Skip the "Magnificent Seven": 2 Underappreciated Memory Stocks With More Room to Run

SK Hynix reported third-quarter revenue up 328% to $78.9 billion and operating profit up 632% to $62.3 billion; Micron’s fiscal fourth-quarter revenue rose 380% year over year to $54.2 billion, with non-GAAP EPS up about 1,000% to $33.42. Micron guided to fiscal 2027 Q1 sales of $61.5 billion at the midpoint, while both companies said memory demand exceeds supply; their shares are down about 20% over the past month. Bank of America forecasts data-center spending will more than double to $2.1 trillion by 2028, and the article argues the stocks look inexpensive at P/E ratios of 11 for SK Hynix and 13 for Micron, versus a tech-sector average of 24.

Analysis

The investable question is not whether AI needs memory, but whether constrained HBM supply can sustain pricing and returns after capacity catches up. A shift toward HBM can lift mix and bargaining power for leading suppliers, while also diverting investment from conventional DRAM and raising input costs for server makers and other memory buyers. The risk is that hyperscaler capex pauses or new capacity arrives just as customers optimize memory use; long-dated capacity agreements only reduce that risk if volumes and economics are binding, not merely forecasts. Treat the quoted P/Es cautiously: peak-cycle earnings can make memory stocks look optically cheap before margins normalize. More importantly, the article's reported revenue levels and growth rates warrant reconciliation against company filings before trading; the periods and scales are not sufficiently reliable here to anchor estimates. Near term, the recent selloff may create an entry window, but the 1–3 month test is guidance, HBM shipments/pricing and customer capex commentary. Over 6–18 months, supply discipline and returns on new capacity matter more than headline AI-spending forecasts. The bullish thesis fails if memory pricing rolls over, inventory rises, or hyperscalers defer projects.

Key Decisions for Investors

• Do not underwrite the article's earnings growth or valuation claims until the figures are reconciled to SK hynix and Micron filings, including currency, reporting period and GAAP/non-GAAP scope.
• If verified results and guidance remain intact, consider a staged long in MU and SKHY rather than buying the full position after a sharp drawdown; size as a cyclical trade, not a secular-growth multiple. Reassess on the next earnings updates and HBM pricing/supply disclosures.
• Track hyperscaler capex commentary from AMZN, GOOG, META and MSFT alongside memory inventory and contract pricing. Broad capex deferrals or sequential pricing deterioration would invalidate the demand-led thesis and argue for reducing exposure.
• Watch Samsung Electronics and memory-equipment capacity plans as competitive/supply signals; expanding supply or weaker HBM yields could pressure returns even if aggregate data-center demand keeps growing.