Over the past seven days, the account of well-known crypto and equities investor Serenity has shed 49.4% of its peak value. The portfolio, which once returned 4502% in 2026, now sits on a brutal correction. Yet Serenity remains publicly optimistic, calling the drawdown "noise" and doubling down on his thesis that a revenue inflection point in AI infrastructure will materialize in the second half of 2027.
The core of Serenity's bet rests on three small-cap stocks: AXTI (AXT Inc.), SIVE (Sivers Semiconductors), and AAOI (Applied Optoelectronics). All three occupy what he calls "bottleneck and restricted areas" of the AI supply chain—optical interconnects and specialty semiconductor substrates. From the noise of 2017 to the signal of today, the playbook has shifted from chasing hot Layer-1s to mining the picks-and-shovels of the AI revolution. Serenity is not alone, but his conviction stands out.
Context: Why Now? Serenity's account drawdown comes amid a broader sell-off in AI infrastructure small-cap names. The market is pricing in skepticism about whether the 2027 H2 revenue inflection point will actually arrive. His three holdings—AXTI (phosphide and gallium arsenide substrates), SIVE (silicon photonics chips), and AAOI (high-speed optical modules)—are all leveraged plays on the upgrade cycle from 800G to 1.6T optical transceivers. According to industry forecasts, 1.6T modules will begin small-volume shipments in late 2026 and ramp in 2027. Serenity has aligned his timeline with that consensus, but the stock market is a discounting machine, and any delay in orders can trigger violent re-ratings.
Core: The Technical and Market Reality Speed runs require foresight, not just reaction. Serenity's thesis is built on observable data: the AI cluster bottleneck is moving from GPU compute to interconnects. As training clusters expand to 100,000 GPUs, the cost and power consumption of optical transceivers become the limiting factor. AAOI's vertical integration (in-house laser chip production) gives it a cost edge over Chinese rivals like Zhongji Innolight. AXTI dominates the InP substrate market, serving as the material foundation for those lasers. SIVE's silicon photonics, though nascent, promises lower latency and power for intra-cluster links.
However, the ledger does not lie, but it rewards patience. A deeper dive into the fundamentals reveals risks that Serenity's optimistic public statements may downplay. AAOI's gross margins are under pressure from price competition. AXTI faces capacity expansion challenges and long customer qualification cycles. SIVE's AI revenue is almost nonexistent—its current income comes from defense contracts. The 2027 H2 breakout requires all three companies to execute flawlessly. Based on my audit experience in AI supply chains, such synchronized success is rare. A single miss—like AAOI losing a major customer to Coherent—can turn the inflection point into a disappointment.
Contrarian Angle: The Unspoken Risks The narrative of 'bottleneck investing' is alluring, but it obscures a critical blind spot: the possibility of technological disruption. Co-packaged optics (CPO) could render traditional pluggable modules obsolete, threatening AAOI's entire product line. SIVE's photonics technology could be leapfrogged by Intel's integrated photonics. Meanwhile, the market's skepticism is not just noise—it reflects a rational reassessment of timing. Serenity's 4502% return in 2026 was partly due to buying at multi-year lows during the 2025 crypto winter. His cost basis is so low that a 49% drawdown still leaves him with massive gains. This is not a strategy a retail crypto investor can replicate without deep pockets and extreme risk tolerance.
Furthermore, the analysis of the original source material reveals a hidden pessimism: the article only cites Serenity's optimism, without providing concrete evidence such as customer orders or guidance from the companies. The 2027 H2 inflection point is a widely held expectation, but it is already priced into the stocks. If actual revenue falls even slightly short, the multiple compression could be severe. Serenity's ability to "handle higher volatility" is a luxury of his past gains, not a signal that the thesis is sound.
Takeaway: What to Watch For crypto investors eyeing similar plays in the AI-blockchain convergence—think DePIN projects like Render Network or Filecoin—this case holds a stark lesson: hardware plays are subject to brutal cycles. The inflection point may arrive in 2027 H2, or it may slip to 2028. When it does, the winners will be those who positioned early and survived the drawdown. But for most, the speed of capital can be merciless. The real question is not whether Serenity is right, but whether the market's patience will outlast the technology's development curve.
Speed runs require foresight, not just reaction. From the noise of 2017 to the signal of today, the ledger does not lie, but it rewards patience.