Herman Jin
Formerly in Asia FICC at Goldman Sachs, Macro & Semiconductor Investing
Applied Optoelectronics' surge was driven purely by American investors' faith, not Shanghai operators, whose money left long ago. It is a packaging company whose technology trails Zhongji Innolight's, and at $200 it never belonged on a buy list. @au_xbt notes SemiAnalysis has warned investors to exit at the top. x.com x.com x.comÂ
Applied Optoelectronics is set to harvest the capacity it invested in over prior years, and the market's post-earnings reaction looks too pessimistic. The key metric is how much capacity it adds, since in a shortage, companies criticized for capex take off once they have capacity to sell. The selloff mostly reflects how far the stock had already run. x.com x.comÂ
Applied Optoelectronics' heavy spending keeps the risk of more dilutive share sales alive. Its latest $600M raise took 5.69M shares at an average $105.36, versus 2.95M shares for $550M at $186.69 under the May program, lifting the share count from 38.9M in May 2024 to about 90.3M. Commitments keep growing: a $94.1M Houston cleanroom, more Houston leases, a $26.8M building, a Ningbo factory lease and a doubled Ningbo credit line, plus $335M of H1 capex. The next 10-Q will show what cash remains. … Show more
Applied Optoelectronics' repeated at-the-market share sales keep the stock capped, a reason to trim into strength. Per @aleabitoreddit, it finished a $500M ATM in April and $600M in June, then filed another $600M in August; that much selling pressure makes it hard to break out past $100, $150 or $200 despite an operationally bullish 2027. @aleabitoreddit expects strong outperformance only if the ATMs stop. x.comÂ
Applied Optoelectronics' heavy share issuance funds a capacity build into demand that exceeds supply, making risk-reward very attractive. It raised about $1.7B through three ATMs this year to expand plants in Houston and China and in-house InP lasers. Management expects demand to exceed capacity through mid-2027 even as monthly 800G/1.6T capacity rises from about 200,000 to 650,000 by year-end and over 930,000 by end-2027, and 1.6T sells for about twice 800G. Execution and dilution are risks. … Show more
Applied Optoelectronics is back in the junk bucket after dropping dilution news in Friday after-hours trading, a reminder of exactly who management is. Lumentum and Coherent are the calmer optical names to own instead. x.comÂ
Nvidia is more attractive than Applied Optoelectronics, which is up 200% year to date at 136x forward earnings with insiders selling, per TipRanks. Insider selling matters more for small and mid caps than for giants. Anyone long AAOI might wait for $50 or $60 before adding, while many stocks down 30-50% offer the highest upside when their moment comes. x.comÂ
Applied Optoelectronics has removed a major overhang by completing its $600M ATM offering, whose mechanical selling pressure likely made it lag the optical sector. It sold 5.69M shares at an average $105.36 for $588M net, bringing 2026 ATM proceeds to about $1.7B, roughly 85% of the $2B Nvidia invested in each of Coherent and Lumentum, as it scales production. x.com x.comÂ
Sivers' risk-reward looks worse than Applied Optoelectronics', which offers more growth, less risk and a cheaper multiple. Sivers guided to 25-30% CAGRs, implying only 8-10% conversion of its $1.2 billion pipeline; even a bullish 15-20% conversion gives $69-84 million of 2029 revenue, leaving it near 9.8x FY29 revenue. AAOI could grow about 95% a year with capacity expansion, or 80%-plus after heavily discounting that, yet trades below 2x 2029 modeled revenue. x.com x.comÂ
Applied Optoelectronics is way too risky to buy here, with risk-reward far from ideal. A failed breakdown last week on heavy buy volume and a lower wick off its all-time anchored VWAP give some fuel, but the logical stop sits 25% lower at last week's low. The largest dark-pool print on record at $223 sits overhead, and heavy volume there can mark a short-term climax; without momentum the stock can quickly become dead money. A sweep of recent lows to about $160 would be a far better entry. x.comÂ
Applied Optoelectronics' long-term logic remains intact despite a sharp pullback in optical module stocks. Q1 revenue grew more than 50% year over year, but after a big run this year expectations for AI data center optical links are high and the trade is crowded, so any shortfall or profit-taking brings violent swings, as after an earlier earnings report on guidance. It stays a long-term tracking position, not a heavy bet. x.comÂ
Semtech is a lower-risk way to play the photonics volume ramp, since it sells the chips inside transceivers that makers like AAOI build. Q2 upside came from 1.6T qualifying early, and 800G TIA share rose from ~18% to over 50%, with >50% expected at 1.6T by January. Content per transceiver jumps from high single digits to $80-90 at 3.2T, which the market hasn't modelled. The CEO says secured capacity may not be enough for FY28, pricing holds, and gross margin starts near 64%. x.com x.comÂ
