Coherent: After AI Optical Interconnect Demand Materializes, Gross Margin Is the Real Gate for Valuation Re-Rating
Claire Weston
Demand isn't the problem anymore — so what is?
Management stated clearly: the constraint is no longer order flow but whether capacity, gross margin, and architecture position can all be delivered at once.
This means → the valuation lens on Coherent needs to shift — from "are orders sufficient?" to "can profits materialize?"
1.6T transceiver ramp has begun before 800G hit peak volume, showing customers aren't waiting for the final architecture — they're scaling 800G while pulling in the next generation.
In plain terms = customers are buying two generations simultaneously; Coherent's revenue backbone through 2026–2027 remains pluggable modules and high-speed optical engines.
Will CPO swallow the pluggable module?
Management directly rejected the "CPO replaces pluggables" narrative, framing the two as parallel paths.
Pluggables serve standardized scale-out deployment; CPO/NPO — near-package optics that place the optical engine closer to the switch chip — serve higher-density use cases.
Timeline: H2 2026 for scale-out revenue ramp, H2 2027 for scale-up revenue ramp — not consistent with the bearish view that CPO is a post-2028 concept.
InP expansion — the hardest proof point over coming quarters?
Coherent plans roughly 4× expansion in InP — indium phosphide, the core material for high-speed lasers — device capacity within two years: doubling by Q3 2026, then more than doubling again by 2027.
Device output to transceiver shipments carries a ~2–3 month lag. This means → a capacity doubling won't show in that same quarter's revenue; the validation window slides one quarter later.
On substrate supply, the company has locked in five geographically diversified suppliers, several capable of 6-inch wafers; management says 6-inch yields already exceed the mature platform.
This reflects a core differentiator: Coherent is internalizing critical laser-source capability rather than relying on outsourced assembly.
The platform spans wide — but where does pricing power come from?
In the CPO ecosystem, Coherent's coverage extends well beyond a single laser supplier — spanning VCSELs, isolators, thermoelectric coolers, PICs, detectors, polarization-maintaining fiber, and micro-lens arrays.
Polarization-maintaining fiber suppliers are essentially Coherent and Corning; management believes its own share is higher.
In plain terms = as CPO moves toward volume production, value may reconcentrate in the materials and component layers Coherent controls — the "picks and shovels" supplier ends up more durable.
OCS optical circuit switching — from niche to AI-cluster interconnect shift?
Coherent recently raised its OCS — optical circuit switching, routing data-center network paths with light instead of electrical signals — SAM to $4 billion; management suggested even that figure may be conservative.
Use cases are expanding: from scale-out spine-switch replacement to cross-cluster interconnect (DCI) and scale-up.
This means → if scale-up becomes the largest long-term opportunity, OCS stops being a small networking category and becomes part of a fundamental shift in how AI clusters connect internally.
Which numbers should investors watch?
Three core validation metrics: ① Are 1.6T transceivers ramping on schedule with ASP and gross margin intact? ② Is 6-inch InP yield still described as better than the legacy platform? ③ Is gross margin moving from ~40% toward 42%+?
The risk: any single link in the expansion chain — epitaxy, device fab, packaging, testing, module integration — slowing down could delay revenue or front-load capex and inventory.
In plain terms = if revenue grows but gross margin stalls, the market will question whether Coherent is simply expanding capacity for customers without capturing profit; only when revenue and gross margin improve together does the valuation have a logical basis to roll forward to FY28.
Content is for reference only, not financial advice.