Micron Earned $24.67 Per Share Last Quarter. Its Dividend Is Still 15 Cents. Something Has to Give.

Micron reported fiscal Q3 earnings of $24.67/share (GAAP) and revenue of $41.46B, up 346% YoY, with net income of $28.24B, driven by high-bandwidth memory demand for AI accelerators. The company guided to about $50B revenue in the current quarter and generated $25.4B operating cash flow and $18.3B adjusted FCF, ending with ~$30.2B cash/investments. While the dividend stays at $0.15/share (payout ratio well under 1%), management said it intends to return 100% of its excess cash over time and plans step-up capital returns later this year.

Analysis

MU’s real signal is not the quarterly dividend gap; it is that a once-cyclical memory name is now trading like a scarcity asset. Near term, that supports multiple expansion because investors will pay up for visibility when the core product is effectively sold out, but that same visibility usually invites aggressive capex across the industry with a lag. The second-order winner is the AI hardware stack that depends on stable HBM supply — especially NVDA, which benefits from fewer memory bottlenecks and lower component volatility in its accelerator roadmap.

The bigger medium-term risk is classic memory discipline failure. If MU and peers keep reinvesting at peak economics, the market will eventually discount 6-12 months of capacity additions before the cash returns show up, which can compress both gross margin and EV/FCF simultaneously. That makes this more of a “stay long until capex or spot pricing inflects” setup than a forever compounder thesis.

The contrarian point is that the market may be underestimating how quickly today’s excess cash can be recycled back into supply. The bullish narrative only holds if HBM remains structurally constrained; if lead times shorten or customers diversify memory sourcing, the current earnings power will look cyclical again. I’d watch for any change in capex guidance, gross margin trajectory, or commentary on booked orders as the cleanest falsifiers over the next 1-3 quarters.

Key Decisions for Investors

• Long MU on 5-10% pullbacks over the next 1-2 weeks; target a 3-6 month hold into the capital-return update. Risk/reward is attractive while visibility stays high, but reduce if management signals capex is rising faster than operating cash flow.
• Pair trade: long MU / short WDC or long MU / short a broader memory basket if available, as MU’s HBM mix should sustain premium pricing longer than commodity NAND/DRAM exposure. Use this as a 1-3 month relative-value expression, not a directional macro bet.
• Buy NVDA as a beneficiary of lower memory-supply friction; the trade works best if you expect AI accelerator demand to remain the binding constraint and memory to stop being a bottleneck. Falsify the thesis if MU commentary implies HBM supply is no longer tight by the next earnings cycle.
• Avoid chasing MU via short-dated calls after the print; implied volatility is likely elevated and the stock is now sensitive to any sign of peak-cycle behavior. If you want convexity, use a longer-dated call spread and size for a 6-12 month capex/return cadence.
• Watch for a reversal trigger: if DRAM/HBM pricing softens or MU raises capex again without a corresponding margin expansion, fade the multiple. That would be the first indication the market is pricing a structural AI annuity that is really just a delayed cyclical peak.