Optical supply chain: lasers, packaging, test, and qualification as bottlenecks

A transceiver contains more than a laser

A laser creates light; a modulator or integrated laser structure places information onto it; coupling and packaging get it into fiber; a receiver converts the incoming signal back into an electrical representation. Drivers, signal processing, control and thermal behavior determine whether the complete module works in a customer's system. Coherent and Lumentum publicly present several layers of optical products. The AAOI annual report describes an issuer's own products and manufacturing business. These sources support different scopes, not a single interchangeable “optics” market.

1. 1Laser/device fabrication
2. 2package and coupling
3. 3module assembly
4. 4test and burn-in

1. 1Customer qualification
2. 2accepted shipment
3. 3cash collection
4. 4further capacity funding

Identify the production stage that limits good units

Material growth, wafer processing, die testing, packaging alignment, assembly and final module test can constrain different products. A laser that passes one laboratory test is not necessarily a qualified production part in a module. Power, wavelength, thermal behavior and lifetime must fit the intended design. An external laser source and a laser integrated into a transceiver create different replacement and reliability boundaries. Do not infer that a component product announcement proves a complete module shipment or market share.

Lumentum's accessed product presentation includes optical switching, laser, external-source and transceiver categories. Coherent's datacenter presentation also covers multiple optical layers. An undated page is not counted as a September release. It helps map a bill of materials, but a current product label cannot establish customer adoption, shipment volume or realized margins. Read the specific product's documentation before using a numerical reach or power limit.

Yield compounds across stages

In an original teaching example, start with 1,000 possible units. Suppose device yield is 80%, package yield is 90% and final test pass rate is 95%. Good units are 1,000×0.8×0.9×0.95, or 684. Treat these as hypothetical independent factors; real defect mechanisms can be correlated. Improving package yield to 95% raises good units to 722, a gain of about 5.56%, without increasing starts. Doubling starts is a different investment decision and may expose a later test bottleneck.

Now suppose test equipment can process only 700 units during the same period. The improved line can make 722 potential good units, but the test queue constrains output. More fabrication capacity would not remove that limit. Longer qualification can introduce another queue after final test. Capacity must therefore name product, yield, period, test capability and customer-acceptance boundary. Factory area is not an output metric.

The cost of a ramp reaches the balance sheet

Higher production can require equipment, spare parts, material inventory, engineering and receivables before collections. AAOI's filing-based financial lessons separately examine those quantities. A supplier with strong demand can still need external funding. Inventory growth is ambiguous: it may support a scheduled ramp, reflect long lead-time components or signal mismatched demand. Inspect composition and valuation rather than selecting the explanation that fits a favorable thesis.

A hypothetical selling price of 100 and unit cost of 70 leaves gross profit of 30. If price falls to 90 while cost stays 70, gross profit falls to 20 and gross margin becomes 22.22%, down from 30%. Increasing volume can offset some total profit loss while requiring more working capital. The example explains how growing optical demand and weaker supplier margins can coexist. It is not a forecast of any named company's price or cost.

A supplier-constraint reading lab

Build a row for each stage: device production, package, module, test, qualification and collection. Record the owner, the metric that measures throughput, the evidence date and the observation that would disprove scarcity. For example, improved final test output can disprove a test-capacity concern but cannot alone disprove customer delays. Compare the same product generation and fiscal period; do not rank companies using incompatible product labels.

For an offline design exercise, choose a link with a defined reach and host interface, then map its components to candidate vendor product families. Leave unverified suppliers blank. Record repair method and the failed part's replacement scope. Finally connect only disclosed company exposure to revenue, margin and cash. The purpose is to locate value capture and its dependencies, not to assert that all optical vendors benefit equally.