Co-packaged optics transfers margin to substrate integrators, not optics specialists

gen-co-packaged-optics · conviction — · status open · horizon — · as of 2026-08-07

Robust to undisclosed shares. 1 derived input under this thesis; redrawing every supply weight the industry does not publish moves none of them by more than 25%. Computed from evidence at most 17 days old (oldest input: broadcom).

Exhibits

Exhibit 1Relative performance, indexed to 100How the names in this thesis have traded against SOXX.

Series available as data/gen-co-packaged-optics.csv

Exhibit 2What the conviction is actually made ofEach premise and the number it composes to. A conjunction of plausible premises is far weaker than any of them.

Co-packaged optics shifts value capture from optical vendors to package integrators

† 4 premises marked supporting — shown and arguable, but the conclusion does not depend on them, so they are not multiplied into the composed figure. Citing a filed figure should not cost conviction.

One gating premise, so the conclusion is exactly as strong as it. The figure is an ordering device, not a calibrated probability — see how the numbers are made.

Weakest link: Co-packaged optics at 0.88 — Architecture is defined and enabling multiple startups, but production timelines remain uncertain

Coherent and Lumentum become component suppliers rather than system vendors

† 3 premises marked supporting — shown and arguable, but the conclusion does not depend on them, so they are not multiplied into the composed figure. Citing a filed figure should not cost conviction.

One gating premise, so the conclusion is exactly as strong as it. The figure is an ordering device, not a calibrated probability — see how the numbers are made.

Weakest link: Perimeter-vs-area scaling gap (the N-squared vs N fan-out dilemma) at 0.82 — Physical necessity drives CPO adoption, but timing depends on bandwidth inflection

Margin compression for optical suppliers exceeds consensus as CPO scales

† 2 premises marked supporting — shown and arguable, but the conclusion does not depend on them, so they are not multiplied into the composed figure. Citing a filed figure should not cost conviction.

One gating premise, so the conclusion is exactly as strong as it. The figure is an ordering device, not a calibrated probability — see how the numbers are made.

Weakest link: Silicon Photonics at 0.75 — SiPh enables CPO cost structure but maturity and yield remain adoption gates

The variant

Consensus

Co-packaged optics is framed as an optical architecture shift that benefits photonics suppliers. The market expects Coherent and Lumentum to capture incremental content as optical engines migrate from pluggable modules onto switch packages. TAMs model optical dollar growth proportional to bandwidth scaling.

Variant

Co-packaged optics is fundamentally a packaging integration play that compresses optical value capture. Moving the optical engine onto the package converts a discrete module sale—controlled by photonics vendors—into a component embedded in a heterogeneous substrate. The economic beneficiary is the entity that owns the package integration, which sits with semiconductor firms commanding 44-56% operating margins, not optical specialists running sub-13%. Broadcom and TSMC gain pricing authority; Coherent and Lumentum supply into a margin structure they no longer control.

Differentiator

Supply-chain analysts track optical content growth. They miss that co-packaging converts photonics from a complete product into a substrate component, transferring the customer relationship and assembly margin from optical specialists to the integrator. The 30-40 point margin gap between semiconductor integrators and photonics suppliers reveals who captures incremental dollars when the architecture changes hands.

Falsifiers

claim: Coherent or Lumentum operating margin declines by 200+ bps from current levels · criterion: Trailing twelve-month operating margin below 10.0% for Coherent or 7.5% for Lumentum · horizon: 2027-12-31 · settles: confirmed
claim: Broadcom discloses co-packaged optics revenue or design wins in networking segment · criterion: Earnings call or filing quantifies CPO contribution or names hyperscale design win · horizon: 2027-06-30 · settles: confirmed
claim: Optical specialists retain module-level pricing in CPO architectures · criterion: Coherent or Lumentum reports CPO ASP within 15% of comparable pluggable module · horizon: 2028-03-31 · settles: refuted

Open questions

What fraction of optical engine value transfers to the package integrator versus remaining with photonics die supplier?
Does CPO adoption accelerate fast enough to reset margin mix before optical specialists reposition business models?
Can Ayar, Celestial, or Lightmatter own integration and retain optical economics, or do they become IP licensors to TSMC/Broadcom?

Reasoning chain

Co-packaged optics shifts value capture from optical vendors to package integrators VALID

premises

Co-packaged optics0.88 strong
Architecture is defined and enabling multiple startups, but production timelines remain uncertain
Broadcom Inc.1.00 strong
Broadcom controls switch ASIC integration where optical engines land
Taiwan Semiconductor Manufacturing Company1.00 strong
TSMC provides advanced packaging substrate where photonics integrate
Broadcom Inc. — operating margin 44.2%1.00 strong
Establishes integrator margin baseline significantly above optical specialists
Coherent Corp. — operating margin 12.1%1.00 strong
Shows optical supplier margin 32 points below semiconductor integrator

88% probability integrators capture packaging margin that photonics vendors previously retained as module assemblers

Coherent and Lumentum become component suppliers rather than system vendors VALID

premises

Coherent Corp. supplies Co-packaged optics1.00 strong
Coherent participates but as input provider, not package owner
Lumentum Holdings supplies Co-packaged optics1.00 strong
Lumentum supplies components into architecture it does not control
Perimeter-vs-area scaling gap (the N-squared vs N fan-out dilemma)0.82 strong
Physical necessity drives CPO adoption, but timing depends on bandwidth inflection
Co-packaged optics — input-cost pressure at least 54.43511.00 strong
High input cost pressure constrains optical vendors' ability to defend pricing

82% probability optical specialists lose pricing power and customer control when photonics embed into substrate

Margin compression for optical suppliers exceeds consensus as CPO scales VALID

premises

Silicon Photonics0.75 strong
SiPh enables CPO cost structure but maturity and yield remain adoption gates
Taiwan Semiconductor Manufacturing Company — operating margin 56.1%1.00 strong
Integrator retains 44-point margin advantage over optical component suppliers
Lumentum Holdings — revenue growth $691.00 strong
Current growth masks future margin structure shift as architecture transitions

75% probability optical vendors face structural margin reset as they lose module assembly economics to integrators

Write-up

Pre-filled skeleton: gen-co-packaged-optics.md