Lumentum: Sold Out Through 2029
Disclaimer: This article is for informational and educational purposes only. It does not constitute financial, legal, or investment advice. Views are based on public information as of the publication date and are subject to change.
On October 9, Lumentum CEO Michael Hurlston told Bloomberg in Tokyo that the company’s optical component capacity is fully booked into early 2029, that roughly 70% of next year’s demand for some products cannot be met, and that customers have started locking in 2030 orders. In April, Hurlston said capacity would be committed through 2028 within two quarters. Six months later the date is early 2029.
Lumentum closed up 6% at a record 1,133 dollars. Coherent and AAOI followed. The stock is up more than tenfold in twelve months.
Readers who follow semiconductors already know the AI bottleneck is shifting from compute to interconnect.
This piece takes three narrower questions:
- Within the optical interconnect supply chain, which layer captures the money?
- Why do only two companies occupy that layer? And of the two, why is Lumentum the better one?
- After a tenfold run, what it is worth.
Where the money stops
Open an 800G or 1.6T transceiver and you find a DSP, drivers, photodetectors, packaging, and a laser. Every photon in the module comes from that laser.
The laser is built on an indium phosphide (InP) chip. That chip sets the module’s speed ceiling, power draw, and yield. There is no TSMC for InP. Every supplier runs its own fab. InP crystals grow slowly, 6-inch volume production is hard, yields are low, and new capacity takes two to three years from decision to shipment.
The bottom layer is InP laser chips. High-speed EML lasers at volume come from Lumentum, Broadcom, Mitsubishi Electric, and Sumitomo Electric; the top three hold about 72% of the market. In 980nm pump lasers, a narrower segment, Lumentum alone holds 70% to 80%.
The middle layer is module assembly. Innolight, Eoptolink, Coherent, AAOI, and Lumentum’s own Cloud Light all sit here. Innolight’s overseas capacity in Thailand and Taiwan alone runs 2.8 million modules a month. Revenue at these companies is growing fast, and none of them can buy enough lasers. Each of the top three US module makers flagged laser shortages on every earnings call from Q3 2025 through Q1 2026. AAOI described an industry-wide shortage of InP laser manufacturing capacity.
The top layer is contract manufacturing, such as Fabrinet. It gets the volume, not the price.
Module output is set by laser supply. Laser supply is set by InP fab capacity. The constraint sits at the bottom, and so does the excess profit. InP substrate demand in 2026 is roughly 2.6 to 3.0 million 2-inch-equivalent wafers against capacity of 0.6 to 0.75 million, a gap above 70%.
Two fabs, one shortage
Three conditions define the bottom layer:
- an in-house InP fab,
- in-house laser production,
- and a presence in both modules and optical switching.
Two companies meet all three. Lumentum and Coherent.
Broadcom makes EML and CW lasers but supplies its own CPO and ASIC ecosystem, not the open market. Mitsubishi and Sumitomo do not build modules or OCS and have no voice at the system level. Innolight and Eoptolink are large, and they are laser buyers. Coherent’s CEO put it this way: silicon photonics or EML, “both need InP capacity.”
The contest is not who ships the most modules. It is who owns InP wafers. On that axis, two public companies exist.
Coherent’s edge is wafer size. It built the world’s first volume 6-inch InP fab in Sherman, Texas. Six-inch delivers four times the capacity of 3-inch and cuts die cost by 60%. Coherent just received up to 50 million dollars under the CHIPS Act to expand, plans to more than double InP output by end-2027, and will bring a third 6-inch fab online in the first half of 2027. Nvidia invested 2 billion dollars in it.
Five reasons Lumentum, not Coherent
In order of weight.
First, Lumentum’s mix is purer exposure to the scarce layer.
Of Lumentum’s 3.01 billion dollars in fiscal 2026 revenue, the Components segment (laser chips, pumps, narrow-linewidth lasers) was 67%. Coherent is a 6-billion-dollar diversified company with industrial lasers, materials, and display businesses alongside optical, and module assembly is the bulk of its networking revenue. Buying Coherent buys a blend. Buying Lumentum buys a more concentrated claim on InP lasers. When the supply gap is widening, concentration is leverage.
Second, Lumentum is close to a monopoly in pump lasers and narrow-linewidth lasers.