Micron: More Great Earnings Don't Change Our Mind

Micron, valued at roughly $1.2T, is characterized as offering only S&P 500-like prospective returns despite strong DRAM and NAND demand. Strategic Customer Agreements lock in more than 35% of revenue through 2030 with price floors and ceilings, but have not been tested in a severe downturn. Heavy capacity investment by Micron, Samsung, SK hynix and CXMT could weaken pricing power and compress margins after 2028.

Analysis

The central analytical issue is not near-term demand but whether MU is exchanging commodity upside for revenue visibility. Price ceilings would cap operating leverage during the next tight-memory phase, while floors are only valuable if counterparties remain obligated through a recession; absent disclosure on termination rights, volume commitments, and credit support, the agreements should not be capitalized as annuity-like revenue. The stated valuation premise also requires verification before any multiple-based conclusion: the referenced market-cap figure appears inconsistent with public-company reality and could materially distort return assumptions.

A synchronized capacity build is initially supportive for semiconductor-capital-equipment utilization, but it creates a 2028-30 risk that bit-supply growth outpaces AI/server demand normalization. MU is more exposed than logic semis to a small change in utilization: a modest decline in DRAM/NAND pricing can flow disproportionately through gross margin, inventory write-downs, and FCF. SK hynix (000660.KS) has relatively stronger HBM positioning, so a broad memory oversupply thesis is not automatically a clean short-SK hynix expression; the more likely divergence is HBM pricing resilience versus conventional DRAM/NAND compression.

Consensus may be underestimating the possibility that customer agreements lower both downside and upside volatility, warranting a lower beta but not necessarily a lower earnings multiple. The bearish structural thesis is premature without evidence that industry capex is translating into usable leading-edge supply rather than technological-node transitions. Near-term confirmation should come from quarterly bit-supply guidance, MU gross-margin trajectory, customer prepayment/deposit disclosures, and whether contract language is independently described in filings rather than management commentary.

Key Decisions for Investors

• No outright MU short solely on the long-dated capacity thesis; the relevant oversupply window is likely 2028-30, while a near-term memory upcycle can overwhelm valuation concerns over the next 1-3 months.
• Set a MU risk alert around the next two earnings reports: reassess bearish exposure if gross-margin guidance falls more than 300 bps sequentially, inventory days rise, or capex guidance accelerates while DRAM/NAND pricing indicators flatten. These would be earlier evidence that supply is catching demand.
• Conditional 3-6 month pair: long SK hynix (000660.KS) / short MU only after verifying that HBM revenue mix and pricing remain stronger at SK hynix while MU guides weaker conventional-memory margins. Target a 10-15% relative return; stop if MU closes the HBM qualification/share gap or contract disclosures demonstrate enforceable take-or-pay protection.
• Watch AMAT and LRCX for a separate 6-12 month capex read-through rather than treating them as direct MU hedges. Sustained memory-WFE order strength supports their backlog near term, but materially raises the later memory-price downside; reduce equipment longs if customers defer fab-tool delivery dates rather than merely lower spending guidance.