AAOI Fell 30% After a $600M Share Sale — the $98 Line Now Decides Bear Trap vs. Broken Chart

A year ago Applied Optoelectronics (AAOI) traded near its 52-week low of $18.50. By May it had peaked at $233.67. Last week it fell roughly 30% in five sessions, and the final leg of that fall was engineered by the company itself: a disclosure that it would sell up to $600 million more of its own stock. The shares now trade near $107, down about 14% from Friday's close of $124.82, and that puts the chart within one hard day's move of its 200-day moving average at roughly $98. Here is the level that confirms the setup and the level that breaks it — and the thing most of the commentary is getting backwards.

Start with what actually happened, because the two shocks are different animals. On August 18, management spoke at an industry conference and warned that near-term margins are under pressure from higher manufacturing costs, expedite fees, and the decline of the legacy high-margin 100G business; AAOI fell 15% that day inside a broader optical sell-off that began with Fabrinet's report. Then, after Friday's close, the company disclosed a new at-the-market equity offering of up to $600 million. That is the third large raise of 2026 — $500 million in April, $600 million in May, and now another $600 million — with the first two programs having drawn roughly $1.05 billion in exchange for nearly 7.8 million shares. Monday, shares plunged 17% to near $102 on elevated volume, breaking below the 20- and 50-day averages while holding above the 200-day, and retail sentiment on Stocktwits flipped from neutral to bearish.

The market is reading the offering as "management sells at the top." That framing is half right and half wrong, and the half that is wrong is the part that matters for the price.

Here is the half that is right. The capital is genuinely dilutive, the pattern is real, and the bear case does not stop at the share count. Q2 revenue was a record $191.9 million, up 86% year over year, but the company still lost $22.8 million on a GAAP basis, and the non-GAAP profit of $0.06 per share was helped along by foreign tax benefits and government subsidy income. Gross margin is running around 30%, against a long-term target of 40%. The company spent $565.5 million on capital expenditures in a single quarter — including $280 million in prepayments — to fund a capacity build-out from roughly 200,000 units a month today toward 650,000 by year-end 2026 and 930,000 by the end of 2027. That money comes from selling stock. And the concentration is extreme: ten customers make up 99% of revenue, one CATV customer is 42%, and two data-center customers add another 26% and 24%. At a market cap near $9.1 billion against a ~$1.1 billion full-year revenue target, this trades at roughly 8x forward sales with thin-margin economics. Nothing about that is cheap. The stock is not the demand story; it is a claim about margins and customer durability that has not been earned yet.

But here is the other half, and it is the half the technical level is actually testing. Management did not cut its demand number. Q3 revenue guidance was held at $255 million to $290 million through both the margin warning and the offering. At the midpoint, that is up roughly 130% year over year. The stated reason the margin picture is messy is not that orders are slowing; it is mix and cost — a temporary 100G slowdown tied to a customer's component issue that costs $20–25 million in sales, expedite fees to ship faster, and a deliberate trade of near-term margin for share in a market where the company says demand for 800G and 1.6T transceivers outruns its production capacity through at least mid-2027. "If we could produce more, we could ship more right now," management said on the Q2 call. A company that raises $600 million to fund the exact capacity that is gating its revenue is an ugly headline, but it is a different claim than a broken thesis. The market has fused the two: dilution arithmetic got treated as if it were demand destruction.

That distinction is what makes the $98 line tradable rather than random. The stock is down about 54% from its May peak after a move of more than 12x off the low, so the damage is real. But the 200-day moving average at roughly $98 is not just an average — it is the same zone where AAOI was trading about a month ago, before this month's surge, and Monday's low of $102 held just above it. That is a twice-tested demand zone at the very line that separates a correction within an uptrend from the end of one.

The rules are narrow, and the volatility means the test comes quickly. A daily close below $98 invalidates the setup: the 200-day gives way, the parabolic run's correction resumes, and the next platform is the June-July base below the current price, with little structural support in between. A reclaim of the 50-day, now near $128, confirms that the $102–$98 zone marked the shakeout low — the bear trap — and that the November earnings report, not this week's tape, gets to set the tone. In between, the stock is unresolved, and with a 14-day average true range near $16, either level can be reached within a session or two. Note the RSI: at roughly 43 after a 30% one-week fall, the flush is not yet washed out, which argues for letting the level decide rather than front-running it.

That is also why Nvidia's earnings Wednesday afternoon matter here. AAOI is not Nvidia, but the entire AI-optical complex has been trading as a single sentiment trade, and the market has spent the last week in a sell-the-news mood over exactly the kind of infrastructure spending and co-packaged optics doubts that will be center of Nvidia's call. A strong print is the kind of catalyst that can flip the sector clock into the "capitulation done" column. A weak one can supply the final flush that makes the low.

So the honest read, and I want to be direct about it: on the fundamentals, this is not a stock I'd buy at $107. The multiple prices in margin and customer outcomes that are unproven, the equity funding is visibly watering the per-share story, and GAAP profitability is still a promise. But the week's sell-off has now reached a level where the marginal seller's argument — dilution — is weaker than the buyer's evidence — a demand guide that never moved. That is precisely the kind of gap a technical test exists to adjudicate. Hold $98 on a real volume flush and the setup is a patient entry near a tested zone with a confirmed reclaim of $128 as the go-level; close below $98 and the chart says the correction has more room, full stop. Nvidia prints tomorrow, and the next few sessions will pick one. The line is $98 — I'd let the tape cross it before I decide the stock is done or down.