MU Micron Technology

Research status: Primary-source verified through Micron fiscal Q4/FY2026 results released 30 Sep 2026. Valuation basis: $1,065.11 close on 30 Sep 2026.

Micron exited FY2026 at extraordinary AI-memory economics: fiscal Q4 revenue was $54.23B, GAAP gross margin 86.8% / non-GAAP 87.0%, non-GAAP EPS $33.42, operating cash flow $43.97B, and adjusted FCF $33.20B. FY2026 revenue reached $133.19B and adjusted FCF $62.31B. Management guided FQ1 FY2027 to $61.5B ± $1.5B revenue, approximately 86.25% non-GAAP gross margin, and $38.15 ± $1 non-GAAP EPS.

The operating evidence strengthens the structural-HBM thesis and argues against an imminent collapse back to legacy memory economics. It does not remove cyclicality risk: 86–87% gross margins are exceptional and should not be capitalized as a permanent steady state. PowerFund therefore continues to treat MU as an AI-memory benchmark and a potential correction entry rather than a chase at peak-cycle profitability.

Investment case

AI accelerators and AI data centers are driving exceptional demand for HBM, high-performance server DRAM and storage. Micron is one of only three scaled global DRAM suppliers, and its current results show that constrained qualified supply plus AI mix can produce economics far above historical memory-cycle norms.

> Updated PowerFund thesis: HBM and AI-server memory appear to be raising Micron's structural earnings floor and cash-generation durability, but today's 86–87% gross margin remains too extreme to underwrite as normal. Capital should be based on normalized-cycle earnings power, HBM/platform share and supply discipline—not on annualizing the current quarter.

Verified FY2026 / FQ4 baseline

• FQ4 revenue: $54.23B.
• FQ4 GAAP / non-GAAP gross margin: 86.8% / 87.0%.
• FQ4 non-GAAP EPS: $33.42.
• FQ4 operating cash flow: $43.97B.
• FQ4 adjusted FCF: $33.20B.
• FY2026 revenue: $133.19B.
• FY2026 adjusted FCF: $62.31B.
• FQ1 FY2027 guide: $61.5B ± $1.5B revenue, ~86.25% non-GAAP gross margin, $38.15 ± $1 non-GAAP EPS.

Normalized underwriting

PowerFund does not annualize the current 86–87% gross margin. A useful normalization ladder is:

• ~50% gross margin: treat as a severe but still structurally improved memory normalization; earnings power could fall toward roughly $45–60/share depending on revenue and capex.
• ~60% gross margin: structurally stronger HBM-cycle case; roughly $65–85/share normalized EPS is plausible.
• ~70% gross margin: sustained AI-memory scarcity / mix case; roughly $90–110+/share normalized EPS is plausible.

These are PowerFund estimates, not company guidance, and are intended to prevent peak-margin capitalization.

PowerFund valuation scenarios from $1,065.11

24 months

• Bear (30%) — $700–850: industry supply catches up, gross margin compresses toward ~50%, HBM premium narrows and the market de-rates memory-cycle earnings.
• Base (45%) — $1,200–1,450: HBM remains structurally tight, gross margin normalizes toward ~60%, and Micron sustains a materially higher earnings floor than prior cycles.
• Bull (25%) — $1,800–2,200: HBM4/HBM4E and AI-server demand sustain ~70% normalized margins longer than expected with disciplined supply and strong per-share cash compounding.

Probability-weighted midpoint is roughly $1,325–1,350, implying only low-teens annualized expected return over two years from the current close—good, but not obviously enough for peak-cycle/crowding risk.

60 months

• Bear — $800–1,000: legacy cyclicality substantially reasserts itself.
• Base — $1,600–2,100: HBM establishes a durable higher earnings floor and cash generation compounds.
• Bull — $2,600–3,300+: AI memory remains a persistent bottleneck and Micron sustains leading platform participation.

At the current price, MU is watch / buy-on-correction, not a chase. A materially better entry should come either from a price dislocation or from further evidence that normalized margins and HBM share deserve another upward revision.

• FQ1 FY2027 delivery versus $61.5B revenue / ~86.25% non-GAAP gross-margin guidance.
• Evidence that HBM4 / HBM4E qualification and customer share broaden beyond a single lead platform.
• Strategic Customer Agreements translating into more durable demand/pricing visibility.
• Continued strong FCF and balance-sheet compounding through FY2027.
• Capacity / advanced-packaging plans remaining disciplined relative to contracted AI-memory demand.
• AI-server DRAM and data-center SSD demand remaining strong alongside HBM.

• Peak-margin normalization: 86–87% gross margins are extraordinarily high and likely above sustainable steady-state economics.
• Supply response: extreme profitability can induce DRAM/HBM capacity additions and packaging expansion.
• HBM competition: SK hynix and Samsung remain formidable and qualification share can move quickly.
• Customer/platform concentration: hyperscaler and accelerator-platform concentration creates bargaining and qualification risk.
• Capex intensity: technology transitions and manufacturing expansion require very large capital commitments.
• Crowding / valuation: MU is now an obvious AI-memory winner; a good company can still be a poor entry if normalized returns are insufficient.
• Classic memory cyclicality: if supply outruns demand, pricing and margins can compress far faster than trailing multiples imply.

Warning — freeze additions

• HBM qualification/share loss at a major accelerator platform.
• Gross-margin guidance falls below ~60% materially earlier than the normalized framework expects.
• Inventory grows materially faster than revenue for two quarters.
• Capex / capacity plans accelerate while pricing or contract visibility weakens.
• Sequential revenue decline before HBM4E and broader platform ramps mature.

Reduce

• Gross margin normalizes below ~50% with no evidence of a structurally higher HBM earnings floor.
• HBM pricing premium compresses rapidly or major strategic-customer agreements fail to translate into durable economics.
• Industry supply growth materially exceeds AI-memory demand.
• Operating cash flow / FCF deteriorates sharply despite reported earnings.

Invalidate

> Invalidate if HBM proves to be only another short-lived memory spike—specifically if Micron loses HBM/platform share while oversupply drives margins and cash returns back toward prior-cycle economics.

Hard invalidators include a major HBM reliability/qualification failure or material balance-sheet deterioration into a memory downturn.

The memory market remains an oligopoly: SK hynix, Samsung and Micron. HBM narrows the field further because qualification, advanced packaging and thermal/power performance matter.

Micron's strategic opportunity is to combine HBM with broad DRAM/NAND/server-memory relationships. The moat is real but cyclical: customers need qualified suppliers, yet memory bits are ultimately fungible enough that supply discipline remains central.

1. Build a cleaner segment/HBM bridge from prepared remarks and 10-K: estimate HBM revenue, DRAM/HBM mix and gross-profit contribution separately from commodity DRAM/NAND.
2. Track HBM4/HBM4E qualifications and share by accelerator platform.
3. Compare Micron capacity, wafer-start and advanced-packaging expansion with SK hynix and Samsung.
4. Re-run the 50% / 60% / 70% gross-margin normalization after FQ1 FY2027 and whenever pricing/capacity evidence changes.
5. Monitor inventory, strategic customer agreements and contract duration for evidence that cyclicality is genuinely lower.
6. Reassess entry zones after a material correction or if normalized earnings power is revised upward.