Report contents
At a glance
- Business model: Applied Optoelectronics sells optical transceivers, lasers and subassemblies to hyperscale data center operators, cable networks and telecom carriers. In-house indium phosphide laser production is a genuine offset to customer leverage. But buyers qualify two or three vendors per speed generation and negotiate price down as volume ramps, so Applied Opt does not set the price of its finished module.
- Financial quality: Revenue grew 84% in FY2025 and roughly doubled again in the September 2026 quarter, yet operating cash flow was negative in each year from FY2021 through FY2025 and free cash flow hit -$360 million in Q3 FY2026. Financing of about $1.6 billion funded the gap; cash of $600 million now sits against $280 million of debt, with liabilities at 28% of assets.
Business model
Business model: a merchant supplier selling into other companies' capex cycles
Applied Optoelectronics designs and manufactures the optical transceivers, lasers and subassemblies that carry data across fiber. Its customers are organizations, not consumers: hyperscale data center operators, cable and broadband network owners, and telecom equipment vendors and carriers. It sells physical units against purchase orders, recognizes revenue on shipment, and carries no subscription book. The useful classification is manufacturer and merchant component supplier, currently in an expansion stage: it is adding data center transceiver capacity and pushing into each new speed generation while its cable television (CATV) and telecom lines provide a smaller, slower base. AAOI does not create its own demand. Revenue follows a handful of network build schedules, and profit depends on how much of each generation's cost improvement the company keeps.
Who pays, and who sets the price
Customer concentration is the defining feature of the model. The annual report's disclosure of customers above 10% of revenue actually understates the exposure, because the disclosed names are hyperscalers whose own programs are lumpy. A large data center operator qualifies two or three module vendors per speed generation, tests them into its network, then negotiates price down as volume ramps. The buyer's alternative is a competitor module in the same form factor and the same chassis, which limits what AAOI can charge for a given part. Revenue here is won program by program, and the customer holds most of the leverage once a design is qualified and a second source is in place.
The genuine offset is vertical integration. Applied Optoelectronics grows and processes its own indium phosphide lasers instead of buying optics from a third party, which matters when laser supply is the binding constraint in a ramp and when in-house wafers lower unit cost and shorten lead times. Whether that shows up as pricing power or merely as a cost advantage the customer captures at the next negotiation is the open question. A supplier with a proprietary component can defend a margin for a generation or two; it rarely sets the price of the finished module.
Three businesses, one dominant profit driver
Data center transceivers drive growth and volatility. They sell into a bid market where the customer defines the specification, so volume can double while price per unit falls. CATV equipment (lasers, transmitters and amplifiers for cable networks) is steadier but tied to a small set of multiple-system operators and their upgrade cycles, which are infrequent and lumpy. Telecom and fiber-to-the-home sales are the smallest leg and depend on carrier capital budgets. Segment mix therefore moves consolidated gross margin more than cost control does, and a strong quarter can reflect a hyperscaler pulling orders forward rather than a durable step up in demand.
The manufacturing footprint adds a structural wrinkle. Production is split among Sugar Land, Texas; Taipei; and Ningbo, China, so tariff policy, export controls and the cost of shifting volume between sites all feed directly into unit economics. Building capacity ahead of a forecast converts a program delay into inventory and underabsorbed fixed cost. That is the core tradeoff in this model: capital committed to win the next hyperscale program is sunk before the volume arrives, and the customer knows it.
What would overturn the judgment
The price-taker characterization would be wrong if Applied Optoelectronics held gross margin roughly flat while data center volume expanded across two speed generations, and if the number of customers above 10% of revenue rose. Either outcome would mean the in-house laser supply is buying real commercial leverage. The opposite evidence is equally clear: revenue growth accompanied by sliding gross margin, rising inventory, and a stable or narrowing customer list would confirm that volume is being bought with price.
Financial quality
Financial quality: the ramp is real, the free cash flow is not
Earnings quality: Weak. Applied Optoelectronics grew revenue 84% in FY2025 and roughly doubled it again year over year in the September 2026 quarter, yet operating cash flow was negative in every fiscal year shown from FY2021 through FY2025 and free cash flow was -$360 million in Q3 FY2026 alone. Investors funded the gap: financing brought in about $1.6 billion across the first three quarters of FY2026, and equity rose from $1.11 billion at March 31, 2026 to $2.24 billion at September 30, 2026 while cumulative net losses were only about $30 million. That difference is issued stock.
The latest quarter's profit did not come from selling product. Gross profit of $80 million roughly matched operating expenses, leaving operating income near zero, so net income of about $6 million was earned below the operating line, consistent with interest on the $600 million cash balance. The prior quarter still lost roughly $23 million at the operating line.
Margins have stopped improving while volume climbs
Gross margin rose from 15.1% in FY2022 to 30.0% in FY2025, then sat between 27.7% and 29.1% for three straight quarters even as quarterly revenue went from $150 million to $270 million. Scale is not producing incremental margin. With operating expenses near $80 million a quarter, a 28% gross margin needs roughly $290 million of quarterly revenue just to break even at the operating line. The company has grown into its cost base without yet growing through it, and revenue would have to rise another 7% or so, at today's margin, before operations contribute anything.
Capital intensity is outrunning the revenue it is meant to serve
Capex, derived from free cash flow less operating cash flow, was about $180 million in FY2025 (39% of revenue) and roughly $360 million in Q3 FY2026, more than the quarter's entire revenue. Total assets went from $1.17 billion to $3.14 billion in nine months while annualized revenue went from about $520 million to $1.08 billion. Asset turnover, 0.53 in FY2025, is running near 0.34 on the latest quarter's annualized revenue, and ROA for that quarter was 0.2%. Net PP&E is not broken out in the disclosures, so ROIC cannot be calculated; what is visible is capital employed compounding faster than sales. Until utilization and returns appear, the expansion is a claim on future cash flow, not evidence of one.
Working capital is absorbing the cash the income statement reports
Inventory rose from $100 million in Q1 FY2025 to $380 million, and receivables from $170 million to $410 million, against revenue growth of 2.7x over the same span. Inventory is up 3.8x. On quarter-end balances, inventory equals roughly 180 days of cost of goods sold and receivables about 138 days of sales; comparing a rising revenue base against period-end balances overstates both, but the direction is unambiguous. Either the company is stocking ahead of orders it can see, or the ramp is being financed on its own balance sheet. Liquidity is not the constraint: liabilities are 28% of assets, the current ratio is 2.34, the quick ratio 1.77, and cash of $600 million sits against $280 million of debt. That cushion was bought, and it is what makes the burn survivable.
What to check next
The FY2024 loss is unexplained. A net loss of $190 million against an operating loss of $70 million and EBITDA of -$160 million leaves roughly $120 million arising below the operating line, and goodwill is not disclosed. The 10-K would show whether that was an impairment, an inventory write-down, or debt extinguishment, and whether current inventory is carried at recoverable value.
The Q2 FY2026 investing outflow. Investing used $570 million against $280 million of capex, so about $290 million went elsewhere. Short-term investments would be harmless; an acquisition or minority stake would change the reading of the equity raise.
Inventory composition and customer terms. Raw materials staged for a ramp and finished goods nobody wants look identical on a balance sheet. Inventory days, reserve levels and customer concentration separate them.
One discrepancy. The leverage series in the indicator data runs several points below liabilities divided by assets on the balance sheet (9.0% versus 28.3% at September 30, 2026). The tables below use the balance-sheet calculation.
Table 1: annual metrics, FY2021 to FY2025 plus latest interim
| USD m unless noted | FY2021 | FY2022 | FY2023 | FY2024 | FY2025 | Q3 FY2026 | Assessment |
|---|---|---|---|---|---|---|---|
| Revenue | 210 | 220 | 220 | 250 | 460 | 270 | Strong |
| Gross margin % | 17.8 | 15.1 | 27.1 | 24.8 | 30.0 | 28.2 | Adequate |
| Net margin % | -25.6 | -29.8 | -25.8 | -74.9 | -8.4 | 2.3 | Very weak |
| ROE % | -20.3 | -30.2 | -28.1 | -84.1 | -7.9 | 0.3 | Very weak |
| ROA % | -11.6 | -15.4 | -14.1 | -39.9 | -4.5 | 0.2 | Very weak |
| Liabilities/assets %† | 44.4 | 53.7 | 43.6 | 58.2 | 36.8 | 28.3 | Adequate |
| Current ratio | 1.78 | 1.32 | 1.85 | 1.77 | 2.63 | 2.34 | Strong |
| Quick ratio | 0.93 | 0.75 | 1.16 | 1.25 | 1.91 | 1.77 | Adequate |
| Asset turnover | 0.45 | 0.52 | 0.55 | 0.53 | 0.53 | 0.10 | Weak |
| Inventory turns | 1.7 | 2.2 | 2.2 | 2.5 | 2.4 | 0.6 | Weak |
| Receivables turns | 4.3 | 3.8 | 4.0 | 3.0 | 2.5 | 0.7 | Weak |
| Operating cash flow | -10 | -10 | -10 | -70 | -170 | ≈0 | Weak |
| Free cash flow | -20 | -20 | -20 | -110 | -350 | -360 | Very weak |
| Capex (FCF − OCF)‡ | 10 | 10 | 10 | 40 | 180 | 360 | Very weak |
| Capex/revenue % | 5 | 5 | 5 | 16 | 39 | 133 | Very weak |
*Three months; ROE, ROA, asset turnover and turns are single-quarter figures, not annualized. †Total liabilities / total assets, computed from the balance sheet; the indicator series runs 8 to 22 points lower and the two cannot be reconciled here. ‡Derived from rounded cash flow figures, so small values are approximate.
Table 2: quarterly metrics, Q1 FY2025 to Q3 FY2026
| USD m unless noted | Q1 25 | Q2 25 | Q3 25 | Q4 25 | Q1 26 | Q2 26 | Q3 26 | Assessment | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Revenue | 100 | 100 | 120 | 130 | 150 | 190 | 270 | Strong | Rising |
| Gross margin % | 30.6 | 30.3 | 28.0 | 31.2 | 29.1 | 27.7 | 28.2 | Adequate | Flat |
| Net income (approx.) | -9 | -9 | -18 | -2 | -14 | -23 | 6 | Very weak | Improving |
| Total assets | 650 | 800 | 980 | 1,170 | 1,570 | 2,300 | 3,140 | Weak | Rising fast |
| Total liabilities | 340 | 370 | 420 | 430 | 460 | 630 | 890 | Adequate | Rising |
| Net PP&E | n.d. | n.d. | n.d. | n.d. | n.d. | n.d. | n.d. | Not disclosed | n.a. |
| Cash | 50 | 70 | 140 | 220 | 450 | 500 | 600 | Strong | Rising |
| Inventory | 100 | 140 | 170 | 180 | 210 | 280 | 380 | Weak | Building |
| Receivables | 170 | 210 | 220 | 240 | 300 | 310 | 410 | Weak | Building |
| Operating cash flow | -50 | -70 | -30 | -30 | -80 | 10 | ≈0 | Weak | Volatile |
| Free cash flow | -80 | -90 | -80 | -130 | -140 | -270 | -360 | Very weak | Worsening |
| Capex (FCF − OCF) | 30 | 20 | 50 | 100 | 60 | 280 | 360 | Very weak | Worsening |
| Current ratio | 1.88 | 2.11 | 2.31 | 2.63 | 3.83 | 2.78 | 2.34 | Strong | Falling |
Net PP&E is not disclosed in any period shown, so fixed-asset intensity is measured here through capex. Net income for single quarters is estimated from reported net margin times revenue because the underlying income statement is rounded to $10 million.
Questions worth following
- Can AAOI fund its capacity build without issuing more equity?
Capex in the September 2026 quarter exceeded that quarter's revenue, and equity sales covered the gap. If operating cash flow can carry the remaining build, dilution stops and the balance sheet becomes a cushion instead of a dependency.
- Can AAOI earn an operating profit before interest income fades?
The September 2026 net profit arrived with operating income near zero, so interest on the cash raised is the likeliest source. Whether the company can earn a profit from selling product decides if the equity raise bought an asset or merely delayed another raise.
- Can AAOI turn its inventory and receivables build into cash?
Inventory and receivables together reached about $790 million against $270 million of quarterly revenue. Converting them at expected prices separates a self-funding business from one that keeps returning to shareholders, and discounting to move stock would hit a gross margin that has already flattened.
Generated by DeepVal AI. Data and models may contain errors or be outdated. For research only; not personalized investment advice.
Continue reading the full report
- Key risksWhat could break the investment case
- Valuation and margin of safetyWhat the current price assumes, and how much room for error it leaves
- SummaryOur overall judgement and what to watch next
Free to sign up · 20 research points · unlocking this report uses 10
20 points cover 2 full reports. No card needed.
Research reflects its generation date. Preview the first two modules.
More on Applied Opt
Applied Opt
Business, moat and key questions
Open company page