The product carrying the entire AI thesis generated less than $4 million last quarter while the CEO forecast $100 million in orders arriving within three months.
Thesis: AAOI owns a genuine bottleneck in in-house laser fabrication during a global laser shortage, with external suppliers quoting one-year-plus lead times. That is real leverage. But the AI revenue driving the multiple is under $4 million per quarter right now, the entire Q2 ramp hinges on firmware completing in March after already slipping once, and the $1B 2026 target requires tripling revenue off a product line that is not yet shipping at scale. The capacity build ($209M capex vs $120-150M guide) is real capital conviction, but it is being funded while operating losses widen and share count jumps 9% in one quarter.
Verdict: HOLD — Conviction: MEDIUM
Catalyst: Firmware completion in March unlocks Q2 800G volume ramp. Management guided Q2 800G at $25M+ with customer demand at $35-40M. Second hyperscaler 800G orders expected within weeks per CEO.
Key Risk: Firmware slips again or Q2 800G misses the $25M floor, invalidating the entire 2026 ramp thesis. Capacity expansion of 5.5x (90K to 500K units/month) with clean room construction, equipment lead times, and qualification gates that have historically slipped for this company.
The Tell: CEO said: 'we should receive more than $100 million of 800G transceiver within a few months, maybe one month, two months, for sure less than three months, we should receive more than $200 million of 1.6T transceiver.' CFO Murry immediately interjected: 'Transceiver orders. We're not buying the transceivers. We're receiving the orders.' The CEO was presenting expected orders as near-certain revenue events. The CFO's real-time correction exposed the gap between forecast and commitment.
Friction Level: HIGH_FRICTION — Bull sees hyperscaler loading forecasts through 2027 as demand locked in. Bear sees forecasts with zero signed orders, CFO correcting CEO on whether they are commitments. The call itself contained the disagreement in real time.
2025Q4 · Feb 26, 2026
No majority across 5 readings2 LONG · 1 AVOID · 1 HOLD · 1 SHORTRead the full call
800G Revenue Under $4M, CEO Projects $100M in Orders Coming
AAOI owns a genuine bottleneck in in-house laser fabrication during a global laser shortage, with external suppliers quoting one-year-plus lead times. That is real leverage. But the AI revenue driving the multiple is under $4 million per quarter right now, the entire Q2 ramp hinges on firmware completing in March after already slipping once, and the $1B 2026 target requires tripling revenue off a product line that is not yet shipping at scale. The capacity build ($209M capex vs $120-150M guide) is real capital conviction, but it is being funded while operating losses widen and share count jumps 9% in one quarter.
CEO said: 'we should receive more than $100 million of 800G transceiver within a few months, maybe one month, two months, for sure less than three months, we should receive more than $200 million of 1.6T transceiver.' CFO Murry immediately interjected: 'Transceiver orders. We're not buying the transceivers. We're receiving the orders.' The CEO was presenting expected orders as near-certain revenue events. The CFO's real-time correction exposed the gap between forecast and commitment.
CEO stated 'we should receive more than $100 million of 800G transceiver within a few months' and 'at least two or three' customers will order. CFO Stefan Murry had to clarify: 'Transceiver orders. We're not buying the transceivers. We're receiving the orders.' Forecasts are being presented as backlog. 800G actual revenue was under $4M in Q4.
Management cites 'loading forecasts' from two to three hyperscalers through 2027. CEO explicitly stated: 'They only give us the loading forecast, but to make sure we guarantee what we promise, there's some kind we need to allow agreement.' No signed 800G orders beyond the fourth volume order from one customer, and that one is firmware-blocked.
CapEx came in at $209M versus $120-150M guided, 39% over the high end. $300M committed to tripling Texas laser production. This is real money deployed against a stated bottleneck. Cash rose to $216M from $150.7M.
Sugar Land facility groundbreaking weeks ago, Texas laser production set to triple, in-house indium phosphide capability during a global laser shortage with external suppliers quoting one-year-plus lead times. Capacity is the stated constraint on the $1B target, and management is spending to remove it.
800G revenue was under $4M in Q4. Data center growth of 69% YoY was driven by 400G (+141% YoY) and 100G legacy products. The AI-labeled product line contributing to the narrative generated almost nothing last quarter.
CEO projected $378M monthly transceiver revenue by mid-2027, a 9x revenue growth path, while confusing 2026 and 2027 timelines on the call. CFO interjected 'Not this year' and CEO responded 'No misleading.' Absolute claims: 'minimum 99% confident,' 'demand is much, much bigger.'
+62
CONFIDENT"CEO used absolute language: 'minimum 99% confident,' 'demand is much, much bigger,' 'we should receive more than $100 million of 800G.' Voluntary specificity on order timelines. CFO repeatedly interjected to temper claims: 'Not this year,' 'No misleading,' clarifying forecasts are not orders in hand."
Rich transcript with specific capacity numbers, segment revenue breakdowns, and customer concentration data. Management gave hard targets and the CFO corrected real-time. Gap between signed orders and forecasts is explicitly visible in the exchange.
Firmware completion in March unlocks Q2 800G volume ramp. Management guided Q2 800G at $25M+ with customer demand at $35-40M. Second hyperscaler 800G orders expected within weeks per CEO.
Firmware slips again or Q2 800G misses the $25M floor, invalidating the entire 2026 ramp thesis. Capacity expansion of 5.5x (90K to 500K units/month) with clean room construction, equipment lead times, and qualification gates that have historically slipped for this company.
“Our total revenue was a record $134.3 million, which increased 34% year-over-year and increased 13% sequentially off a strong Q3 and was in line with our guidance range of $125 million-$140 million.”
“We expect Q1 revenue to be between $150 million and $165 million, accounting for a sequential increase in CATV revenue, as well as a sequential increase in our data center revenue.”
Figures are extracted from the transcripts and checked in code: every quote appears word for word. Descriptive only: none of this feeds the verdict.
Revenue hit $134.3M with GM at 31.4% beating the high end. Data center surged 70% QoQ to $74.9M with 400G up 141% YoY. Q1 guide of $150-165M is a 12-23% sequential jump with zero 800G contribution from the March firmware. The capacity build is real: $209M capex against $120-150M guide, Sugar Land groundbreaking, Texas laser tripling. In-house laser fabrication during a global shortage with one-year-plus external lead times is a genuine bottleneck position. Three hyperscale customers are engaged with two loading forecasts through 2027. CATV at $245M nearly tripled with a credible $300M run rate. Cash at $216M funds the build. If March firmware lands and Q2 800G hits $25M+, the stock reprices.
800G revenue was under $4M in Q4. The CEO admitted it came in 'a lot below' the $4-8M guide. The entire AI thesis rests on a product that is firmware-blocked with no signed orders beyond one customer. Share count jumped from 70.3M to 76.4M, a 9% dilution in one quarter. Operating losses widened to $7.1M from $2.5M YoY. GM guided DOWN to 29-31% for Q1. CapEx ran 39% over guide while opex rose to 37% of revenue. Three customers represent 91% of revenue. The CEO confused 2026 and 2027 targets and the CFO had to correct him live. The $1B target requires tripling revenue off a sub-$4M product line with signed orders still pending. Management is selling forecasts as commitments.
Say/Do Gap
CEO presented $100M+ 800G and $200M+ 1.6T orders as arriving 'within months.' CFO corrected live that these are expected orders, not in hand. Actual 800G revenue was under $4M. The ramp from under $4M to $25M+ in one quarter depends entirely on firmware that already slipped once.
AI Label Without Substance
Data center growth of 69% YoY was driven by 400G and 100G legacy products. The AI product (800G) contributed under $4M. Management's own breakdown shows 51% of data center revenue from 100G products, 41% from 200G/400G, and only 8% from 10G/40G. The AI narrative is not yet supported by the AI product line.
Unsustainable Trend Confidence
CEO used 'minimum 99% confident,' 'demand is much, much bigger,' confused 2026 and 2027 on air, and CFO had to interject 'Not this year' and 'No misleading.' Absolute forward claims from a company whose key product slipped a quarter and whose orders are unsigned.
Backlog Quality
Management explicitly cited 'loading forecasts' rather than orders. CEO stated customers 'only give us the loading forecast' and agreements are still needed. The $1B 2026 target is built on forecasts from customers with no contractual commitment behind them.
Capacity Ceiling
Management stated revenue is 'limited by our production capacity and supply chain, not market demand.' Expanding from 90K to 500K units/month is a 5.5x jump with clean room construction, equipment lead times, and qualification gates. Historical precedent for this company includes slippage.
HOLD. The asymmetry runs through firmware completion in March. If 800G hits $25M+ in Q2, the capacity build and hyperscaler engagement converge and this reprices higher. If it slips again, the Say/Do Gap becomes a credibility event with 9% dilution already baked in. I cannot size the risk until I see signed orders and a working product. The in-house laser bottleneck position is real and worth owning, but not at a price where the AI narrative is doing all the work on a sub-$4M revenue line. Wait for Q1 close and the March firmware milestone. No edge yet.