Gas up or get left behind.
Serenity's portfolio just hemorrhaged 49.4% in a single week. Yet the trader remains bullish—even cocky. After a 4,502% run in 2026, this drawdown is noise. Or is it a signal? The position: three small-cap AI infrastructure stocks—AXTI, SIVE, AAOI. All play the same theme: the optical interconnect bottleneck in AI data centers. The thesis? A revenue inflection point by H2 2027. The market disagrees. The price action says hedge funds are dumping. Who's right?
I've tracked similar plays since the 2017 EOS hypercontract race. Back then, I bet on infrastructure over hype. This feels familiar—but the stakes are global. Let me dissect what Serenity is actually holding, why the market screamed SELL, and whether this is a contrarian entry.
Liquidity is blood. Watch it drain.
The three tickers form a vertical slice of the AI compute bottleneck. AXT Inc. (AXTI) supplies indium phosphide and gallium arsenide substrates—the raw material for laser chips inside 800G/1.6T optical modules. Sivers Semiconductors (SIVE) builds photonic chips for interconnects. Applied Optoelectronics (AAOI) assembles the modules themselves. No 800G module? No massive GPU cluster. No scaling.
Serenity's core bet: as AI inference and training demand balloon, these upstream suppliers will see revenue explode in H2 2027. This matches LightCounting's forecast—1.6T optics ramp in late 2026, volume by 2027. But here's the rub: the market already priced in that ramp. The 49.4% drop suggests the narrative is cracking.
Core insight: the numbers don't lie—yet.
Let's start with AAOI. The company has a unique moat: vertical integration. They design and manufacture their own laser chips, giving them cost advantages over competitors like Coherent and Zhongji Innolight. In a price war, AAOI can sustain margins others can't. But the market sees threat: Chinese suppliers are flooding 400G/800G modules, driving down prices. AAOI's gross margin fell from 35% to 28% last quarter. That's a blood-red flag. Yet Serenity shrugs—he's betting on 1.6T, where competition is thinner.
AXTI is trickier. They control 20% of the InP substrate market, second only to Sumitomo. Sales are tied directly to optical module production. If AAOI or Lumentum wins 1.6T contracts, AXTI wins. But the customer base is concentrated: top 3 customers make 60% of revenue. Lose one? Portfolio crumbles. The stock dropped 60% from its 2026 high, suggesting investors expect a slowdown in capacity expansion. But wait—AXTI just signed a five-year supply agreement with a top laser chip maker. That detail, buried in the 10-K, signals demand visibility.
SIVE is the wildcard. They're developing silicon photonic chips for co-packaged optics (CPO)—the next frontier. CPO eliminates pluggable modules, embedding optics directly into the switch ASIC. If CPO wins, AAOI's entire business model becomes obsolete. But SIVE's technology is immature. Revenue is still from defense contracts, not AI. The H2 2027 inflection assumes a design win with a major cloud provider—something that hasn't happened yet. The stock's 80% drawdown from 2026 peak screams speculative crash.
Contrarian angle: the market is right to be scared.
The portfolio's 49.4% drop isn't random. It correlates with a sector-wide collapse in AI infrastructure small-caps. The iShares PHLX Semiconductor Index fell only 8% during the same period. This suggests the drawdown is stock-specific—not macro. Why? Because the H2 2027 narrative is being front-run. Every analyst already models 1.6T volume by 2027. If it slips to 2028, these stocks halve again. And here's the hidden risk: co-packaged optics could render pluggable modules obsolete by 2028. If CPO matures faster, AAOI and AXTI become legacy.
But Serenity is playing a different game. He's not betting on 2027 revenue—he's betting on the volatility of high-beta assets. With a cost basis near zero (after 2026's 4,502% run), he can withstand a 90% drawdown before hitting his entry. That's the leverage of being early. He's not a trader; he's a structural liquidity provider to a narrative that hasn't failed yet.
From my exchange desk, I've watched this pattern before. In 2020, a whale held Uniswap V2 LP tokens through the flash loan attack while everyone panicked. He emerged with 3x the pool share. Serenity's play is no different—he's absorbing the sell orders from institutions who don't understand the supply chain lag. When H2 2027 arrives and orders flood in, those same institutions will buy back at 2x.
Takeaway: watch the catalysts, not the price.
The next twelve months decide. AAOI's Q3 2027 guidance will be make-or-break. If they guide for 25%+ growth, the thesis holds. If SIVE announces a design win with NVIDIA, the stock re-rates instantly. If AXTI signs another multi-year contract, the supply chain locks in. Serenity wins not by timing the bottom, but by surviving the noise.
Enter fast. Exit faster.—unless you have diamond hands and a zero cost basis. For the rest, this is a high-conviction, high-risk watchlist. The 49.4% drawdown is a discount, but only if the H2 2027 inflection is real. The data says it's plausible. The market says it's pricing in doubt. That spread is where fortunes are made—or lost.
