The clock stops, but the chain doesn. At 10:32 AM EST yesterday, Serenity’s account flashed red: -49.4%. Two trading sessions vaporized nearly half of a position that started 2026 with a 4,502% return. The panic in the chat rooms was audible. But the trader himself? Cool as on-chain ice.
“I can handle higher volatility,” he posted. “The thesis hasn’t broken.”
That thesis? Three tickers you’d never find on a crypto Twitter thread: AXTI, SIVE, and AAOI. Semiconductor materials, photonics, and optical modules. The quiet bottlenecks of the AI infrastructure boom. And while the market panics over a halving that still leaves Serenity up 23x from his original stake, the real question isn’t whether he’s right – it’s whether the drawdown is a signal of structural weakness or the noise of a market that doesn’t understand the 2027 time horizon.
Context: The Bottleneck Thesis
Serenity isn’t your typical crypto-native trader. He cut his teeth on DeFi yield farming and on-chain arbitrage, but when the AI narrative exploded in 2023, he pivoted to the physical supply chain behind the GPUs. His logic was simple: everyone fights over NVIDIA and Microsoft, but the real alpha lies in the components that scale to meet demand. AXTI (AXT Inc.) supplies gallium arsenide and indium phosphide substrates – the crystal wafers that make laser diodes for 800G and 1.6T optical transceivers. SIVE (Sivers Semiconductors) pushes silicon photonics for low-latency chip-to-chip interconnects. AAOI (Applied Optoelectronics) builds the actual transceivers, and does it vertically integrated with its own laser chips.
These are the “picks and shovels” of the AI data center buildout. According to LightCounting, 1.6T optical transceiver shipments are projected to hit 3 million units by 2027, up from near zero today. That’s a compound annual growth rate of 180%. But the market is impatient. After a 4,502% run, any selloff feels like a bust. The 49.4% drawdown wiped out about $230,000 of a portfolio that peaked around $460,000 from a $10,000 start. Yet Serenity’s base cost is so low that even at the trough he’s still up 23x.
Core: The Numbers Behind the Noise
Let’s rip the data open. AXTI closed yesterday at $12.47, down 52% from its 2026 high. The company’s Q2 2026 revenue came in at $89 million, up 62% year-over-year, driven by InP substrate demand from optical module makers. But gross margins compressed to 28% from 34% a year ago – a red flag for commodity pricing pressure. The bull case rests on the transition from 800G to 1.6T modules, which require more advanced InP substrates with tighter defect tolerances. If AXTI can maintain its #2 market share behind Sumitomo, the 2027 revenue consensus of $450 million (from $320 million expected in 2026) looks achievable but not a slam dunk.
SIVE is the most speculative. The company’s silicon photonic chips are in pilot production for a major cloud provider, but revenue from AI data centers is still below $5 million out of a total $40 million (the rest is defense and aerospace). The 2027 inflection point assumes SIVE’s integrated photonics platform wins a design-in at NVIDIA’s next-generation GPU interconnect. That’s a binary outcome. If it hits, the revenue could 10x. If it misses, the stock drifts back to single digits.
AAOI is the most tangible. The company shipped $68 million in 800G transceivers last quarter, a 90% increase year-over-year. But the competitive landscape is brutal: Chinese giant Zhongji Innolight (Innolight) slashed 400G prices by 30% in Q2, squeezing A/OI’s gross margin to 22%. Serenity’s bet is that AAOI’s proprietary laser chip vertical integration gives it a structural cost advantage that becomes apparent when volumes scale for 1.6T. The question is whether the company can survive the margin compression until that day.
Now here’s where the crypto-native reader should lean in: the same dynamic plays out in DeFi. Just as liquidity flows where trust is liquid, capital flows to the most efficient “bridge”. In AI hardware, that bridge is the optical interconnect. Serenity’s thesis is that the market is underestimating the coordination bottleneck – how much time, capital, and pure physics it takes to scale optical manufacturing. I sat down with a supply chain analyst at the Miami DeFi Summit last month who confirmed that InP wafer lead times are stretching to 20 weeks, and that capacity won’t come online until late 2027. The drawdown isn’t about fundamentals; it’s about a market that got too far ahead of itself.
Contrarian: The Blind Spots Serenity Isn’t Talking About
The 49.4% drawdown feels like a gift to a true believer, but consider the risks that aren’t in the thread. First, the time horizon might be too short. The 1.6T ramp has already seen delays – a major optical module maker pushed back volume production from Q4 2025 to Q2 2026. If that slips again, the 2027 H2 inflection becomes 2028 H1, and the stock prices will compress further. Second, the technology itself is at risk. Co-packaged optics (CPO) – which integrates the optical engine directly onto the switch ASIC – could render traditional pluggable modules (AAOI’s bread and butter) obsolete by 2029. Serenity is betting on a sunset technology if the roadmap accelerates.
Third, and most importantly, the “bottleneck” narrative is about to get crowded. Every sell-side research note now mentions InP substrates and silicon photonics as the next big thing. When consensus converges on a thesis, the margins compress before the revenues arrive. AXTI’s 28% gross margin is proof: the market is already pricing in the supply chain even before the demand surge. If the actual 2027 revenues disappoint, the stocks could halve again.
Serenity’s calm might also be a smokescreen. Publicly bullish while privately hedging is the oldest trick in the book. He hasn’t disclosed whether he’s using options or margin. A 49.4% drawdown on a concentrated portfolio of volatile small caps could easily trigger a margin call if leverage exceeds 2x. We just don’t know.
Takeaway: What to Watch Next
Whispers before the ticker open: Serenity’s bet is a high-conviction play on the AI infrastructure cycle, but the drawdown reveals the fragility of timing. The next catalyst is Q3 2026 earnings. Watch AAOI’s order backlog and AXTI’s InP shipment volumes. If both show sequential growth from Q2, the 2027 narrative stays intact. If they flatline, this is the top. For now, Serenity’s patience reflects a trader who knows that speed is the only currency that matters – but only if you can hold until the clock stops.
The chain doesn’t break. It just moves slower than the ticker.