mstack

AAOI

4 min read

AAOI — Applied Optoelectronics

TL;DR

AAOI is a vertically integrated optical transceiver and InP laser component maker with a historically bad execution reputation — but Irrational Analysis upgraded it to “worth chasing” in March 2026. The bull case rests on one claim: “massive internal InP/laser capacity” at a moment when InP is structurally scarce (Lumentum and Coherent each received $2B Nvidia equity investments and are still scrambling to ramp). The near-term catalyst is completing 800G qualification — optical reliability (GR-468) and manufacturing line qual are done; the remaining interop failure is diagnosed as a solvable firmware/Tx FIR coefficient problem, not a physics failure. The CPO laser pitch is not yet credible — OFC 2026 showed 11% wall-plug efficiency vs. Lumentum’s 13%, and no linewidth or RIN data disclosed.12

Business

Makes optical transceivers (qualifying 800G, developing 1.6T) and high-power InP laser components for potential CPO applications. Primary customers are hyperscale datacenters. Key asset: internal InP/laser fab capacity — vertically integrated when most optical vendors source InP externally. Historical track record is poor (“legendary levels of incompetence” per Irrational Analysis); management credibility is low.1

Thesis

  • Internal InP capacity is the structural bull claim. InP supply is constrained across the industry. Lumentum and Coherent received $2B Nvidia equity investments each and are still capacity-constrained. AAOI’s internal InP fab makes it a capacity-credible second source for hyperscalers seeking to diversify away from the Lumentum/Coherent duopoly — if qualification completes.1
  • 800G qualification is 80% done. The hardest steps — optical reliability (GR-468) and manufacturing line qual — are complete. The remaining issue is an interop failure diagnosed by Irrational Analysis as a Tx FIR coefficient mapping nonlinearity (when Rx requests a specific Tx FIR coefficient vector and gets a slightly different result due to circuit PVT variation, link training fails). Author frames this as “a normal problem that happens all the time” and “easy to fix” at the firmware level — not a fundamental optical defect.1
  • 800G BER data at OFC 2026 was encouraging. ~e-12 pre-FEC all lanes, test ran >2 hours (not cherry-picked). 1.6T BER was “a touch lower than I would like but still decent.” Positive signal that the underlying optical hardware is functional.2

Risks

  • Execution track record. “AAOI has a horrific reputation. Legendary levels of incompetence.” The interop problem was botched — CEO framing (“not our problem”) was read charitably by the analyst but could reflect management denial of a real issue.1
  • CPO laser is not competitive. OFC 2026: AAOI’s high-power CPO laser showed 11% wall-plug efficiency (vs. Lumentum’s 13% at comparable conditions). Critically, AAOI disclosed no linewidth or RIN data. Linewidth is the critical spec for ring-modulator CPO (the architecture Nvidia chose and mass-adopted). Lumentum is the only vendor publicly disclosing linewidth at OFC 2026. “No noise performance == messa not interested.” CPO laser TAM is inaccessible until this disclosure gap is closed.2
  • 1.6T transceiver BER was marginal. Better-performing lanes at OFC 2026 were in RTLR/LRO (linear receive only) mode, not full retimed — author noted this as a flag. Not a disqualifier, but below best-in-class performance.
  • No valuation anchor. No revenue, earnings, or multiple data in the corpus. The thesis is qualitative and catalyst-dependent.

Recent catalysts

  • 2026-03-27 — OFC 2026 Irrational Recap: AAOI booth visit — 800G BER ~e-12 pre-FEC (good), 1.6T BER “decent,” CPO laser at 11% WPE with no linewidth/RIN data (disqualified for CPO).2
  • 2026-03-01 — Irrational Analysis “AAOI Probably Worth Chasing”: post-Q4 earnings analysis — 800G qual status, interop failure root cause diagnosis (Tx FIR coefficient nonlinearity), author buy intention based on InP capacity bull case.1

Second-order reads

  • 2026-03-27 — Irrational Analysis, OFC 2026 Irrational Recap — AAOI’s CPO laser weakness (11% WPE, no linewidth) contrasts with Lumentum’s differentiation (13% WPE + disclosed linewidth) → reinforces the LITE bull case on CPO laser quality as a moat.
  • 2026-03-01 — Irrational Analysis, AAOI Probably Worth Chasing — InP scarcity framing here mirrors the supply constraint thesis in LITE and COHR; if AAOI qualifies, it adds transceiver supply diversification that benefits hyperscalers sourcing outside the Lumentum/Coherent duopoly.

Valuation & positioning

No multiples or financial figures in the corpus. Irrational Analysis expressed intent to buy after Q4 earnings. Thesis is binary — completion of 800G interop qualification vs. continued execution failure.1

Sources

LITE — dominant EML/transceiver maker and CPO CW laser leader; AAOI’s primary qualification competitor COHR — second major vertically integrated optics vendor; similar CPO CW laser positioning race MTSI — MACOM; photodetector/TIA vendor across transceivers including AAOI’s target markets SMTC — CTLE analog amplifier; goes into the same transceiver architectures AAOI qualifies into MRVL — DSP vendor; AAOI’s 800G transceivers may use Marvell DSPs for retimed variants

Footnotes

  1. Irrational Analysis — AAOI Probably Worth Chasing — 2026-03-01 ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7

  2. Irrational Analysis — OFC 2026 Irrational Recap — 2026-03-27 ↩ ↩2 ↩3 ↩4

Graph View

Backlinks