Highlights
- The optical transceiver market is projected to surge from $7 billion in 2024 to $24 billion by 2030, driven by AI data center expansion.
- Germanium prices hit nearly $5,000/kg in September 2025 while rare earth dopants like erbium and ytterbium remain opaque, privately negotiated markets dominated by Chinese pricing agencies.
- China's leverage lies not just in mining but in refining and purifying specialty optical inputs like erbium, ytterbium, and thulium used in active fiber amplifiers and lasers.
- A Chinese export restriction on specialty dopants and optical chemicals could disrupt qualification and delivery schedules for hyperscale data centers, subsea cables, and medical photonics.
- Preform manufacturing capacity takes 18 to 24 months to build, meaning supply cannot respond quickly to geopolitical shocks at the germanium and rare earth chemistry layer.
Fiber optics is no longer just a telecom story; it is now an AI infrastructure story. The clearest hard number on the active-optics side is the optical transceiver market, which LightCounting estimated at $7 billion in 2024 (opens in a new tab)and projected to reach $24 billion by 2030 as AI data centers scale. On the passive-fiber side, demand is also surging: CRU-cited industry data show data-center fiber demand grew about 76% in 2025 and could reach 30% of global fiber demand by 2027, up from less than 5% in 2024. Meta’s multiyear agreement to buy up to $6 billion of Corning optical fiber, cable, (opens in a new tab) and connectivity products, plus Nvidia’s $500 million investment to expand Corning’s U.S. fiber capacity, (opens in a new tab) show how quickly hyperscalers are locking up supply.
China remains the deepest manufacturing base, with Hengtong (opens in a new tab) and FiberHome (opens in a new tab) reportedly running at full capacity and YOFC (opens in a new tab) now demonstrating advanced hollow-core fiber. The United States remains anchored by Corning (opens in a new tab) and OFS (opens in a new tab) (Furukawa Electric); Japan by Sumitomo Electric (opens in a new tab) and Furukawa; Europe by Prysmian (opens in a new tab); and India by STL (opens in a new tab). The practical takeaway is that the market is global, but scale, speed, and marginal capacity are still heavily influenced by Chinese producers.
The Supply Chain Overview
The fiber-optics supply chain begins with ultra-pure silica and specialty chemical inputs, moves into preform manufacturing, then fiber draw, polymer coating, cabling, connectors, and finally system integration into data centers, telecom backbones, subsea cables, factories, and medical devices. The real bottleneck is not cable assembly; it is preform capacity, which industry sources say typically takes 18 to 24 months to build. That matters because data centers can consume radically more fiber than conventional server architectures, and supply cannot be turned on like a software switch.
There are really two fiber economies. The first is standard passive transmission fiber, which mainly depends on high-purity silica plus germanium chemistry to tune refractive index in the core. The second is the specialty “active fiber” economy, where rare earth ions are deliberately doped into the glass so the fiber can amplify or generate light. In telecom, the dominant example is the erbium-doped fiber amplifier, which works in the 1550 nm low-loss window of silica fiber. In industrial and medical photonics, ytterbium is central to high-power fiber lasers, thulium to mid-infrared and surgical/sensing applications, and neodymium to specialty near-infrared laser architectures. MCVD and related preform techniques are used to incorporate these dopants into the glass core.
The Hidden Mineral Dependency
The critical-minerals story is sharper than many data-center investors appreciate. Germanium is typically recovered as a byproduct of zinc processing and then refined into germanium dioxide or germanium tetrachloride, both important for optical applications and especially for fiber-preform chemistry. Rare earth dopants such as erbium, ytterbium, neodymium, and thulium are mined from rare-earth ores or ion-adsorption clays, chemically cracked and leached, then separated through solvent extraction into high-purity oxides suitable for optical-grade downstream processing. China’s leverage is not simply in mining; it is in the refining, purification, and pricing ecosystem that converts obscure minerals into specification-grade optical inputs.
The best visible price benchmarks underscore the risk. NdPr oxide—a broader rare-earth benchmark relevant to Western pricing power—fell to about $63/kg in early 2025 and then rallied to about $123/kg in February 2026, above the $110/kg floor price used in recent U.S. defense-backed supply agreements. Germanium became even more acute: reported prices near $5,000/kg in September 2025, and prices were up 10% more amid renewed defense demand and scarcity. By contrast, erbium, ytterbium, and thulium markets are smaller and more opaque; prices are often negotiated privately, and Western benchmarking still leans heavily on Chinese agencies such as Asian Metal and Shanghai Metal Market. Specialist intermediaries and traders such as Tradium and Strategic Metal Investments remain important precisely because these are not deep, transparent commodity markets.
Unlike copper, gold, or oil, rare earths and many critical minerals do not trade in deep, transparent global commodity markets. Most transactions occur through confidential, negotiated contracts where price depends on purity, specification, volume, delivery terms, geographic origin, qualification status, and long-term strategic relationships. Consequently, neither China's published domestic price indices—which operate within a state-shaped market influenced by production quotas, export licensing, and industrial policy—nor Western price reporting agencies should be viewed as definitive market prices.
Instead, they serve as useful reference points or directional indicators. In practice, every commercial transaction is unique, particularly for high-purity separated oxides, specialty metals, alloys, and magnet feedstocks. The true market remains highly fragmented, largely over-the-counter, and often opaque, making price discovery itself a strategic advantage. This lack of transparent pricing is one reason Rare Earth Exchanges continues to advocate for more robust Western market infrastructure and independent price discovery mechanisms, and above all, more transparency.
Chinese Gates
If China sharply tightened access again, the first shock would hit specialty dopants and optical chemicals, not just “fiber cable” in the abstract. Standard telecom fiber could keep moving for a time where non-Chinese preform and fiber plants already exist, but lead times would lengthen, costs would rise, and premium products for AI data centers, subsea systems, fiber lasers, sensing, and medical devices would tighten first. Erbium-Doped Fiber Amplifiers (EDFAs), specialty laser fibers, and some preform chemistries are exactly the kind of small-volume, high-consequence inputs that can snarl an entire downstream sector. In other words, a China cutoff would not merely raise glass costs; it would threaten throughput, qualification, and delivery schedules across hyperscale data centers and adjacent high-tech manufacturing.
For Rare Earth Exchanges, the warning is straightforward: fiber optics looks like a diversified industrial sector, but the mineral-intensity of the preform and active-fiber layers gives Beijing meaningful leverage over one of the West’s most strategic growth markets. This is the same analytical error REEx has identified elsewhere in critical minerals: policymakers and boards often confuse a signed contract or plant expansion with true supply-chain sovereignty. In fiber optics, that mistake could leave the West believing it secured AI infrastructure while remaining exposed at the germanium and rare-earth chemistry layer.
Register today: REEx Marketplace™ (opens in a new tab)
0 Comments
No replies yet
Loading new replies...
Moderator
Join the full discussion at the Rare Earth Exchanges Forum →