the new driver): a 30,000-GPU AI building requires roughly 12,240 core-km of fiber for its internal network alone (KAD, 2026). A single AI GPU rack (e.g., NVIDIA NVL72) needs up to 1,152 fibers — vs 15–30 fibers in a traditional cloud rack, a 38–77x increase per rack (AFL Hyperscale). A large AI campus can consume over 1 million fibers just for GPU-to-switch interconnects. CRU's 2025 industry report identified AI/hyperscale data centers as the "strongest growth engine" for optical fiber, overtaking traditional telecom.
• FTTH / broadband (the secular floor): the US BEAD (Broadband Equity, Access, and Deployment) program allocates $42.45B for broadband buildout; first demand impact expected 2026–2027, though permitting and BABA (Build America, Buy America) compliance may delay execution (CRU, 2025). FTTH in India growing at 13.8% CAGR; China at 12.8% (GM Insights).
• 5G backhaul & small cells: every 5G base station needs fiber backhaul. The US has ~1 million small cells deployed vs 5+ million projected need for full 5G densification est.. 5G deployment growth is moderating post-initial buildout.
• Undersea cables: new submarine cable systems (each containing 12–24 fiber pairs spanning thousands of km) are being ordered by Google, Meta, and Microsoft for inter-continental AI traffic. Backlog is 3–4 years (500-stocks networking scan). This adds specialty fiber demand (ultra-low-loss, long-haul).

Sources: Corning press releases (Jan, Mar, May 2026); CRU "Top ten industry shifts" (2025); GM Insights fiber optic cable market report (2024); AFL Hyperscale UHFC white paper; KAD "AI Data Centers Driving Fiber Demand" (2026); 500-stocks networking scan.

Supply

Current capacity

Global fiber production is estimated at 550–650 million fiber-km per year est.. The five largest producers — Corning, Prysmian, YOFC, Sumitomo, Furukawa — together control over 40% of the market by revenue (GM Insights). Corning alone claims more than 1 billion km of fiber produced cumulatively and operates two of the world's largest fiber/cable factories in Hickory, NC (Corning corporate). Chinese producers (YOFC, Hengtong, ZTT) dominate volume but face US tariffs, EU anti-dumping duties, and BABA compliance barriers for government-funded US projects (CRU, 2025).

Capacity expansion in progress

• Corning: building three new manufacturing plants (NC + Texas), funded partly by NVIDIA's $500M investment. Expanding "optical production by 10x and fiber capacity by 50%" (Corning/NVIDIA, May 2026). Separately, broke ground March 2026 on a cable manufacturing expansion at Hickory, NC (Trivium Corporate Center) with $170–268M investment and 750–1,000 new jobs. Plans to increase NC employment by 15–20% (Lightwave Online, Mar 2026).
• Prysmian: acquired Encore Wire (WIRE ticker, $290/share cash) in July 2024, adding US manufacturing capacity for copper cable and expanding its North American footprint. Expanding Encore's McKinney, TX plant (Recycling Today). Prysmian is the world's largest cable company by revenue (fiber + power cable).
• Sterlite Technologies (STL): began US fiber cable production in 2024 for BEAD-funded projects (GM Insights). Indian supplier entering the US market.
• Sumitomo Electric: acquired major shares of Germany's Südkabel in June 2024 to expand European production (GM Insights).

Bottlenecks

• Preform manufacturing: the most capital-intensive and technically complex step. Only a handful of companies globally can make high-quality preforms. Expanding preform capacity takes 18–24 months est.. This gates total fiber production more than draw-tower capacity.
• Specialty fiber (G.657.A2): CRU's 2025 report noted a sharp demand rise for G.657.A2 (bend-insensitive) fiber for data centers and drone deployment. This created tightness in the supply of standard G.652.D fiber in China as manufacturers shifted production (CRU).
• Lead times: ribbon fiber cable lead times exceeded 60 weeks as of March 2026. Loose-tube outdoor fiber was backordered into Q3 2026 (Lightwave Online).
• Trade barriers: US tariffs on Chinese fiber, EU anti-dumping duties on Indian and Chinese imports, and BABA requirements for federally funded US projects all constrain the effective supply available in Western markets (CRU).

Sources: Corning press releases; CRU 2025; GM Insights; Lightwave Online (Mar 2026); Prysmian/Encore Wire press release (Jul 2024).

The gap

Observed signals point to demand exceeding supply in the near term, particularly for specialty fiber and high-count data-center cable:

When could it flip: The large capacity expansions underway (Corning's 10x optical production ramp, Prysmian's North American additions, Chinese exporters pushing into new markets) could overshoot if either AI capex or BEAD spending stalls simultaneously. The concentrated buyer base (a few hyperscalers account for most contracted demand) means a spending pause could create a glut quickly, as happened in the 2001 telecom bust. The European fiber market is already contracting with tightening margins (CRU, 2025). This is a forecast — timing is unknown.

Sources: Lightwave Online; KAD; CRU 2025; AFL Hyperscale; Corning press releases.

The players

Important: the tickers assigned to this group (COHR, WIRE, CLS, OFS) mostly do not make optical fiber or cable. The table below shows what each ticker actually does, alongside the dominant fiber & cable companies that are not in the ticker list.

Sources: Corning FY2025 10-K; Coherent Q3 FY2026 earnings; Celestica Q1 2026 earnings; Prysmian/Encore Wire press release (Jul 2024); OFS Capital (stockanalysis.com); GM Insights (market share). Market caps as of Jun 2, 2026.

The price of exposure

Because the assigned tickers are a poor match for this product, the arithmetic below separates the dominant fiber play (Corning/GLW) from the tickers actually in the group.

Corning (GLW) — the closest pure play on optical fiber & cable

• Revenue vs market value: Corning's FY2025 total core revenue was $16.4B; market cap is $172B. That is roughly $10.6 of market value per $1 of current annual revenue (P/S 10.6). Optical Communications alone was $6.3B — the remaining $10.1B of revenue comes from display glass, automotive, specialty materials, and other segments. So only ~38% of the revenue base is fiber/cable/connectivity.
• Earnings vs market value: trailing P/E of 96x; forward P/E of 59x. GAAP net income in FY2025 was $1.6B. At $172B market cap, the stock needs sustained rapid earnings growth to justify the price arithmetically.
• Cash generation: FY2025 adjusted free cash flow was $1.72B (FCF yield ~1%). Operating cash flow was $2.7B. Corning carries $9.9B in total debt vs $1.8B cash (net debt ~$8.1B). Debt-to-equity ratio is 0.80; interest coverage 7.3x.
• Growth trajectory: Corning's "Springboard" plan targets $11B in incremental annualized sales by end of 2028 (vs Q4 2023 run rate) and $40B total sales by 2030. Optical Communications grew 35% in FY2025 and 36% in Q1 2026. The $18B+ in contracted hyperscaler deals provides unusual forward visibility.

Coherent (COHR) — adjacent exposure via transceivers

• What you own: transceivers and laser components that plug into fiber — the active-optics layer, not the passive fiber/cable layer. Revenue $5.8B FY2025, market cap $84B (P/S ~14.4x). Datacom/comms is ~75% of revenue and growing 40%+ YoY.
• Money-in/out: consolidated net income turned positive in FY2025 at $30M (marginal). Quarterly run rate improving: Q3 FY2026 GAAP EPS $0.97. Carries $3.2B long-term debt (down from $4.2B at merger close). Cash $1.6B. Deleveraging.

Celestica (CLS) — indirect exposure via hardware

• What you own: a contract manufacturer building AI servers and network switches. Not a fiber company. Q1 2026 revenue $4.05B, 80% from CCS (compute and connectivity solutions). Forward P/E ~40x. Market cap $54B.

WIRE and OFS — no viable exposure

• WIRE: delisted July 2024. Cannot be bought.
• OFS: a $46M BDC. Buying it exposes you to a portfolio of middle-market loans, not fiber optics.

Sources: Corning FY2025 10-K, stockanalysis.com; Coherent Q3 FY2026 earnings; Celestica Q1 2026 earnings. All prices and market caps as of Jun 2, 2026.

What to deep-dive next

• Corning (GLW) is the obvious first deep-dive for this product — it is the largest optical fiber maker globally, the only US-listed company with dominant share, and has $18B+ in contracted hyperscaler revenue providing visibility through the end of the decade. Key questions: how much of the $172B market cap rests on the Springboard plan's $40B-by-2030 revenue target actually materializing, and what happens if hyperscaler spending pauses.
• Prysmian (PRY.MI) is the second: world's largest cable company, now including Encore Wire's North American copper/building-wire business. Not US-listed (Milan Stock Exchange). Key question: does the combined fiber + power cable + copper wire portfolio offer a more diversified entry into the "wiring the AI build-out" theme.
• Preform supply is the real bottleneck. Understanding who makes preforms (Corning, Sumitomo, Shin-Etsu, a few Chinese firms) and their capacity expansion timelines would tell you more about when the supply gap closes than anything else in this sub-sector.
• COHR and CLS are legitimate companies worth analyzing — just not as optical fiber plays. They belong in the "Optical Transceivers" and "Contract Electronics Manufacturing" groups respectively.

Sources & confidence

• Hard / filed: Corning FY2025 10-K and FY2024 10-K (segment revenue, net income, FCF, debt); Corning Q1 2026 earnings release (Optical Comms $1.85B, +36%); Coherent Q3 FY2026 earnings press release (revenue $1.81B, segment mix); Celestica Q1 2026 earnings (revenue $4.05B, CCS $3.24B); Prysmian/Encore Wire acquisition press release (Jul 2, 2024, $290/share); OFS Capital (stockanalysis.com — revenue $39M, net loss $37M, mkt cap $46M).
• Contracted / announced: Corning–Meta $6B deal (Corning press release, Jan 2026); NVIDIA–Corning $500M equity + $2.7B warrants + three new plants (Corning/NVIDIA, May 2026); Corning NC expansion $170–268M (Lightwave Online, Mar 2026); two additional unnamed hyperscaler deals "similar in size and duration" to Meta (Corning Q1 2026 earnings commentary via finsee.ai).
• Industry reports (hard but third-party, not SEC filings): GM Insights fiber optic cable market ($13B 2024, $34.5B by 2034, 10.4% CAGR); CRU "Top ten industry shifts in 2025" (AI as growth engine, G.657.A2 tightness, pricing uptick, European contraction); AFL Hyperscale UHFC white paper (1,152 fibers per AI rack, 38–77x vs traditional); KAD "AI Data Centers Driving Fiber Demand" (12,240 core-km per 30k-GPU building, China fiber pricing 40 RMB +50% YoY).
• Approximate / not live-verified (every est. tag): global production volume 550–650M fiber-km; Corning >40% global share; draw tower cost $50–100M; preform yield ~8,000 km; 5M small-cell target; preform expansion lead time 18–24 months; Prysmian market cap and revenue.