10.05

Who makes specialty glass & glass fiber for AI data centers

Specialty glass & glass fiber (optical fiber, T-glass, glass cores)

Purpose-engineered glass and glass fiber — optical fiber, low-loss T-glass fabric, and glass-core material — used in network cabling and inside advanced chip packages.

This part covers two related product lines. Optical fiber, drawn from specialty glass preforms, is the physical medium for the network layer's transceivers and cabling (Layer 3) and for site fiber entrance (Layer 9). Separately, low-dielectric-loss glass fabric (T-glass) and, in newer designs, solid glass cores are used inside the substrate that sits under large AI accelerator packages (Layer 1), where they carry high-speed signals between the die and the board with less loss than standard substrate materials. Both draw on specialty glass-melting and fiber-drawing capacity that is distinct from ordinary glass manufacturing and concentrated among a handful of producers. AI clusters are unusually fiber-intensive relative to their size — the scale-out and scale-up network fabric that ties thousands of GPUs together runs on far more fiber per rack than a general-purpose enterprise cluster — and the largest packages are the first to adopt glass-core substrates to keep signal loss down as package sizes grow. For investors, the fiber side is a function of optical-fiber and cable capacity and glass-preform lead times; the substrate side is a smaller, newer market where qualification with a handful of substrate and package makers is the gating step rather than raw material availability.

AI delta: AI clusters run far more optical fiber per rack than general-purpose compute because of the added scale-out and scale-up network fabric, and the largest accelerator packages are moving to glass-core and T-glass substrates to control signal loss as package size grows.

Appears at:Edge / microEnterprise / small coloLarge colo / hyperscale buildingAI campusGW-class campus· cooling:any — Optical-fiber demand scales with network fabric density at every tier and is highest in hyperscale, campus and gigawatt-class clusters; the glass-core substrate use applies only to the largest chip packages, independent of site size.

Companies on this part (8)

CompanyRoleOwnershipTickerPrice1dConfidenceNote
AGCmanufacturerpublic5201.T5447 JPY-0.4%disclosed · source
CorningmanufacturerpublicGLW124 USD+0.8%inferred
Jushimanufacturerpublic600176.SS32.90 CNY+1.0%inferred
Nippon Electric Glassmanufacturerpublic5214.T4859 JPY-2.9%disclosed · source
Nittobomanufacturerpublic3110.T2867 JPY-3.4%disclosed · source
Owens CorningmanufacturerpublicOC114 USD-1.9%inferred
Taishan · unit of CNBMmanufacturersubsidiary3323.HK*inferred
Schottmanufacturerprivateinferred

Who competes here

The same table read the other way: every company above competes for specialty glass & glass fiber sockets. Confidence tags matter — an inferred placement is a lead, not a confirmed rivalry.

Nearby parts

Latest feed

Questions about specialty glass & glass fiber

What is specialty glass & glass fiber?
Purpose-engineered glass and glass fiber — optical fiber, low-loss T-glass fabric, and glass-core material — used in network cabling and inside advanced chip packages.
Which companies make specialty glass & glass fiber for AI data centers?
This catalog lists 8 companies on Specialty glass & glass fiber, 7 of them with listed equity: AGC, Corning, Jushi, Nippon Electric Glass, Nittobo, Owens Corning, Taishan, Schott.
How does AI change demand for specialty glass & glass fiber?
AI clusters run far more optical fiber per rack than general-purpose compute because of the added scale-out and scale-up network fabric, and the largest accelerator packages are moving to glass-core and T-glass substrates to control signal loss as package size grows.
How reliable are these specialty glass & glass fiber supplier placements?
The 8 placements on this part break down as 3 disclosed, 5 inferred. "disclosed" means the company or a filing says so; "reported" comes from trade press or a reference-design list; "inferred" is a taxonomy lead with no citation yet.

Sources: placements without a source URL come from the taxonomy seed and are tagged accordingly. How this data is built.· Markdown · JSON