September Effect: 2 Memory Stocks to Buy on a Potential Market Dip

The article flags a potential September-October equity pullback in a midterm-election year, noting the S&P 500 fell at least 5% in 15 of the past 24 such cycles. It identifies Micron and Sandisk as potential dip-buy candidates, citing forward P/E multiples of roughly 6x and below 6x fiscal-2028 estimates, respectively. The bullish thesis rests on AI-driven HBM, DRAM and NAND demand, constrained memory supply, and multiyear customer contracts; Sandisk targets mid- to high-teens revenue growth and approximately 80% adjusted gross margin from fiscal 2028 to 2030.

Analysis

The actionable read-through is not simply "AI demand": HBM capacity allocation can tighten conventional DRAM because advanced-node wafer starts, packaging capacity, and engineering resources are shared. MU has greater operating leverage to that spillover than NVDA, whose valuation already embeds sustained accelerator scarcity; each incremental DRAM/NAND price increase should flow disproportionately to MU gross margin and FCF while utilization remains high. The key near-term risk is that a broad risk-off tape compresses the multiple before estimates move, creating a better entry point rather than invalidating the earnings setup.

SNDK is the higher-beta expression of NAND discipline, but its stated long-range margin framework should be treated as an aspiration, not an underwriting input. NAND is more exposed than DRAM to hyperscaler digestion, SSD inventory corrections, and renewed price competition from Samsung, SK Hynix, and Kioxia; multiyear contracts may reduce volume volatility but do not necessarily eliminate price-reset risk. Before sizing SNDK, confirm its net leverage, contract pricing structure, customer concentration, and whether consensus estimates already capitalize a peak-cycle margin.

Seasonal election-year weakness is a poor standalone timing signal: its value is as a liquidity event for entering a supply-tightness thesis. Over the next 1-3 months, monthly contract-price data, hyperscaler capex commentary, and MU's next gross-margin/bit-shipment guide matter more than calendar effects. Over 6-18 months, the thesis fails if suppliers convert HBM investment into broad DRAM/NAND capacity faster than demand absorbs it, or if AI server deployment shifts toward lower-memory architectures.

Contrarianly, the market may be underestimating the duration of conventional-memory tightness, but it may also be overvaluing the durability of peak margins. Long-term supply forecasts several years out have limited predictive value in memory: a single capex reversal or demand pause can rapidly change spot pricing and equity multiples.

Key Decisions for Investors

• Buy MU only on a 10-15% market-driven pullback or after a post-earnings guide reset, using a 6-12 month horizon. Prefer equity over calls given cycle timing uncertainty; target 20-30% upside if DRAM pricing and gross-margin guidance continue rising. Exit if MU guides to declining DRAM pricing or materially weaker bit shipments for two consecutive quarters.
• Use a relative-value expression: long MU / short SOXX in equal dollar amounts for 3-6 months. This isolates memory-price operating leverage from a broad AI semiconductor derating; cover the short if MU underperforms SOXX by 10% after an earnings guide that confirms stable pricing.
• Keep SNDK on watch rather than establish a full position until leverage, customer concentration, and contract economics are verified. If those checks are favorable, initiate at half-size following a NAND-price pullback, with a smaller risk budget than MU because normalized margins can compress faster in a downturn.
• Monitor quarterly commentary from NVDA, Microsoft, Amazon, Alphabet, and Meta for AI-capex digestion. A synchronized reduction in accelerator or server-deployment guidance is the earliest cross-sector signal to reduce both MU and SNDK, even before memory spot prices weaken.