Markets | S&P 500 · Nasdaq · BTC Conviction
← Portfolio
MU
Technology  ·  Updated 2026-07-07
Monitoring
6/10
Overall
7
Fundamental
6
Valuation
7
Analyst Align
8
Macro
5
Durability
Current Price
Today

Thesis

# MICRON TECHNOLOGY (MU) — UPDATED EQUITY RESEARCH NOTE

**Date of Update:** 2026-07-07

**Prior Thesis Date:** 2026-05-10 | **Prior Status:** Watchlist (5/10)

**Current Price:** $930.93 | **Market Cap:** $1.05T | **52W Range:** $103.38–$1,255.00

---

WHAT HAS CHANGED SINCE LAST THESIS

**Price:** Down ~10% from prior update ($1,035.50 → $930.93); down 19.3% in past week. Volatility remains extreme (Beta 2.14).

**New catalysts:** Ford long-term automotive memory supply agreement (2026-07-06) and GM strategic customer agreement — meaningful diversification into automotive/embedded, a higher-margin, longer-cycle segment.

**New risk:** Class action-style litigation (2026-07-05) alleging price gouging on DRAM amid AI boom. This is reputational and potentially regulatory — worth monitoring but historically antitrust suits against memory oligopolists (2018 DRAM cases) settled without existential damage.

**Insider selling accelerated:** CEO Sanjay Mehrotra sold ~120,000 shares across May–June 2026 totaling >$106M in proceeds. This is a material red flag on valuation. Prior thesis flagged this concern; it has intensified.

**Fundamentals:** Revenue growth of 345.7% YoY (TTM) reflects the extraordinary DRAM/HBM up-cycle. Forward P/E has compressed to 6.22 vs. TTM P/E 21.05 — the market is pricing in continued super-cycle earnings but is skeptical of durability. Status remains **monitoring**; the risk/reward has not yet crossed into recommend territory.

---

1. THESIS SUMMARY

**What Micron actually does:** Micron designs and manufactures DRAM, NAND flash, and High-Bandwidth Memory (HBM). It is one of only three global DRAM producers (with Samsung and SK Hynix) — an oligopoly with meaningful capital and technical barriers to entry.

**Customers:** Hyperscalers (Microsoft, Meta, Google, Amazon, Oracle) for datacenter HBM/DRAM; NVIDIA (as HBM3E buyer for H200/B100/B200 GPUs); OEMs (Apple, Samsung mobile); auto OEMs (newly signed Ford, GM); industrial/embedded customers. The customer base is highly concentrated at the top — hyperscaler capex swings drive revenue.

**Competitors:** Samsung Electronics (largest DRAM producer, ~40% share, and lead HBM competitor), SK Hynix (HBM leader with NVIDIA relationship, ~35% DRAM share), Kioxia and Western Digital (NAND). Micron is #3 in DRAM (~22–25% share) but has closed the HBM gap materially in 2025–2026.

**Value proposition:** Best-in-class node transitions (leading-edge 1-gamma DRAM), competitive HBM3E product now qualified with NVIDIA, and geographic diversification (U.S.-based fab investment funded partly by CHIPS Act — ~$6.1B awarded).

**Moat:** Oligopolistic industry structure, massive capital intensity ($10B+ per new fab), and cumulative process/yield expertise. This is a **cyclical moat** — it protects margins on the upside but does not eliminate the demand cyclicality that has repeatedly destroyed EPS in prior downturns.

**Founded:** 1978 in Boise, Idaho. **IPO:** 1984 (NASDAQ). **CEO Sanjay Mehrotra:** in role since May 2017 (~9 years). Prior co-founder of SanDisk. Solid operator; navigated the 2019 and 2022–2023 down-cycles.

**Insider ownership:** 0.3% — very low. Institutional 81.4%. CEO's recent aggressive selling ($106M+ over two months) is a signal that management views current levels as full value.

**Core thesis:** Micron is a levered play on the AI capex super-cycle via HBM. The bull case is that HBM demand is structurally different from prior DRAM cycles — supply-constrained, tied to multi-year hyperscaler roadmaps, and priced separately. The bear case is that this is a classic late-cycle memory boom that will normalize in 2027–2028 as capacity floods in.

---

2. COMPANY TIMELINE

**1978:** Founded in Boise, Idaho by Ward Parkinson and others.

**1984:** IPO on NASDAQ.

**2013:** Acquired Elpida Memory, doubling DRAM capacity.

**2017:** Sanjay Mehrotra becomes CEO.

**2019:** Cyclical trough (DRAM oversupply).

**2022–2023:** Steep memory down-cycle; posted GAAP losses. Stock bottomed near $50.

**2024:** HBM3E qualification with NVIDIA marks turning point.

**2025–2026:** Explosive revenue growth (+345.7% TTM per data provided). Stock ran from ~$103 (52W low) to $1,255 all-time high in mid-2026. Ford and GM automotive supply agreements signed in July 2026 signal diversification.

**Recent 12–24 months:** HBM ramp is the dominant narrative. Micron has moved from a laggard to a credible #2/#3 in HBM behind SK Hynix. Pricing power in DRAM is at multi-year highs. The recent price gouging lawsuit (July 2026) is the flip side of that pricing power. The 19% weekly drawdown suggests the market is beginning to question durability.

---

3. PEER & SECTOR BENCHMARKING

| Metric | MU | Samsung (mem div est.) | SK Hynix | Sector Median (Semis) |

|---|---|---|---|---|

| Revenue Growth (TTM) | 345.7% | ~60–80% | ~90% | ~15% |

| Gross Margin | 72.6% | ~45–50% | ~55% | ~50% |

| EBITDA Margin | 75.6% | ~40% | ~55% | ~35% |

| ROIC | 13.0% | ~10% | ~15% | ~12% |

| EV/EBITDA | 16.0 | ~8 | ~9 | ~18 |

| Forward P/E | 6.2 | ~10 | ~7 | ~22 |

**Caveats:** MU's reported margins (72.6% gross, 75.6% EBITDA, 80.4% operating) look anomalously high — the operating margin exceeding EBITDA margin is a data artifact; treat these numbers with caution. Historically MU's peak gross margin is ~55–60%.

**Read-through:** MU trades at a **discount** to sector on forward P/E (6.2 vs. ~22) — the market is pricing in a cyclical peak. Versus direct memory peers, MU is roughly in line. The forward P/E of 6.2 implies the market believes forward EPS ($149.64) is not sustainable. **Direct competitors:** Samsung and SK Hynix. SK Hynix is the best comp — similar cycle exposure, HBM leader, and trades at similar forward multiples.

---

4. CAPITAL ALLOCATION ASSESSMENT

**Buybacks (TTM): $0** — notable given cash generation. Management is preserving liquidity for capex.

**Dividends: $0.52B (5% yield, 1.1% payout)** — the yield figure appears distorted (payout ratio only 1.1% suggests actual yield is much lower on current earnings; likely a data artifact).

**FCF: $7.64B** — solid but modest relative to the $30B+ capex commitments over the next 3–4 years (Boise ID and Clay NY fabs).

**Debt/Equity: 6.33** — high but manageable given cash flows; CHIPS Act awards ($6.1B) and customer prepayments (HBM long-term agreements) offset.

**M&A:** No material recent activity — disciplined.

**Assessment:** Management is appropriately prioritizing fab investment over shareholder returns. The concern is capex-cycle timing: MU is building capacity into what may be peak demand, echoing 2017–2019. Optionality is somewhat constrained by the CHIPS Act commitments and HBM capex — this is not a company that can pivot quickly if demand softens.

---

5. TECHNOLOGY POSITIONING (AI TRANSITION)

Micron is a **direct beneficiary** of the AI infrastructure buildout — not a company being disrupted by AI. HBM is the memory substrate for every leading-edge AI accelerator; each NVIDIA B200 uses 8 stacks of HBM3E. Data points:

Revenue growth accelerating (345.7% TTM per provided data — extraordinary and suggests either fresh cycle acceleration or a data anomaly worth verifying against 10-Q).

Gross margins at historically elevated levels driven by HBM ASPs (HBM commands 3–5x commodity DRAM pricing).

HBM3E qualified with NVIDIA; HBM4 in development.

Ford and GM agreements signal expansion into automotive AI/ADAS memory — smaller today but higher-margin and less cyclical.

**Divergence risk:** MU is not a Salesforce-style pivot story. Its exposure to AI is direct and mechanical — memory is a commodity input to AI compute. The risk is not disruption but **cyclicality**: when hyperscaler capex growth slows, memory pricing collapses. The market narrative currently overweights the tailwind and underweights the cyclical mean reversion.

**Conclusion sentence:** The market narrative on AI as a tailwind is well-supported by operational data, but the market's implicit assumption that this cycle differs from prior memory cycles has not yet been validated by a full cycle.

---

6. BULL CASE

**HBM demand is structurally different:** Multi-year hyperscaler roadmaps, long-term supply agreements, and capacity-constrained supply keep pricing elevated through 2027+.

**Oligopoly discipline holds:** Three-player DRAM market resists the historical race-to-the-bottom capex expansion.

**Automotive/embedded diversification:** Ford, GM, and future auto OEM deals build a lower-cyclicality revenue base.

**Forward P/E of 6.2 is dislocated:** If forward EPS of $149 holds even 60%, the stock is materially undervalued.

7. BEAR CASE

**Memory is memory:** Every prior cycle ended with oversupply and price collapse. Samsung and SK Hynix are aggressively adding HBM capacity through 2027.

**CEO selling $106M+ in 60 days:** Insiders are cashing out at these levels. This is the single loudest signal in the data.

**Litigation and antitrust exposure:** DRAM price gouging lawsuit could presage regulatory action, particularly if AI compute costs become a political issue.

**Customer concentration:** NVIDIA + top 4 hyperscalers likely >50% of revenue; any hyperscaler capex pause hits hard.

**Valuation on TTM P/E (21x) is not cheap for a cyclical at peak earnings.**

8. EXIT CONDITIONS

HBM ASP declines >15% QoQ (signals oversupply).

Hyperscaler capex guidance cuts from 2+ of Microsoft/Meta/Google/Amazon.

SK Hynix or Samsung announce accelerated HBM capacity expansion above current roadmaps.

Gross margin compression below 50% (indicates cycle turn).

Continued insider selling at accelerating pace, or CEO departure.

Adverse legal outcome in DRAM litigation with material financial impact.

9. 5-YEAR EXPECTED OUTCOME RANGE

**Bull ($2,500+):** HBM cycle sustains through 2028, MU captures 25%+ HBM share, automotive scales to 15% of revenue, cycle amplitude structurally reduced. Compound revenue growth 15%+ annualized off a high base.

**Base ($900–$1,300):** Current earnings prove partially durable; a mid-cycle correction in 2027–2028 followed by recovery. Total return roughly flat to +40% over 5 years — respectable but not compelling given cyclical risk.

**Bear ($400–$600):** Classic memory down-cycle in 2027 as capacity floods in; EPS collapses to $10–$20 range; multiple compresses. 50%+ drawdown from current levels.

---

FINAL VERDICT

**Status: Monitoring (6/10 overall).** MU is a high-quality asset in the right place at the right time, but three data points prevent moving to recommend: (1) aggressive CEO insider selling, (2) cyclical history of memory that has never rewarded buying at peak margins, and (3) 19% weekly drawdown suggesting the market is repricing the cycle. I would become interested at $600–$700 with fundamentals intact, or at current levels only with evidence that HBM pricing is holding into 2027 hyperscaler orders. **No recommendation to initiate; maintaining watchlist position.**

*Sources: yfinance (price/fundamentals), SEC EDGAR (Form 4, 8-K), NewsAPI, prior thesis dated 2026-05-10.*

▲ Bull Case

  • **1978:** Founded in Boise, Idaho by Ward Parkinson and others.
  • **1984:** IPO on NASDAQ.
  • **2013:** Acquired Elpida Memory, doubling DRAM capacity.
  • **2017:** Sanjay Mehrotra becomes CEO.
  • **2019:** Cyclical trough (DRAM oversupply).
  • **2022–2023:** Steep memory down-cycle; posted GAAP losses. Stock bottomed near $50.
  • **2024:** HBM3E qualification with NVIDIA marks turning point.
  • **2025–2026:** Explosive revenue growth (+345.7% TTM per data provided). Stock ran from ~$103 (52W low) to $1,255 all-time high in mid-2026. Ford and GM automotive supply agreements signed in July 2026 signal diversification.
  • **Recent 12–24 months:** HBM ramp is the dominant narrative. Micron has moved from a laggard to a credible #2/#3 in HBM behind SK Hynix. Pricing power in DRAM is at multi-year highs. The recent price gouging lawsuit (July 2026) is the flip side of that pricing power. The 19% weekly drawdown suggests the market is beginning to question durability.

▼ Bear Case

  • **1978:** Founded in Boise, Idaho by Ward Parkinson and others.
  • **1984:** IPO on NASDAQ.
  • **2013:** Acquired Elpida Memory, doubling DRAM capacity.
  • **2017:** Sanjay Mehrotra becomes CEO.
  • **2019:** Cyclical trough (DRAM oversupply).
  • **2022–2023:** Steep memory down-cycle; posted GAAP losses. Stock bottomed near $50.
  • **2024:** HBM3E qualification with NVIDIA marks turning point.
  • **2025–2026:** Explosive revenue growth (+345.7% TTM per data provided). Stock ran from ~$103 (52W low) to $1,255 all-time high in mid-2026. Ford and GM automotive supply agreements signed in July 2026 signal diversification.
  • **Recent 12–24 months:** HBM ramp is the dominant narrative. Micron has moved from a laggard to a credible #2/#3 in HBM behind SK Hynix. Pricing power in DRAM is at multi-year highs. The recent price gouging lawsuit (July 2026) is the flip side of that pricing power. The 19% weekly drawdown suggests the market is beginning to question durability.

Exit Conditions

Conviction Timeline

6.0/10 2026-05-10 5.0/10 2026-06-02

Mentioned in Briefs

Change History

reaffirm
50-day rolling review. Conviction: 5/10
2026-06-02
new
AI Supercycle Phase 1 batch report. Conviction: 6/10.
2026-05-10
Chat with Meridian
Ask anything about your portfolio
Hey William 👋 Ask me anything about your portfolio, a specific stock, Bitcoin, or the market. I have context on your current positions and theses.