How Micron Could 3x From Here If AI Memory Demand Keeps Exploding

Micron reported June-quarter revenue of $41.456B, up 345.72% year over year, with non-GAAP EPS of $25.11 versus $20.2843 consensus and GAAP gross margin of 84.6%. Management guided Q4 revenue to $50B and non-GAAP EPS to $31.00, citing AI-driven demand, 16 strategic customer agreements, and roughly $100B in remaining performance obligations. The article frames a path to roughly $3,000/share as highly aggressive, while noting Goldman Sachs’ bear case that margins and pricing power will normalize as supply expands.

Analysis

MU is transitioning from a cyclical memory name into a contract-backed AI infrastructure compounder, and that changes the equity math more than the near-term tape. The key second-order effect is on capital allocation across the memory stack: if take-or-pay HBM demand really hardens, the pricing power migrates from spot-market oscillation to installed-base scarcity, which should keep utilization high and force rivals to prioritize HBM over legacy bits. That is bullish for gross margin durability in the next 12-18 months, but it also invites faster capacity build by the ecosystem, which is usually what seeds the next downcycle.

The market is likely underestimating how much of MU’s upside is already being pulled forward by multiple expansion rather than just EPS growth. At current levels, the stock does not need heroic 2026 numbers to look cheap; it needs confidence that 2027-2028 supply additions will be absorbed without a sharp reset in pricing. The real bear case is not a collapse in AI demand, but the combination of hyperscaler memory optimization and coordinated capex from all three HBM suppliers, which can flatten unit growth just as the market starts capitalizing peak margins.

GS looks structurally right on cycle risk but may be early. The most attractive setup is not a naked long after a huge run, but a time-spread or call-spread expression that captures the next two earnings beats while limiting exposure to a 2027 normalization. The contrarian miss is that HBM may resemble a quasi-standardized bottleneck for longer than skeptics expect: if AI training and inference intensity keep scaling faster than compression/efficiency gains, the market could sustain premium economics well beyond the typical memory cycle window.

Key Decisions for Investors

• Buy MU on 3-5% post-earnings pullbacks with a 6-12 month horizon; target a re-rating toward the high end of sell-side estimates if guidance cadence remains intact, but use a hard stop if HBM commentary weakens or margins fail to expand sequentially.
• Express the bull case with MU Jan-2027 $1,200/$1,500 call spreads rather than stock; this captures upside from another 2-3 quarters of guide raises while capping exposure if the market starts discounting 2027 supply.
• Initiate a relative-value short GS vs long MU basket only if MU proves another quarter of contract-backed demand; the trade benefits from GS being positioned for cycle mean reversion while MU extends margin duration.
• For more defensive exposure, pair long MU with short a memory-capex beneficiary that is more levered to commoditized demand, to isolate HBM scarcity premium from broader semiconductor beta.
• Trim or hedge MU aggressively on any sign that hyperscaler memory compression or supplier capex discipline breaks down; that is the earliest catalyst for multiple compression, likely 6-18 months before actual EPS deterioration.