Applied Optoelectronics

Applied Optoelectronics makes advanced optical transceivers and HFC networking products powering AI datacenters. Primary products: 800G (and next-gen 1.6T) optical transceivers sold directly to hyperscalers for GPU-to-GPU interconnect in AI clusters.

Applied Optoelectronics designs and manufactures optical transceivers and HFC networking gear; the investable story is still datacenter optics for hyperscaler AI clusters, with capacity now being expanded in Houston and backed by China-side working capital facilities.

Applied Optoelectronics designs and manufactures fiber-optic networking products, with the investable story centered on AI datacenter optical modules/transceivers rather than legacy cable-TV gear.

Applied Optoelectronics makes optical transceivers and HFC networking gear, with the investable story centered on high-speed datacenter optics for AI clusters.

• Q2 2026 revenue $191.922M (+86.4% YoY vs $102.952M) — 5th consecutive record quarter (8-K EX-99.1, 2026-08-06).
• Mix: Data Center $107.662M (56.1%), CATV $80.578M (42.0%), Telecom $3.411M (1.8%). Data Center +140% YoY; CATV still 42% (1.8 GHz upgrade cycle also contributing).
• GAAP GM 27.7% (vs 30.3% Q2-25); non-GAAP GM 29.8% (vs 30.4%). GM trending DOWN.
• GAAP net loss −$22.8M (−$0.28/sh); non-GAAP net income +$5.5M (+$0.06 diluted) — first non-GAAP profitability.
• Q3 guide: revenue $255–290M (midpoint $272.5M = +42% QoQ), non-GAAP GM 29–30.5%, non-GAAP NI $10.1–24.0M ($0.11–0.26/sh) on ~92.8M shares.
• Balance sheet (6/30/26): cash $499.7M, total assets $2.301B, liabilities $633.2M, equity $1.668B, A/R $314.0M. Equity +127% since FY25 — heavy equity-funded buildout.
• Concentration: top-10 customers = 99% of revenue; top-3 = 42% / 26% / 24% (92% combined, Q2-26); $211M of $314M A/R due from a single distributor (Digicomm International).

• monetization model: one-time hardware product sales (optical transceivers/modules, subsystems); no subscription/SaaS/license mix disclosed. Source: FY2023 10-K, sec.gov , accession 0001158114-24-000008.
• revenue mix: FY2023: Data Center ~51.8% ($112.6M), CATV ~41.8% ($91.0M), Telecom ~4.6% ($10.1M), FTTH ~1.8% ($3.9M); total $217.6M (10-K revenue/segment note).
• who pays: hyperscale data center operators, cable MSOs, telecom/FTTH network operators, network equipment OEMs (10-K Business/customer concentration).
• margin_value_capture: FY2023 GAAP gross margin 17.4%; operating margin negative; value capture low, tied to design/manufacturing yield (10-K MD&A/Risk Factors).

• customer classes: Hyperscalers, Data Center Operators, cable MSOs, telecom/FTTH network operators, network equipment OEMs (10-K Business/customer concentration, accession 0001158114-24-000008).

Applied Optoelectronics, Inc. (“AAOI”) is a vertically integrated designer and manufacturer of optical components, modules, and subsystems used in data centers, cable television/broadband networks, and telecom networks. The company monetizes through direct sale of hardware products — mostly optical transceivers, laser diodes, photodetectors, and optical subassemblies — not software subscriptions.

The exact current segment mix must be pulled from the FY2023 10-K segment note. As a memory/estimate only — needs primary source validation — AAOI’s revenue in recent years has been dominated by data-center transceivers, with the following rough composition:

This mix matters because data-center transceivers have different margin and cycle characteristics than CATV/telecom components. AAOI’s long-term bull case typically assumes data-center growth offsets CATV decline.

From memory/training data — not verified against EDGAR in this session — AAOI’s fiscal 2023 revenue was roughly $217–218 million and the company was still loss-making at the operating/net level. Exact figures need primary source validation.

AAOI has historically had volatile margins due to factory utilization, product mix shifts, restructuring charges, inventory write-downs, and pricing pressure. A high-quality financial check must use the audited 10-K income statement, not press summaries.

• Who pays: Data-center hyperscalers, cloud operators, network equipment OEMs, cable MSOs, and telecom system vendors pay per unit for optical modules/components.
• Revenue recognition: AAOI generally recognizes revenue when control transfers, typically on shipment or delivery depending on terms. Exact policy is in the 10-K revenue recognition note.
• Contract structure: Purchase orders and master supply agreements; not recurring subscription revenue. This creates lumpy, customer-concentrated revenue.
• Unit economics driver: ASP declines across optical transceivers are a structural headwind; AAOI must reduce cost or shift to higher-speed products to offset ASP erosion.

I cannot retrieve the current customer-concentration footnote in this environment. In historical AAOI filings, a small number of large customers have often accounted for a very large share of revenue. Training-data memory suggests that Amazon and Microsoft have appeared as 10%+ customers in prior periods, but that must be verified against the current 10-K. The current filing may name different or additional customers.

No customer name in this note should be treated as a current verified filing claim. The correct way to satisfy the “named customers from filings only” requirement is to pull the customer-concentration footnote from the latest 10-K.

AAOI sells to a concentrated set of buyers:

• Hyperscale data-center operators / cloud providers: historical filing memory suggests Amazon and Microsoft have been 10%+ customers; current names require 10-K validation.
• Network equipment OEMs: optical module suppliers often sell to OEMs that integrate modules into switches/routers.
• CATV / broadband equipment makers: buyers of lasers and transmitters for cable network upgrades.
• Telecom system vendors: buyers of long-haul/metro optical components.

Buyer archetypes include:

• Data-center network architects and supply-chain managers qualifying 400G/800G optics.
• Cable operator engineering teams upgrading to fiber-deep or DOCSIS 4.0 architectures.
• Telecom OEM procurement teams seeking second-source laser suppliers.

The common thread is high qualification requirements and concentrated order flow; a small number of design wins can determine the growth trajectory.

Applied Optoelectronics, Inc. is a manufacturer of optical devices, modules, and subsystems used in fiber-optic communications. The company’s business is hardware-centric: it sells physical optical components and modules, not recurring software or subscription revenue.

The main disclosed reporting/product categories in AAOI SEC filings have historically included:

1. Data center products — optical transceivers and related modules used inside hyperscale data centers for switch-to-server, switch-to-switch, and sometimes spine-to-spine or leaf-to-spine interconnect.
2. Cable television / CATV products — laser transmitters, receivers, and related modules used in hybrid fiber-coaxial networks.
3. Telecom products — modules used in wireless fronthaul/backhaul and other telecom transport.
4. FTTH / fiber-to-the-home products — optical modules used in passive optical networks.

These segment names are taken from AAOI’s business description and segment reporting structure in its SEC filings; the source of truth for the latest segment taxonomy is the most recent 10-K. Exact segment revenue percentages are needs primary source validation.

The table below is intentionally not filled with unverified memory numbers. Required source: AAOI 10-K segment note, which reports revenue by product category or segment. The exact document identifier is needs primary source validation.

AAOI monetizes through:

• Sale of optical transceivers and modules into data center, telecom, CATV, and FTTH applications.
• Sale of laser diodes, photodiodes, and sub-assemblies either internally consumed or sold to customers.
• Engineering and qualification support for custom designs, especially for large data center customers.

There is no meaningful SaaS-like recurring revenue stream apparent in the company’s disclosures. The model is therefore capacity-, qualification-, and order-flow-dependent.

The ultimate payers are:

• Hyperscale data center operators — for high-volume optical transceivers.
• Network equipment manufacturers / OEMs — who integrate AAOI modules into switches, routers, and transport equipment.
• Cable MSO operators and their equipment suppliers.
• Telecom carriers and infrastructure vendors for fronthaul/backhaul and access-network optics.

From the business description and segment reporting, the customer classes are:

• Hyperscale data center operators and their direct procurement organizations.
• Networking/telecom equipment manufacturers that embed optical modules in their systems.
• Cable operators and cable network equipment vendors.
• Telecom carriers and mobile infrastructure vendors.
• Distributors and resellers in smaller accounts.

Within those customer classes, the relevant buyers are:

• Optical module qualification engineers at hyperscalers or OEMs.
• Supply-chain and procurement teams managing multi-vendor qualification and pricing.
• Hardware engineering teams designing switch/router/transport platforms.
• NPI or new product introduction teams that approve specific module suppliers for new platforms.

These archetypes matter because AAOI’s success depends not just on selling a module, but on being qualified into a platform that will ramp over multiple quarters.

The rule here is strict: customer names must come from filings only.

• The AAOI 10-K historically discloses a customer-concentration note that identifies customers above a threshold, usually 10% of revenue.
• I am not naming a current customer in this draft because the exact note has not been pulled in this session.
• The correct source is: AAOI 10-K, financial statement footnotes, customer-concentration note.
• URL: https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0001158114&type=10-K&dateb=&owner=include&count=40
• Document identifier: needs primary source validation.

Applied Optoelectronics, Inc. (“AAOI”) is a fiber-optic networking company that designs and manufactures optical components, modules, and equipment used in data centers, cable-television/broadband networks, telecom networks, and fiber-to-the-home/enterprise networks. The company is vertically integrated in key optoelectronic processes, including laser design, epitaxy, packaging, and testing.

This qualitative description should be confirmed against Item 1 of the issuer’s most recent 10-K.

AAOI monetizes by manufacturing and selling optical components, transceivers, subassemblies, and network equipment. Revenue is generally recognized when control transfers to the customer under purchase orders or contractual arrangements. The precise revenue-recognition policy, warranty terms, and customer acceptance provisions are:

needs primary source validation

The likely buyer archetypes, to be confirmed from filings, are:

• hyperscale cloud/data-center operators;
• cable multiple-system operators;
• telecommunications carriers;
• enterprise network operators;
• OEMs and optical-system integrators.

None are affirmed here. The current named-customer disclosures are:

needs primary source validation

The correct place to identify them is:

• 10-K, Note on customer concentration
• 10-K, Risk Factors, customer concentration risk

Any investor-presentation customer list should be used only as Tier 2 or lower corroboration, not as a substitute for filing disclosure.

Applied Optoelectronics, Inc. is a vertically-oriented designer and manufacturer of optical communications products. The company’s SEC EDGAR company profile is filed under CIK 0001158114 (SEC EDGAR AAOI filing index, CIK 0001158114, access date 2026-04-26).

The company’s Form 10-K business description states that it designs, manufactures, and sells optical transceivers, optical transmitters, optical receivers, laser diodes, and related equipment for data center, cable television (CATV), telecommunications, and fiber-to-the-home applications. This is a product-sale model, not a subscription or software model: the company recognizes revenue when it ships hardware to customers or distributors, subject to standard revenue-recognition criteria. The payer is generally the purchaser of optical modules or laser components — principally large network operators, hyperscale data center operators, cable multiple-system operators (MSOs), and communications equipment vendors (SEC EDGAR AAOI 10-K index, CIK 0001158114, access date 2026-04-26).

The AAOI 10-K segments revenue into product groups: data center products, CATV products, telecom products, and other products. Qualitative revenue mix is as follows:

• Data center products: largest revenue category; represents the majority of revenue. Exact percentage for FY2023 is needs primary source validation because I could not re-pull the exhibit in this session.
• CATV products: second-largest category; includes headend optics, node transmitters, and related components. Exact percentage is needs primary source validation.
• Telecom products: smaller category; includes metro/long-haul transceivers and components. Exact percentage is needs primary source validation.
• Other products: includes FTTH/PON transceivers and custom devices. Exact percentage is needs primary source validation.

The structural point is that AAOI’s revenue is heavily concentrated in data center and CATV products. This concentration is visible in the Form 10-K segment discussion, even without re-pulling the exact percentage (SEC EDGAR AAOI 10-K index, CIK 0001158114, access date 2026-04-26).

AAOI monetizes by selling physical hardware into high-volume optical networks. The customer classes that pay are:

1. Hyperscale / data center operators — purchase high-speed optical transceivers, originally 100G and now increasingly 400G and above, for Ethernet switch-to-server and switch-to-switch interconnect.
2. Cable MSOs — purchase CATV transmission lasers, optical nodes, and headend optics for hybrid fiber-coaxial networks.
3. Telecom equipment vendors and carriers — purchase telecom transceivers and laser components for long-haul, metro, and access networks.
4. Other equipment makers — purchase laser diodes and photodetectors as components for their own systems.

The 10-K business description supports these buyer classes; the names of the specific end customers are generally not fully disclosed in the Form 10-K, but the filing does disclose customer concentration when any customer exceeds 10% of revenue (SEC EDGAR AAOI 10-K index, CIK 0001158114, access date 2026-04-26).

AAOI’s most important financial vulnerability is customer concentration. The Form 10-K’s risk factors and revenue concentration note indicate that a small number of customers account for a large share of revenue. Exact top-customer percentages are needs primary source validation for this session, but the structural feature is not in dispute: the company is dependent on a few large buyers.

This concentration matters because unit economics and marginal profitability are not spread across a diversified customer base. A single hyperscale qualification win can drive a large revenue step-change; a loss can create a large revenue hole. The worked calculation below formalizes that sensitivity.

Gross margin and operating margin are disclosed in the income statement and MD&A. The heavy mix of standard-form-factor optical transceivers tends to create gross-margin pressure because datacom transceivers often follow rapid annual price declines. The Form 10-K risk factors state that average selling prices have declined over time and may continue to decline.

Without re-pulling the exact figures, I will not assert a gross-margin number. The specific FY2023 gross margin and operating margin are needs primary source validation. They must be read from the FY2023 Form 10-K before feeding any conviction model (SEC EDGAR AAOI 10-K index, CIK 0001158114, access date 2026-04-26).

The 10-K business section supports the following customer archetypes:

1. Hyperscale data center operator — large buyer of 100G/400G transceivers; often designs its own network architecture and qualifies specific module vendors for specific data rates.
2. Cable operator / MSO — buys CATV headend and node optics; values long-term reliability and backward compatibility with existing HFC networks.
3. Telecom equipment manufacturer — buys optical transceivers and laser components as inputs into its own systems.
4. Network equipment OEM — buys FTTx/PON transceivers for broadband access equipment.

The buyer’s procurement team is typically a hardware engineering or supply-chain organization. Purchasing decisions are specification-based, not brand-driven: deliverable optical power, reach, data rate, power consumption, temperature range, and reliability data matter more than vendor name in standard form factors.

The Form 10-K does not always name its largest end customers. In the financial statement notes, companies commonly disclose concentrated customers as “Customer A” or “Customer B” rather than by name. Therefore, I will not invent a customer name.

What can be said from the filing structure:

• The 10-K does disclose that an identified number of customers exceed 10% of revenue. Exact percentage and count are needs primary source validation.
• Among data center customers, the most likely buyer archetype is a U.S.-based hyperscale operator, based on AAOI’s data center revenue and prior qualitative descriptions. However, I cannot cite a customer name from the Form 10-K without direct re-pull.
• Among CATV customers, the likely buyers are cable MSOs and their equipment vendors, but again, the filing may not name them.

If a name is needed for a conviction report, it must be taken from the actual 10-K customer concentration note or from a Tier-1 filing exhibit; it must not be assumed from news coverage. I mark top-customer identity as needs primary source validation.

AAOI monetizes by selling physical units. Revenue is recognized when control of the product transfers to the customer, generally upon shipment or delivery depending on contract terms. Customers typically issue purchase orders rather than long-term take-or-pay contracts. This means revenue visibility is short, order-book disclosures are limited, and the company is exposed to order push-outs or inventory rebalancing by a few large buyers.

Q1 2026 revenue $151.1M (+51% YoY vs $99.9M Q1 2025); 4th consecutive record quarter. Q2 2026 guide $180–198M. Gross margin 29.1% GAAP. Non-GAAP net loss $4.9M ($0.07/share). Source: SEC 8-K EX-99.1 filed 2026-05-07.

FY2025 revenue $455.7M vs $249.4M FY2024 (+82.8% YoY); FY2025 gross profit $136.9M (30.0% gross margin) and operating loss $54.6M (-12.0% operating margin). Q1 2026 revenue was $151.1M. Source: SEC companyfacts / 10-Q filed 2026-05-07.

FY2025 revenue $455.7M (+82.8% YoY); TTM revenue $507.0M; market cap $9.71B; forward P/E 66.48; analysts rate it Buy with a $151.30 target. Source: StockAnalysis overview + SEC 10-K.

Latest quant snapshot shows market cap 10.89B, P/E -205.59, Forward P/E 74.58, institutional ownership 71.53% (daily-picks quant refresh). StockAnalysis financials page shows gross margin 29.64% and operating margin 6.74%; revenue table extraction was not reliable from the live page, so annual revenue needs primary-source validation.

Optical transceivers are the irreplaceable interconnect bottleneck for AI GPU clusters. As AI cluster sizes scale from thousands to hundreds of thousands of GPUs, 800G→1.6T optics become the critical bandwidth constraint. AAOI is a U.S.-domestic manufacturer of these components — a strategic moat in a tariff environment.

Optical interconnect remains a hard bottleneck in AI cluster scale-out, and AAOI is adding cleanroom / fab capacity into the 800G→1.6T transition while funding the ramp with a larger credit line.

AI cluster scale-out keeps pushing optics demand from 800G toward 1.6T, and AAOI is adding cleanroom and fabrication capacity before that wave peaks.

The bottleneck is AI datacenter interconnect bandwidth: as GPU clusters scale, 800G and 1.6T optics remain a necessary enabling component, and AAOI sells directly into that constraint.

• Q1 2026 first volume shipment of 800G transceivers to a large hyperscale customer completed (SEC 8-K 2026-05-07)
• Q1 2026 manufacturing capacity nearly 100,000 units/month of 800G; doubling Houston-area footprint via $102.25M real estate acquisition/lease (3 buildings, ~736K sq ft) (SEC 8-K 2026-05-08)
• Strong Q2 2026 guidance $180–198M implies ~33% sequential growth; "significantly larger growth expected starting in Q3 as additional capacity comes online" (CEO Dr. Thompson Lin, SEC 8-K 2026-05-07)
• FY2025 revenue $455.7M vs FY2024 $249.4M; gross profit $136.9M; operating loss $54.6M, implying 30.0% gross margin and -12.0% operating margin (SEC companyfacts; 2025 10-K).
• Q1 2026 revenue reached $151.1M (10-Q filed 2026-05-07), confirming the growth profile is still running hot.
• 2026-07-01 8-K: AAOI signed a $94.1M design-build agreement with LCC3 Solution Inc. for OMD 3 (FAB4), including 195,591 sq ft of ISO Class 6 cleanroom space plus office, shipping/receiving, quality and reliability testing, and inventory operations; targeted substantial completion Jan. 10, 2027.
• 2026-06-16 8-K: Global Technology, Inc. entered a one-year RMB 500M credit line with Shanghai Pudong Development Bank, doubling the prior RMB 250M facility to support working capital and fixed asset needs.
• 2026-06-05 8-K: shareholders approved the 2026 Equity Incentive Plan, authorizing 2.5M new shares, which supports retention / hiring through the buildout.
• 2026-07-01 8-K: AAOI signed a $94.1M design-build agreement with LCC3 Solution Inc. for the OMD 3 (FAB4) project, covering 195,591 sq ft of ISO Class 6 cleanroom space plus office, shipping/receiving, quality/reliability testing, and inventory operations, with substantial completion targeted for Jan. 10, 2027.
• 2026-05-07 8-K / 10-Q: first volume shipment of 800G transceivers to a large hyperscale customer; Q1 2026 revenue $151.1M (+51% YoY).
• StockAnalysis overview: market cap $9.71B, TTM revenue $507M, forward P/E 66.48, and earnings date Aug. 6, 2026.
• OIF’s June 4, 2026 search results / meeting coverage show new projects around near-package optics, 1600ZR, data center/optical network coordination, and energy-efficient interfaces, which supports the view that the industry is still standardizing the 1.6T transition.
• Quant refresh updated AAOI to market cap 10.89B and inst own 71.53%.
• StockAnalysis live page shows gross margin 29.64% and operating margin 6.74%.

1. Capacity-to-revenue bridge (the binding constraint). 650K units/mo × 12 = 7.8M units/yr of 800G/1.6T. At a blended ASP of ~$500–700 (800G ~$400–800, 1.6T ~$800–1,500), that implies ~$3.9B–5.5B/yr of AI-optics revenue capacity vs Q2-26 datacenter revenue of $107.7M (annualized ~$431M) → ~9–13× headroom. Capacity, not demand, is the limiter.
2. ASP compression (disconfirming). CEO: 800G units "more than doubled sequentially." Datacenter revenue Q1→Q2 rose only $81.4M→$107.7M (+32% QoQ). Unit volume 2× against revenue +32% implies blended ASP −34% QoQ — pricing pressure even while demand exceeds supply.
3. Dilution = cost of the capacity moat. Aug 21 ATM for up to $600M at last price $129.10 → ~4.65M shares ≈ 5.0% dilution (on ~92.8M shares). Amazon warrant adds up to 7.95M shares (~8.5% potential) at $23.70 strike — deeply in-the-money. Total potential dilution ~13–14%, but raised at ~5.4× the warrant strike (cheap capital at peak valuation).

• technical_scarcity: 6/10 — real but contestable. The scarcity is U.S. high-volume cleanroom capacity + hyperscaler qualification + Amazon's strategic warrant, not proprietary silicon; multiple module vendors exist. (Cite: 650K-unit ramp, Amazon warrant, 3.25× capacity build.)
• economic_capture: 5/10 — weak pricing power. GM declining (30.3%→27.7%), GAAP loss −$22.8M, non-GAAP margin only ~2.9%, and ~13–14% dilution ahead. Revenue capture is strong (+86% YoY) but unit economics are thin.
• time_to_monetize: 7/10 — monetizing NOW. Non-GAAP profitable this quarter, Q3 guide +42% QoQ midpoint, capacity 3.25× by year-end, demand > supply through mid-2027.

1. Incremental Houston footprint commitment (single month, Aug 31–Sep 4):

• Buildings 4+5 = 1,093,807 sq ft; Blue Ridge B3 = 153,928 sq ft → ~1.25M sq ft of new space added in one week.
• Year-1 combined rent: ($220,835.32 + $457,325.02) × 12 = $8.14M/yr; steady-state (mo 109–120): ($307,582.40 + $636,968.43) × 12 = $11.33M/yr. Straight-line 10-yr total ≈ ~$97M.
• Total incremental capacity commitment ≈ $97M (leases) + $11.3M (deposits) + $2.55M (steel/solar) + $26.78M (purchase) ≈ ~$138M — the largest single-month capacity commitment in the note's history.

1. Fixed-cost overhang vs profit (the disconfirming math): year-1 rent of $8.14M ÷ Q2 2026 non-GAAP net income of +$5.5M = ~1.5× — the new annual rent alone exceeds the entire quarter's non-GAAP profit. At Q2 GAAP GM 27.7%, ~$2.0M/quarter of fixed occupancy cost must be absorbed before the ramp contributes to profitability. This is the capacity-cycle downside: the ~$138M is a fixed bet that demand outpaces capacity through mid-2027.

1. Capacity-to-revenue bridge (updated): >930K units/mo by end-2027 = 11.16M units/yr; at blended ASP $500–700 (800G ~$400–800, 1.6T ~$800–1,500) → ~$5.6–7.8B/yr revenue capacity vs Q2-26 datacenter revenue $107.7M (annualized ~$431M) → ~13–18× headroom. Capacity, not demand, remains the binding constraint.

1. $200M 1.6T order materiality: $200M ÷ Q2-26 total revenue $191.9M = ~1.04× a full quarter; ÷ Q2 DC revenue $107.7M = ~1.9× the entire datacenter line. A single customer order ≈ one quarter of the whole company.

Toll-road: unchanged 6 STRONG PROXY (bypass 2 / concentration 2 / hyperscaler 2). The $200M single-customer 1.6T order reinforces — rather than diversifies — the top-3 = 92% concentration risk.

\[

Revenue\ impact = \$200M \times 0.45 \times 0.20

\]

\[

Revenue\ impact = \$200M \times 0.09 = \$18M

\]

That would be an 18.0% revenue decline on the illustrative base.

If the latest AAOI 10-K discloses a top customer above 30% of revenue, the risk of an order break is severe because optical module contracts are often platform-specific and cannot be instantly replaced by another customer. This is a load-bearing conviction-scoring input.

The actual customer concentration percentage is needs primary source validation.

`

Variable cost per unit = 400 + 40 + 25 + 10 = 475

Contribution per unit = 800 - 475 = 325

Break-even units = 32,500,000 / 325 = 100,000

`

This table shows that a high fixed-cost, price-competitive transceiver business becomes break-even-sensitive very quickly. It should be recalibrated with actual AAOI gross-margin data from the 10-K. Until then, the AAOI-specific parameters are:

needs primary source validation