AAOI’s 800G/1.6T demand evidence is meaningful, and the $100.15 close is 40% below June 17 and 57% below the 52-week high. The reset makes the stock researchable again, but not yet ownable: spot already sits near the upper end of a reasonable 2027 base range while Q1 free cash flow was deeply negative and the current post-ATM cash and share count remain undisclosed.
AAOI is becoming a two-engine company: high-speed data-center optics and a still-material CATV franchise. The industry tailwind and announced orders are credible; the variant question is whether vertical laser integration, automation and U.S. capacity create better per-share economics than larger peers. The answer is not yet visible in reported margins or cash flow.
Q1 revenue grew 51% year over year to $151.1 million; data-center revenue grew 154% to $81.4 million. AAOI disclosed more than $200 million of 1.6T orders and $124 million of 800G orders, and guided Q2 revenue to $180-$198 million. S2 S3 S15 S15b
Q1 had a $14.3 million GAAP net loss, $85.4 million operating cash outflow and $58.2 million of capex, for negative $143.6 million free cash flow. Accounts receivable reached $299.0 million, inventory $206.2 million, and the top ten customers represented about 98% of revenue. S2
The market may be underestimating the value of internal lasers and domestic qualification capacity—but it may also be overvaluing nameplate unit targets. At $100.15, AAOI is cheaper than optics peers on forward sales but more expensive than their median on forward EBITDA, so margins and dilution are the decisive variables. S1 S17 S17b S17c
At the visible diluted bridge, AAOI trades at about 4.8x forward NTM enterprise value to sales and 29.6x forward NTM EBITDA. The current value also equals roughly 3.3x Fiscal.ai’s mean 2027 revenue estimate. That is not a distressed price: it assumes a substantial 2027 scale-and-margin step.
Q1 is the latest reported quarter. The balance sheet was strengthened by equity issuance, but the manufacturing ramp consumed cash and subsequent property, lease and cleanroom commitments prevent treating reported cash as excess.
Fiscal.ai’s normalized market cap of $8.99B and enterprise value of $8.83B use a stale 78.971 million-share count and a different quote. This report reprices the May 13 issuer-disclosed 80.243 million basic shares at $100.15, then separately models visible awards, the in-the-money convert and vested Amazon warrant. New ATM usage remains unknown. S1 S6 S7
This is the most important unresolved section. The items below should not be mechanically summed: timing overlaps, facilities may be undrawn, lease present values are unavailable, and some contracts may sit inside management’s capex plan. They show why March cash cannot be treated as current excess cash.
| Capital / funding item | Amount | Investor read | Evidence |
|---|---|---|---|
| Q1 cash and restricted cash | $449.4M | Filed balance-sheet liquidity; unrestricted cash was $439.7M. | Q1 10-Q |
| Q1 net cash | $159.3M | Fiscal.ai bridge includes debt carrying values and recorded leases; it is not post-quarter cash. | Fiscal.ai / Q1 10-Q |
| Valuation net-cash bridge | $163.8M | Uses unrestricted cash, recorded loans/acceptances/leases and the convertible’s $125M principal; used for basic EV. | Q1 10-Q |
| Post-quarter prior-ATM cash | ~$107.4M | Mechanical cash received after March 31 from the ATM completed April 2; before Q2 burn. | Q1 10-Q / May ATM filing |
| Pearland property purchase | $58.4M | Purchase closed in April; a real cash use, not an option. | April 8-K |
| Signed cleanroom contracts | $125.0M | $30.9M February design-build contract plus $94.1M July contract; timing and overlap with capex guidance matter. | SEC filings |
| Three new 123-month leases | $5.57M initial annual base rent | Signed after Q1; final annual base rent approaches $7.76M, plus operating costs. Present value was not disclosed. | May 8-K |
| Texas semiconductor grant | $20.85M | Texas CHIPS Act support against more than $279M of planned investment; not a federal CHIPS award. | Texas Governor |
| China credit capacity | RMB500M / ~$73.9M | Revocable facility and financing capacity, not debt until drawn. | June 8-K |
| New ATM authorization | Up to $600M | Usage since May 14 has not been disclosed. At spot, full use would be roughly six million new shares before fees. | May 424B5 |
Management said each remaining 2026 quarter’s “capital investments” should exceed Q1’s $68.7 million, while the filed capex number was $58.2 million. The distinction matters. Q2 must reconcile operating cash flow, capex, property and cleanroom payments, lease commencement, debt draws and ATM proceeds. S2 S4
Nameplate capacity only matters if units are qualified, produced at acceptable yields, shipped, collected and converted into margin. Texas reduces geographic risk but can add startup inefficiency, duplicated infrastructure and U.S. labor cost before utilization scales.
| Share-count item | Shares | Treatment |
|---|---|---|
| May 13 basic shares | 80.243M | Issuer-disclosed; latest basic count available. |
| Reserved RSUs + PSUs | 2.727M | Conservative full-share treatment; performance outcome remains uncertain. |
| 2030 convertible shares | 2.886M | In the money at spot; diluted bridge removes $125M principal from debt. |
| Vested Amazon warrant, treasury-stock method | 1.011M | Only the vested 1.324M warrant tranche; exercise price $23.6956. |
| Visible diluted bridge | 86.867M | Before new ATM usage and unvested success-linked Amazon warrants. |
| Scenario share range | 90M-100M | Explicit underwriting range for stock comp, ATM use and success-linked warrant dilution. |
The visible diluted bridge is approximately 86.9 million shares before any use of the new ATM and before the 6.62 million Amazon warrants that only vest against future purchase thresholds. The valuation scenarios deliberately use 90-100 million shares rather than presenting false precision.
AAOI scores 47/90, or 52/100 normalized. Growth torque and market fit are strong; cash conversion, margin proof and funding quality keep the overall score in watchlist territory.
The investable growth products are 800G and 1.6T transceivers, SiPh light engines, ELSFP and high-power lasers for future CPO/NPO architectures. Management reported only $4.6 million of 800G revenue in Q1, so the announced order book and capacity targets are still much larger than demonstrated revenue. S4 S15 S15b S16
Quantum18/Quantum12 amplifiers, nodes and QuantumLink management software support DOCSIS 4.0 upgrades. Commercial traction is real, but AAOI has not disclosed recurring software revenue, attach rates or standalone economics; it should not receive a SaaS valuation premium without evidence. S20
AAOI makes all laser chips in Sugar Land and uses Taiwan and China for components, transceivers, CATV equipment and labor-intensive assembly. In 2025, 57.5% of revenue was associated with products manufactured in China and 38.2% with Taiwan; Texas expansion is strategic, but not automatically margin-accretive. S5
This is the practical underwriting frame for AAOI from here.
Q2 lands inside or above the $180-$198 million guide with 29%-30% non-GAAP gross margin; Q3 guidance supports roughly $265 million; high-speed volume raises margins; and the cash/share-count bridge shows financing is becoming optional rather than structural.
800G/1.6T revenue grows while yields, receivables, inventory and capex keep cash flow deeply negative; Q3 guidance falls materially below consensus; or the new ATM is used heavily before returns are visible.
Actual 800G units and revenue, initial 1.6T shipment timing, Q3 revenue and margin guide, current basic/diluted shares, ATM proceeds, cash, operating cash flow, capex, receivables and inventory. S14
AAOI fits the AI-infrastructure build because high-radix switching increases optical bandwidth and power constraints. Market relevance is not the debate. The debate is whether AAOI’s manufacturing stack captures durable economics before architectures, pricing or larger suppliers catch up.
Internal InP lasers/light engines, MBE and MOCVD process know-how, automation, long qualification cycles and domestic capacity can create sticky design wins. The moat is moderate and execution-dependent—not unique—because Coherent and Lumentum also have broad SiPh, EML, laser and vertically integrated portfolios. S5 S16 S17 S17b
Management targeted combined 800G/1.6T capacity from about 100,000 units per month at Q1 exit to 150,000 in Q2, more than 650,000 at year-end 2026 and more than 930,000 at year-end 2027. These are company targets, not verified saleable output or yields. S4 S11 S12
Open interfaces, multi-sourcing, concentrated buyer power and rapid product cycles constrain durability. Innolight is the closest scaled transceiver comparator; Lumentum and Coherent offer broader technology portfolios, and Fabrinet is the cleaner manufacturing-execution comparator. S5 S17 S17b S17c
“AI needs optics,” 800G/1.6T scarcity, announced orders and Texas construction are visible. A differentiated thesis now requires evidence of qualified yields, better gross margin, customer diversification and cash conversion—not another capacity press release.
Six reported quarters, the Q2 company-guided midpoint, and three forward-estimate quarters through Q1 2027. Revenue growth is year over year. Earnings use company-reported non-GAAP diluted EPS for history, company non-GAAP guidance for Q2 and Fiscal.ai adjusted-estimate convention thereafter; the table also states the historical GAAP comparator.
Interactive chart requires JavaScript; the same data is in the table below.
Interactive chart requires JavaScript; the same data is in the table below.
| Quarter | Status | Revenue | Revenue YoY | Adjusted EPS | Earnings read | Basis |
|---|---|---|---|---|---|---|
| 2024 Q4 | Reported | $100.3M | +65.9% | $-0.02 | Adjusted loss narrowed | Company-reported non-GAAP diluted EPS; GAAP diluted EPS was -$2.60 |
| 2025 Q1 | Reported | $99.9M | +145.5% | $-0.02 | Adjusted loss narrowed | Company-reported non-GAAP diluted EPS; GAAP diluted EPS was -$0.18 |
| 2025 Q2 | Reported | $103.0M | +137.9% | $-0.16 | Adjusted loss | Company-reported non-GAAP diluted EPS; GAAP diluted EPS was also -$0.16 |
| 2025 Q3 | Reported | $118.6M | +82.1% | $-0.09 | Adjusted loss narrowed | Company-reported non-GAAP diluted EPS; GAAP diluted EPS was -$0.28 |
| 2025 Q4 | Reported | $134.3M | +33.9% | $-0.01 | Near adjusted breakeven | Company-reported non-GAAP diluted EPS; GAAP diluted EPS was -$0.03 |
| 2026 Q1 | Reported | $151.1M | +51.4% | $-0.07 | Adjusted loss widened | Company-reported non-GAAP diluted EPS; GAAP diluted EPS was -$0.19 |
| 2026 Q2 | Company guide | $189.0M | +83.6% | $0.00 | Breakeven guide | Company non-GAAP guide midpoint; Fiscal.ai consensus is $190.5M and $0.015 |
| 2026 Q3 | Analyst estimate | $264.6M | +123.0% | $0.28 | Turns positive | Fiscal.ai mean estimate |
| 2026 Q4 | Analyst estimate | $410.0M | +205.4% | $0.67 | Turns positive | Fiscal.ai mean estimate |
| 2027 Q1 | Analyst estimate | $508.2M | +236.2% | $0.86 | Turns positive | Fiscal.ai mean estimate |
Fiscal.ai exposes 2026 and 2027 third-party forward estimates but does not expose the estimate-set timestamp. The unusually wide ranges are as important as the means. A third year is shown only as an explicit model extension—not consensus.
Interactive chart requires JavaScript; the same data is in the table below.
Interactive chart requires JavaScript; the same data is in the table below.
| Year | Status | Revenue | Revenue Growth | Earnings per share | Earnings Growth | Source |
|---|---|---|---|---|---|---|
| 2025A | Reported | $455.7M | +82.8% | $-0.26 | Adjusted loss narrowed | Revenue from 2025 Form 10-K; adjusted EPS from company/Fiscal.ai consensus-basis history |
| 2026E | Consensus | $1.03B | +126.0% | $0.89 | Turns positive | Fiscal.ai mean; revenue range $886.3M-$1.10B from 5 estimates; adjusted EPS range $0.45-$1.27 from 6 estimates |
| 2027E | Consensus | $2.57B | +149.9% | $4.77 | +437.2% | Fiscal.ai mean; revenue range $1.84B-$3.50B from 4 estimates; adjusted EPS range $3.35-$8.02 from 5 estimates |
| 2028E | Model extension | $3.22B | +25.0% | $6.20 | +30.0% | No accessible Fiscal.ai 2028 consensus; explicit model extension of +25% revenue and +30% adjusted EPS |
The quarterly mean bridge through 2026 totals about $1.016 billion, below the $1.030 billion annual mean and management’s verbal outlook above $1.1 billion. This is a timing/coverage mismatch, not one unified forecast. The 2027 revenue range of $1.84B to $3.50B and adjusted EPS range of $3.35 to $8.02 show how little consensus exists on ramp economics. S1 S4
Peer values were repriced at July 24 closes using Fiscal.ai balance-sheet bridges and closest forward NTM estimates. “NTM vs TTM step-up” is not strict same-period growth; it only shows the magnitude of the forward model versus the latest trailing base.
| Company | Close | Market Cap | EV | TTM Revenue | Forward NTM Revenue | NTM vs TTM Step-up | NTM EBITDA Margin | TTM Gross Margin | EV / NTM Sales | EV / NTM EBITDA | NTM P/E | Role / Read-through |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| AAOI basic | $100.15 | $8.04B | $7.87B | $507M | $1.765B | +248% | 16.1% | 29.6% | 4.46x | 27.7x | 35.3x | Direct subject; highest ramp and dilution risk. |
| AAOI diluted overlay | — | $8.70B | $8.41B | — | $1.765B | — | 16.1% | 29.6% | 4.77x | 29.6x | — | Includes visible awards, in-the-money convert and vested-warrant treasury-stock dilution; excludes undisclosed new ATM use. |
| Lumentum (LITE) | $762.99 | $54.71B | $54.85B | $2.488B | $5.606B | +125% | 41.8% | 40.8% | 9.78x | 23.4x | 41.6x | Core vertically integrated optics peer; far better margin proof. |
| Coherent (COHR) | $282.39 | $55.22B | $56.56B | $6.602B | $9.573B | +45% | 25.9% | 37.0% | 5.91x | 22.8x | 34.1x | Direct optical exposure but much more diversified. |
| Fabrinet (FN) | $476.04 | $17.06B | $16.12B | $4.235B | $5.682B | +34% | 12.6% | 12.0% | 2.84x | 22.4x | 27.5x | Manufacturing-execution comp, not a branded-product peer. |
| Ciena (CIEN) | $390.96 | $55.36B | $55.54B | $5.569B | $7.417B | +33% | 24.0% | 43.0% | 7.49x | 31.2x | 45.2x | Systems/software demand read-through, not a clean module comp. |
AAOI’s visible diluted 4.77x forward-sales multiple is below the 6.70x peer median, but its 29.6x forward-EBITDA multiple is about 28% above the 23.1x median. It looks cheap only on sales because the margin, funding and dilution risk are worse. Lumentum and Coherent are the core technology comparators; Fabrinet and Ciena are secondary execution and demand read-throughs. S1 S17 S17b S17c
Forward enterprise value to sales is the primary framework because current earnings and cash flow are distorted by the ramp. Forward EBITDA is the sanity check. The result below is a decision range—not a target price—and explicitly sensitizes revenue, margins, multiples, cash retention and dilution.
| Case | 2027 Revenue | EBITDA Margin | EV / Sales | Enterprise Value | Diluted Shares | Net Cash | Value / Share | Return vs $100.15 | What must be true |
|---|---|---|---|---|---|---|---|---|---|
| Bear | $1.836B | 12%-14% | 2.0x-2.5x | $3.67B-$4.59B | 94M-100M | $0-$0.2B | $37-$51 | -63% to -49% | Ramp slips while inventory, receivables and capex stay high; ATM becomes necessary funding. |
| Base | $2.479B-$2.574B | 16%-18% | 3.0x-3.5x | $7.44B-$9.01B | 90M-95M | $0.2B-$0.5B | $80-$106 | -20% to +6% | Median-to-mean revenue, improving yields and margin, controlled but non-zero dilution. |
| Bull | $3.502B | 20%-23% | 4.0x-5.0x | $14.01B-$17.51B | 90M-100M | $0.4B-$0.8B | $144-$203 | +44% to +103% | Near-high estimate delivery, successful 800G/1.6T yields, customer diversification and rerating. |
Every 0.5x on $2.5 billion of revenue changes enterprise value by about $1.25 billion, or roughly $13-$14 per share on 90-95 million shares.
Every five million incremental shares reduces value per share by roughly 5%-6% around a 90 million-share base.
Every $250 million of retained net cash adds roughly $2.6-$2.8 per share. A five-point EBITDA-margin miss on $2.5 billion of revenue removes $125 million of EBITDA and would likely compress the multiple too.
Q2 cash, ATM usage, current shares, cleanroom-payment timing, lease liabilities, working capital and saleable yields are unknown. The range is designed to show which assumptions control the outcome. Spot near the upper base means upside requires execution closer to the bull case—not merely continued enthusiasm for optics.
Evidence-linked thesis claims are separated from company claims and analyst judgment.
Verified: Q1 data-center revenue grew 154% and AAOI has disclosed substantial 800G/1.6T orders. Company claim: capacity can exceed 650,000 units per month by year-end. Judgment: internal lasers, automation and U.S. capacity could support sticky wins, but only if they deliver peer-competitive margin and cash returns. S2 S4 S15 S15b
The market may underweight vertical integration and domestic laser capacity, but the broader optical shortage, announced orders and Texas build are already visible. There is no confirmed edge until reported yields, margin, diversification and cash conversion separate AAOI from a capacity-funded cycle.
Q2 revenue below $180 million, non-GAAP gross margin below 29%, or Q3 revenue guidance materially below roughly $265 million.
800G volume materially misses the expected near-fourfold sequential increase, or 1.6T shipments slip beyond Q3/Q4.
Receivables, inventory and capex keep operating/free cash flow deeply negative while the new ATM is used materially.
Gross margin fails to inflect toward the mid-30s by Q4, making management’s greater-than-40% exit aspiration implausible.
Sugar Land or Pearland delays, cost overruns or poor yields undermine saleable output; larger peers win the relevant SiPh/EML sockets.
Digicomm receivable concentration worsens, announced purchase orders are reduced/rescheduled, or CATV proves a cyclical hardware peak.
These are thesis-specific risks rather than generic small-cap hardware risks.
The May program authorizes up to $600 million of additional ATM issuance. Visible awards, the in-the-money convert and vested Amazon warrant already lift the modeled count from 80.24 million to about 86.87 million before new ATM use. S2 S6
Top-ten customers were about 98% of Q1 revenue. Digicomm represented 44.1% of revenue and 74.5% of receivables, while Microsoft was 28.8% of 2025 revenue. A single timing or collections issue can dominate a quarter. S2 S5
Q1 capex was $58.2 million, and management expects much higher spending through the ramp. Cleanroom construction and Texas startup add schedule, yield, labor-cost and underutilization risk before nameplate capacity becomes saleable. S2 S11
Purchase orders are generally short term, cancelable or reschedulable, and AAOI says backlog is not a reliable predictor. The Amazon warrant’s unvested portion requires future purchases; it is not a committed order. S2 S5 S15
SiPh, EML, CPO/NPO and pluggable architectures evolve rapidly. Larger peers can multi-source, spend more on R&D/capacity and pressure module prices before AAOI’s cost curve matures. S16 S17 S17b
China and Taiwan remain central to manufacturing. Texas reduces geopolitical exposure but can add U.S. labor cost and duplicated infrastructure; China assembly also does not automatically determine tariff origin for every product. S5
CATV remains roughly 44% of quarterly revenue. QuantumLink may improve stickiness, but recurring software economics are undisclosed; a DOCSIS hardware peak would offset part of the optics ramp. S20
Historical GAAP losses coexist with forward non-GAAP estimates. Stock compensation, warrant contra-revenue, convertible treatment and expansion depreciation can make adjusted EPS overstate per-share cash economics.
Primary filings and the full Q1 event package are strong. Evidence confidence remains medium because the decisive variables—post-May ATM usage, current cash, saleable yields, customer ramp durability and segment economics—will not be resolved until Q2.
| ID | Source | Type | Date | Coverage | Confidence | Use |
|---|---|---|---|---|---|---|
| S1 | Fiscal.ai local wrapper output | Company statistics, financials and third-party estimates | Retrieved July 26, 2026 | Overview, statements, annual/quarterly estimates and peer data; estimate-set timestamp not exposed | Medium-high | Core metrics, forecast distributions and peer valuation. Current market cap and EV were independently repriced. |
| S2 | AAOI Q1 2026 Form 10-Q | SEC filing | Filed May 7, 2026 | Quarter ended March 31, 2026 | High | End-market revenue, cash flow, capex, customer concentration, Amazon warrant, debt and leases. |
| S3 | AAOI Q1 2026 results, Exhibit 99.1 | SEC-filed company release | Filed May 7, 2026 | Q1 results and Q2 outlook | High | Reported and non-GAAP results, Q2 guide and management capacity claims. |
| S4 | AAOI Q1 2026 Fiscal.ai event package | Transcript + report package | Event May 7; retrieved July 26, 2026 | Full transcript and seven-page report PDF checked; no presentation slides in package | Medium-high | Management claims, capacity cadence, 2026 outlook and capital-spending commentary. |
| S5 | AAOI 2025 Form 10-K | SEC filing | Filed February 26, 2026 | Fiscal year 2025 | High | Business model, end-market mix, customer concentration, manufacturing footprint and purchase-order terms. |
| S6 | May 2026 ATM prospectus supplement | SEC filing | Filed May 14, 2026 | $600M ATM and May 13 share count | High | Basic share count, awards, ATM terms and dilution risk. |
| S7 | Yahoo Finance historical price | Market data | Through July 24, 2026 close | AAOI and peer closes and AAOI price history | Medium | Spot price, June-to-July drawdown and market-cap repricing. |
| S8 | April Pearland property purchase 8-K | SEC filing | Filed April 8, 2026 | $58.4M property acquisition | High | Post-quarter committed cash use. |
| S9 | May Houston leases 8-K | SEC filing | Filed May 13, 2026 | Three 123-month leases totaling 736,216 square feet | High | Post-quarter rent commitments and footprint expansion. |
| S10 | June China credit agreement 8-K | SEC filing | Filed June 16, 2026 | RMB500M revocable facility | High | Funding capacity; not treated as debt until drawn. |
| S11 | July cleanroom design-build 8-K | SEC filing | Filed July 1, 2026 | $94.1M contract; January 10, 2027 target completion | High | Manufacturing commitment, schedule and buildout risk. |
| S12 | Pearland construction-start announcement | Company release | July 14, 2026 | Construction execution milestone | Medium | Confirms work began; does not independently verify yields or saleable capacity. |
| S13 | Texas Semiconductor Innovation Fund award | Government release | April 29, 2026 | $20.85M grant and >$279M planned investment | High | Public incentive and investment scale; specifically a Texas, not federal, award. |
| S14 | Q2 earnings-date announcement | Company IR release | July 16, 2026 | August 6, 2026 results call | High | Next hard catalyst. |
| S15 | 1.6T volume-order announcement | Company release | March 9, 2026 | >$200M order with expected Q3-Q4 shipments | Medium | Demand evidence, subject to filed purchase-order cancellation/rescheduling risk. |
| S15b | Upsized 800G order announcement | Company release | April 2, 2026 | $71M order; $124M total disclosed orders from that customer since mid-March | Medium | Demand evidence and initial 10,000-unit shipment claim. |
| S16 | AAOI silicon-photonics roadmap | Company technical article | April 15, 2026 | 400G/800G/1.6T platform and automation claims | Medium | Company-claimed architecture and manufacturing advantages. |
| S17 | Coherent OFC 2026 product announcement | Peer company release | March 17, 2026 | 1.6T products across SiPh, InP EML, CW laser and VCSEL | Medium-high | Competitive technology and scale context. |
| S17b | Lumentum Q3 FY2026 results | Peer company release | May 5, 2026 | $808.4M revenue and 47.9% non-GAAP gross margin | Medium-high | Optics growth and margin benchmark. |
| S17c | Fabrinet Q3 FY2026 Form 10-Q | SEC filing | Filed May 5, 2026 | Contract-manufacturing economics | High | Manufacturing-execution benchmark. |
| S18 | Nasdaq short-interest API | Market data | July 15, 2026 settlement | 9.94M shares short; 1.14 days to cover | Medium-high | Positioning context; not part of fundamental valuation. |
| S19 | CNY/USD history | Market data | July 24, 2026 | 1 CNY = $0.1477 used for RMB facility translation | Medium | Currency conversion only. |
| S20 | AAOI HFC networking portfolio | Company product pages | Accessed July 26, 2026 | Quantum18/12, QuantumWave and QuantumLink | Medium | CATV product and software functionality; no recurring-revenue economics disclosed. |
Underwriting status: preliminary watchlist initiation; not ownership-ready. Evidence confidence: medium. Market data through the July 24, 2026 close; primary fundamentals through Q1 2026. United States dollar reporting throughout; the only currency conversion is the RMB500 million credit facility, translated at 1 CNY = $0.1477.