AAOI: Down 37% From Its Peak. Opportunity or Trap?

The Laser Self-Supply Advantage, CPO as the Proving Ground, and Profit That Still Isn’t Proven

AAOI is up more than 550% over the past year. It peaked at $233.67 on May 13, and it has since fallen back to around $147. That’s a 37% drop from the high. Look only at the one-year chart and you see a stock that exploded. Look only at the last month and you see a stock that’s caving in. The same name shows two opposite pictures at the same time.

So the question for anyone looking at this stock right now comes down to one thing. Is this drop a pullback that cooled off an overheated move, making now the place to buy? Or is it the early stage of a bubble starting to deflate, a place to avoid? This is a company posting record revenue for four straight quarters and still losing money, with a GAAP gross margin of 29.1%, trading at 21x sales. Can you buy it just because it’s fallen 37% from its high?

This article exists to answer that question.

I’ll work through what AAOI actually sells (not transceivers), which customers that demand reaches, what in the backlog is real revenue, and what kind of future is already priced into today’s level. Then at the end I’ll lay out my own investment view on this name.

Contents

1. What AAOI actually makes
2. From 800G to 1.6T, and the laser bottleneck
3. The real value and limits of vertical integration: from margins to the CPO transition
4. Two pillars of demand: hyperscaler orders and the AMD possibility
5. The quality of the backlog: what’s revenue and what’s intent
6. The risk isn’t a single layer
7. Valuation and my call

What AAOI actually makes

Most optical module makers buy lasers from outside and assemble them into modules. AAOI makes the laser chips itself. It buys InP (indium phosphide) substrates, grows the laser active layer directly on them, bundles those laser chips into subassemblies, and then completes them into transceiver modules. From chip to module, everything is vertically integrated inside one company. Among US-based optical communication companies, the ones with meaningful in-house laser production capacity can be counted on one hand.

Why that structure matters comes later. First, the revenue mix. AAOI has four markets: data center, CATV (cable TV), telecom, and FTTH. What’s driving the stock right now is data center. In Q1 2026, data center revenue was $81.4 million, up 154% year over year. The big driver was that 800G transceivers for AI data centers began shipping in volume to a large hyperscaler for the first time.

The other axis of revenue is CATV. Q1 CATV revenue was $66.8 million, the second-largest segment after data center. The market views CATV as a tired legacy business, but in my read this is the part that puts a floor under revenue when the data center ramp wobbles. Riding the DOCSIS 4.0 upgrade cycle, large cable operators like Spectrum and Mediacom are deploying AAOI’s QuantumLink and smart amplifier gear. Because of that, the whole company isn’t tied to AI optics alone.

That said, treating it as a cash-flow safety net comes with a caveat. As of Q1, a single cable distributor, Digicomm, accounted for 74.5% of accounts receivable, and the company gave this customer longer-than-usual payment terms. Revenue being stable and cash actually arriving on time are two different things.

The income statement is a different story. Q1 revenue was a record $151.1 million, but the company posted a GAAP net loss of $14.3 million. Gross margin was 29.1%. Adjusted EBITDA came in at positive $1 million, so it has climbed to roughly breakeven.

Gross margin is positive, so why the loss? A 29.1% gross margin is the line where you subtract only the direct cost of making the product from revenue. Sell a dollar of product and 29 cents is left after materials and manufacturing cost. The problem is the costs that come below that. R&D and SG&A, the operating expenses of running the company. In Q1, operating expenses ran up into the high 30s as a percentage of revenue. The remaining 29 cents can’t cover that, so there’s already a loss at the operating line, and once items like stock-based compensation are added on top, the final net loss prints at $14.3 million.

So where does this loss come from? Not from a rotting business, but from the cost of building out fast, outrunning revenue. Two things overlap. The first is product mix. Data center revenue surged 154% in Q1, and the engine of that, 800G, is in the very early stage of ramping production, so its cost is still high relative to price. The fast-growing side is the low-margin side, the classic early-ramp window. That’s why total margin sits at 29% even with steady CATV underneath. The second is capacity investment. The company is taking 800G and 1.6T production capacity from 100,000 units a month up to 650,000 by year-end and 930,000 by 2027, and Q1 investing cash outflow was $68.1 million, of which equipment and property spending was $58.2 million. The factory gets built first and revenue comes later, so depreciation and labor run ahead of revenue.

So this loss looks more like growing pains than structural rot.

Once 800G reaches volume scale and the capacity investment starts coming back as revenue, the structure is one where margins rise and the loss shrinks. But for that read to hold, future revenue growth has to actually translate into margin recovery and operating leverage. The company sees that inflection in the very next quarter. Q2 guidance is a range from a $2.5 million loss to a $2.8 million profit on a non-GAAP basis, right at the threshold of turning profitable.

Revenue is exploding while the company is still in the stretch where it doesn’t make money. Whether that loss is growing pains, and whether margins recover as the company plans, I look at separately later.

For now, the only point is that this gap is the first axis for judging AAOI.

From 800G to 1.6T, and the laser bottleneck

The optical transceiver market right now is in the middle of a generational shift. The data center workhorse is still 100G and 400G. In Q1 data center revenue, 100G was 41% and 200G plus 400G was 47%, while 800G, just starting volume production, was only 5.6%, or $4.6 million. The company said it would push 800G shipments in Q2 to nearly four times the Q1 level, but it also sees 400G demand staying strong over the same period. This isn’t 800G replacing the workhorse in one shot. It’s 800G climbing fast on top of a base that 400G still supports. On top of that sits 1.6T. The 1.6T transceiver already has its first volume order booked, with deliveries set to start as early as Q3 and wrap up by year-end.

The reason this transition favors AAOI lies in the light source. As transceiver generations go up, the difficulty of the laser that generates the signal rises sharply. The industry keeps signaling that upstream laser light-source supply is the bottleneck above 800G. High-performance lasers like the EML (electro-absorption modulated laser) can only be made in a limited number of places, and supply can’t keep up with surging demand. Module makers are lining up to get lasers.

AAOI doesn’t have to stand in that line. It pulls lasers from its own fab. In a transition where silicon photonics (SiPho) based transceivers are growing toward 40 to 45% share at 800G and around 60% at 1.6T, self-supplying a stable CW (continuous wave) laser is more than a cost advantage. It’s a supply advantage. AAOI can build when others can’t because they can’t source. For a hyperscaler, that becomes a way around the supply chain jam.

The company is pushing this advantage hard through capacity. It has a plan to grow laser fab capacity 350% by 2027, and it’s expanding its Sugar Land headquarters to 900,000 square feet. It also received a $20.85 million grant from the Texas semiconductor innovation fund. On the supply chain side, it cut the China-sourced portion of component value in its 800G and 1.6T products to under 10%, reducing tariff exposure. The direction is to grow US production capacity while shaving geopolitical risk at the same time.

This is the picture the market sees of AAOI.

A US optical company that self-supplies lasers, an 800G and 1.6T transceiver maker riding AI demand. The real call gets decided one layer down. The most important piece is this: when transceivers move to CPO, the module revenue AAOI eats today gets pulled into the switch chip, and the question is whether the laser self-supply strength survives even then. That’s where it splits between today’s price being just before compression or at the start of a bubble.