AI Data Center PartsSearch
Home / Parts / Networking & interconnect / Optics contract manufacturing
3.05

Who makes optics contract manufacturing for AI data centers

Contract manufacturers that assemble and test optical transceivers on behalf of module brands and switch vendors.

Optics contract manufacturing sits between the component layers that feed a transceiver (lasers in 3.06, DSPs and drivers in 3.07, silicon photonics in 3.08) and the finished, branded modules described in 3.04. Rather than every module brand running its own assembly lines, a handful of contract manufacturers build modules — including a GPU vendor's own branded optics — to the module brand's design, similar to how EMS providers build servers to an OEM's specification elsewhere in the stack. AI-driven optics demand has concentrated volume in this small group of manufacturers, since qualifying a new assembly and test line for the highest optical speeds is a slow, capital-intensive process. An investor should watch how tightly a given contract manufacturer is tied to a single large customer's volume, since that concentration cuts both ways: it can mean a durable, high-utilization relationship or a single-customer dependency.

AI delta: Surging optical module demand from AI clusters has concentrated assembly volume in the small group of contract manufacturers already qualified at the highest optical speeds.

Appears at:Large colo / hyperscale buildingAI campusGW-class campus· cooling:any — Volume follows transceiver demand (3.04); relevant wherever high-speed optics are being built at scale, immaterial below hyperscale-level deployments.

Companies on this part (5)

CompanyRoleOwnershipTickerPrice1dConfidenceNote
CelesticamanufacturerpublicCLS317 USD+0.5%inferred
FabrinetmanufacturerpublicFN433 USD-4.3%inferredbuilds Nvidia's modules
FlexmanufacturerpublicFLEX115 USD+1.1%inferred
JabilmanufacturerpublicJBL312 USD-0.3%inferred
SanminamanufacturerpublicSANM203 USD+1.2%inferred

Who competes here

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

Nearby parts

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