Micron: HBM-led supercycle still mispriced as a traditional memory trade

Stevie AI on Micron Technology, Inc. (MU-USA | microntechno)

10/26/2026

Summary

Micron manufactures DRAM and NAND memory for data center, PC, mobile, and embedded markets, and is riding an unprecedented supply-demand imbalance driven by HBM for AI accelerators. The structural insight is that this isn't a typical memory cycle peak — clean-room construction lead times (new capacity online late 2028+) and the rising silicon trade ratio of HBM versus conventional DRAM mean supply cannot catch demand for at least two more years, even as agentic AI workloads push server unit growth into the high teens. The market continues to price Micron on trough-cycle memory multiples despite management signaling margin expansion ahead and >75% of FY2027 shipments already committed under long-dated agreements. The historical numbers validate the scale of this inflection: revenue went from $37.4B in FY2025 to $133.2B in FY2026, with net income exploding from $8.5B to $85.0B and EPS from $7.59 to $74.33, as gross margins reached 80.7% and net margins hit 63.8%. This is not a modest upcycle — it is one of the largest single-year profitability re-ratings in semiconductor history, driven almost entirely by HBM mix shift and an industry-wide supply shortfall that Micron did not create and cannot quickly solve even if it wanted to. Our price targets are built off the EPS path: FY2027 EPS of $98.87, FY2028 $116.52, FY2029 $117.58, and FY2030 $112.33, as margins normalize off peak but remain structurally elevated versus pre-AI history. We apply a 9x forward P/E — below typical semiconductor-cycle peak multiples of 12-15x reflecting the market's (correct) skepticism that 80% gross margins are permanent, but above trough-cycle multiples of 5-6x given genuine structural demand support from HBM and multi-year SCA lock-ins. This yields price targets of $889.83 (FY2027), $1,048.68 (FY2028), $1,058.22 (FY2029), and $1,010.97 (FY2030) — modest upside near-term but meaningful compounding of per-share value as net cash builds toward $278B by FY2030, setting up eventual capital return.

Thesis

1. **Structural Supply Constraint, Not a Typical Cycle Peak**: The conventional memory cycle narrative — supply catches demand within 12-18 months — does not apply here because the bottleneck has shifted from equipment lead times to clean-room physical construction. Management has been explicit that new clean-room capacity will not come online until late calendar 2028 at the earliest, meaning the industry cannot meaningfully add bit supply for roughly two more years regardless of capital commitments. This is a multi-year structural constraint, not a cyclical one, and it is the single most important reason this upcycle should be valued differently than prior DRAM/NAND peaks in 2017-2018 or 2021. 2. **HBM Trade Ratio Dynamics Compound the Shortage**: As the industry transitions from HBM3 to HBM4/HBM4E, the silicon trade ratio — the amount of wafer capacity consumed per effective bit of usable memory — is rising, meaning each unit of HBM demand consumes disproportionately more capacity than conventional DRAM. Combined with diminishing bit-growth returns from each successive node transition, this means Micron's effective supply growth is decelerating even as nominal capex rises, reinforcing pricing power for the foreseeable future. 3. **Revenue and Earnings Visibility Via SCAs**: Management's disclosure that over 75% of FY2027 shipments are already contractually committed gives unusual visibility for a historically commoditized, spot-price-driven business. This de-risks the near-term earnings path meaningfully relative to prior cycles where Micron's revenue was almost entirely exposed to spot pricing volatility, and it justifies a premium to trough multiples even though the market continues to treat memory earnings as inherently unreliable. 4. **Margin Trajectory Still Underappreciated**: Despite Q1 FY2027 guidance for single-digit sequential bit growth and double-digit cost increases, management expects margin expansion for the remainder of FY2027 as pricing (albeit decelerating), favorable mix, and efficiency gains offset start-up costs from new fabs. The market appears to be anchoring on Q1 cost inflation as a signal of margin peak, when management's own guidance points to sequential improvement through the fiscal year. 5. **Balance Sheet Inflection Toward Capital Return**: Net cash is forecast to grow from a de minimis base to $97.9B (FY2027), $172.0B (FY2028), $231.1B (FY2029), and $278.2B (FY2030). This is an extraordinary balance sheet transformation for a company that has historically run with meaningful net debt through capex-heavy periods. Once this net cash position is firmly established, management has signaled capital return via buybacks will begin — a catalyst the market is not yet pricing given the current focus on cyclical risk. 6. **Valuation Mismatch**: At $886.70, the stock trades at roughly 12x FY2026 EPS of $74.33 and under 9x our FY2027 EPS estimate of $98.87 — multiples consistent with a business expected to see immediate margin collapse. Given the structural supply constraints through FY2028 and still-elevated (if moderating) margins projected through FY2030, we believe a 9x forward multiple on a smoothed, still-high-but-normalizing EPS base is the appropriate anchor, offering upside as the market digests that this cycle has a longer runway than historical memory supercycles.

Risks

1. **Cyclical Pricing Reversal Risk**: The entire thesis rests on supply remaining constrained through FY2028. If clean-room construction accelerates faster than guided, if competitors (Samsung, SK Hynix) add capacity more aggressively, or if AI capex growth decelerates even modestly, pricing could reverse quickly and the extraordinary 63.8% net margin of FY2026 could compress toward historical mid-cycle levels of 15-25%, invalidating both our earnings and multiple assumptions simultaneously. 2. **Customer Concentration and Hyperscaler Capex Discipline**: Micron's demand is increasingly tied to a small number of hyperscale AI customers. Any pause, delay, or reallocation of AI infrastructure capex by major cloud providers — whether due to ROI scrutiny on AI spending, macro tightening, or a shift toward more efficient inference architectures requiring less memory bandwidth — would disproportionately hit Micron's highest-margin HBM revenue stream. 3. **Execution Risk on Capacity Expansion**: The Idaho fab's first wafer output is targeted for mid-calendar 2027, and the broader clean-room buildout is a multi-year, capital-intensive construction program. Delays, cost overruns, or equipping decisions misaligned with actual demand trends (management has stated capacity will be equipped

📈 Price Targets