Back to Portfolio
MU

Micron Technology

$487 550B market cap April 15, 2026
Micron Technology Inc MU BUFFETT / MUNGER / KLARMAN SUMMARY
1 SNAPSHOT
Price$487
Market Cap550B
2 BUSINESS

Micron Technology is the #3 global memory semiconductor company, executing brilliantly at the peak of a historic HBM/AI demand supercycle. FQ2 2026 gross margins of 75% and revenue of $23.9B are the highest in company history, driven by AI data center demand for HBM, server DRAM, and enterprise SSDs. HBM represents a genuine structural improvement that will elevate Micron's mid-cycle earnings above historical averages. However, the memory industry remains fundamentally cyclical with commodity economics, zero switching costs, and inevitable capacity overshoot. At $487 (7.6x book, 23x peak EPS, 49-61x mid-cycle EPS), the stock prices in a permanent supercycle that has never occurred in memory. Over 28 years, Micron posted negative EPS in 11 of them. The correct approach is to wait for the inevitable cyclical downturn -- which history suggests arrives 6-18 months after peak margins -- and buy aggressively below $250 when the market capitulates on memory stocks, as it did in late 2022 and early 2025.

3 MOAT NARROW

3-player DRAM oligopoly (Samsung 40%, SK Hynix 30%, Micron 25%) with massive barriers to entry ($15-20B per fab). Technology leadership in 1-gamma DRAM node (4 consecutive node leads). HBM provides temporary differentiation. However, products are fungible commodities with zero switching costs.

4 MANAGEMENT
CEO: Sanjay Mehrotra

Mixed -- excellent technology investment but capex is driven by competitive necessity, not choice. Token dividend ($0.60/yr). Buybacks suspended. CHIPS Act grants help fund US fabs.

5 ECONOMICS
67.6% Op Margin
20.2% ROIC
39.8% ROE
23x P/E
1.7B FCF
3.5% Debt/EBITDA
6 VALUATION
FCF Yield0.3%
DCF Range175 - 300

Overvalued by 63-178%. At $487, priced for permanent supercycle. Mid-cycle fair value ~$200-250.

7 MUNGER INVERSION
Kill Event Severity P() E[Loss]
Memory cycle downturn -- margins revert from 75% gross to negative, as happened in FY2023 (-9% gross margin) HIGH - -
HBM competition -- Samsung catching up, margins will compress from 3-5x to 1.5-2x regular DRAM MED - -
8 KLARMAN LENS
Downside Case

Memory cycle downturn -- margins revert from 75% gross to negative, as happened in FY2023 (-9% gross margin)

Why Market Right

Massive industry capex ($20-25B MU alone, plus Samsung/SK Hynix) creates oversupply risk for 2027-2028; AI capex deceleration -- if hyperscaler spending growth slows even 10-20%, HBM demand drops sharply; Samsung HBM4 qualification closing technology gap, normalizing HBM margins; China market permanently lost -- Micron exiting China server chips

Catalysts

HBM TAM growing to $100B by 2028 (40% CAGR) -- raises mid-cycle earnings structurally; Full calendar 2026 HBM supply sold out under long-term agreements; 1-gamma DRAM technology leadership extends cost advantage; Data center SSD revenue exceeded $1B quarterly -- diversifying beyond DRAM

9 VERDICT WAIT
B+ Quality Moderate - Solid at peak but balance sheet deteriorates in downturns. Issued $5B debt during FY2023 trough. Not fortress-grade.
Strong Buy$175
Buy$250
Fair Value$300

Do not buy at current prices. Set alerts for $250 (Accumulate) and $175 (Strong Buy). Be prepared to act quickly when the cycle turns -- memory troughs are sharp but short-lived.

🧠 ULTRATHINK Deep Philosophical Analysis

Micron Technology (MU) - Deep Philosophical Analysis

A Meditation on Memory, Cycles, and the Seduction of Peak Earnings

The Core Question: Can a Commodity Producer Become a Franchise?

Warren Buffett famously said he would not own a semiconductor company. Charlie Munger elaborated: "The trouble with semiconductors is that the minute you build the best plant in the world, your competitor is building a better one." This observation applies to no company more directly than Micron Technology.

Micron makes memory chips. DRAM and NAND. These are the most commoditized products in all of technology. A DRAM chip from Micron performs identically to one from Samsung or SK Hynix. No customer cares who made the memory in their server, phone, or laptop. There is no brand loyalty, no switching cost, no ecosystem lock-in. Memory is bought on price and availability, full stop.

And yet, here we are in April 2026, with Micron posting the highest gross margins in its 48-year history -- 75% in the most recent quarter. That is Apple territory. That is LVMH territory. It is utterly extraordinary for a company whose products are, by definition, interchangeable with its competitors'.

The question that must be answered before any investment decision: Is this a permanent change in the nature of the memory business, or is this the most seductive cyclical peak the industry has ever produced?

Moat Meditation: The Oligopoly Illusion

There is an argument that DRAM has evolved into a natural oligopoly with structural pricing discipline. Only three companies remain: Samsung, SK Hynix, and Micron. The last meaningful new entrant was Elpida, which went bankrupt in 2012. The barriers to entry are staggering -- $15-20 billion for a single fab, five years from groundbreaking to production, and technology requirements that are among the most advanced in all of manufacturing.

This oligopoly argument is seductive, and it contains a grain of truth. The DRAM industry IS more rational than it was in the era of 10+ players in the 1990s and 2000s. The boom-bust cycles have moderated somewhat in amplitude, and the floor of profitability in troughs has risen (though it is still deeply negative, as FY2023 demonstrated).

But here is what the oligopoly bulls miss: all three players are investing $20-30 billion per year in new capacity right now. Samsung, SK Hynix, and Micron are collectively pouring $60-80 billion annually into new fabs and equipment. This is not the behavior of disciplined oligopolists protecting margins. This is the ancient, compulsive pattern of commodity producers investing at the peak of the cycle because demand appears insatiable and they fear losing share if they do not keep pace.

The mechanism that creates the next downturn is already in motion. Idaho (2027). New York (2030). Samsung's expanded Pyeongtaek complex. SK Hynix's expanded Icheon facilities. The capacity that is being built today will come online in 2027-2029, and it will be delivered into a market where AI capex growth has inevitably decelerated from the current torrid pace.

The HBM Question: Permanent Premium or Temporary Windfall?

HBM (High Bandwidth Memory) is the specific driver of Micron's current extraordinary performance. HBM stacks multiple DRAM dies vertically with through-silicon vias and bonds them to a base logic die, creating a product with dramatically higher bandwidth than conventional DRAM. It is essential for AI accelerators like NVIDIA's GPUs and Google's TPUs.

HBM genuinely IS different from standard DRAM in several important ways:

  • It is more complex to manufacture (advanced packaging, thermal management)
  • It requires closer customer collaboration (custom configurations for each accelerator)
  • It has a 3:1 wafer consumption ratio versus standard DRAM
  • It currently sells at 3-5x the price per bit of standard DRAM

These factors suggest HBM will carry a structural premium over standard DRAM for the foreseeable future. I accept this premise. The question is not whether HBM earns premium margins, but HOW MUCH of a premium, and for how long the current extraordinary margins last.

History provides the answer, and it is humbling. Every time memory has experienced a structural demand driver -- PCs in the 1990s, smartphones in the 2010s, data centers in the 2020s -- the initial scarcity premium was competed away within 2-4 years as all players expanded capacity to meet the new demand. The demand driver was real and permanent. But the super-margins were temporary.

I believe HBM will settle at a 50-100% ASP premium over standard DRAM (1.5-2x), not the current 300-500% premium (3-5x). This is still attractive and structurally elevates Micron's mid-cycle earnings. But it is not 75% gross margins.

The Owner's Mindset: Would Buffett Own This for 20 Years?

Absolutely not, and he would explain why with clarity:

"I want to own businesses where the competitive landscape is clear, where my business will be stronger in 10 years than it is today, and where I don't have to make brilliant capital allocation decisions just to keep up. Micron is the opposite of this. They must spend $15-20 billion per year merely to stay in the game. One wrong technology bet and they fall behind for years. One prolonged downturn and their balance sheet comes under stress. And the product they sell has no brand, no loyalty, and no price premium except what supply scarcity temporarily grants."

Micron is a technology investment, not a business ownership investment. The distinction matters enormously. Technology investments can be wildly profitable, but they require timing -- buying cheap and selling dear, navigating cycles, and accepting that the economic moat is always under competitive assault.

The Buffett portfolio approach -- buy and hold forever -- does not work with cyclical commodity producers, no matter how brilliantly managed. Micron under Sanjay Mehrotra is the best-managed it has ever been. Technology leadership, operational discipline, strategic HBM positioning -- all excellent. But management quality cannot override industry economics. Samsung and SK Hynix are also brilliantly managed, and they are investing just as aggressively.

Risk Inversion: What Could Destroy This Business?

Let us ask Munger's inversion question: "Tell me where I'm going to die, so I'll never go there."

Scenario 1: AI Winter / Capex Pullback (20% probability over 3 years) If hyperscaler AI capex growth decelerates sharply -- due to diminishing returns from scaling, regulatory intervention, or economic recession -- HBM demand could stall while massive new capacity comes online. This would create a downturn more severe than FY2023, potentially driving MU below $100.

Scenario 2: Chinese Memory Independence (10% probability over 5 years) CXMT and YMTC achieve cost-competitive DRAM and NAND production at scale. While currently 2-3 generations behind, Chinese memory makers have massive government support and captive domestic demand. If they succeed, Micron faces a new competitor in what is supposed to be a closed oligopoly.

Scenario 3: Technology Disruption (5% probability over 10 years) A post-DRAM memory technology (persistent memory, processing-in-memory, optical memory) reduces demand for conventional DRAM/NAND. Unlikely in the near term but should be monitored.

None of these scenarios would literally destroy Micron, but each could cause a 60-80% stock price decline from current levels. The most likely scenario -- a normal cyclical downturn without any structural disruption -- would itself cause a 50%+ decline based on historical patterns.

Valuation Philosophy: The Price of Cyclicality

The challenge with valuing cyclical businesses is that every traditional metric lies to you at the peak. Micron's forward P/E of 7.7x appears cheap. But that "forward" number assumes FY2027 earnings of ~$63/share, which requires 75%+ gross margins to persist through an entire year. This is like valuing a homebuilder at the 2006 earnings run rate, or an oil producer at $150/barrel spot prices.

The correct approach for cyclical businesses is one of three methods:

  1. Normalized earnings P/E: Use average earnings over a full cycle (7-10 years). Micron's 10-year average EPS is ~$4.30. At $487, that is 113x normalized earnings. For a cyclical with no structural moat, this is indefensible.
  2. Book value floor: Micron has never traded below 1x book value for long. Current book of $64 provides a theoretical floor. At $487, you are paying 7.6x book -- the highest in history.
  3. Through-cycle FCF yield: Cumulative FCF over 7 years was ~$5B on a current market cap of $550B. That is a 0.13% annual FCF yield. Atrocious.

All three methods converge on the same conclusion: Micron at $487 is egregiously overvalued relative to its through-cycle economics. The only justification is that HBM has permanently elevated the cycle, and 75% gross margins are the new floor. I assign less than 10% probability to that outcome.

The Patient Investor's Path

The beautiful thing about deeply cyclical stocks is that they give you extraordinary buying opportunities on a regular schedule. Micron has traded below $50 in 2012, 2016, 2019, 2022, and 2025. That is five buying opportunities in 13 years -- roughly one every 2.5 years.

The patient investor's playbook for Micron:

  1. Wait. Do not buy at $487. The expected value of buying here is deeply negative.
  2. Prepare. Set alerts at $250 (Accumulate) and $175 (Strong Buy). Understand the business well enough to act decisively when the opportunity arrives.
  3. Act at the trough. When memory is in crisis -- when sell-side notes say "avoid," when the stock has fallen 60%, when gross margins are heading toward zero -- that is when you buy. Not at 75% gross margins.
  4. Size appropriately. Even at trough prices, limit position to 3-5% of portfolio. Cyclicals can stay depressed longer than expected, and timing the exact bottom is impossible.
  5. Sell at the peak. This is not a Buffett "hold forever" stock. When margins are expanding and the market celebrates, trim or exit. The cycle will turn.

The most dangerous words in memory investing are: "This time is different because of [AI/smartphones/cloud/PCs/etc.]." The demand driver may be real. The super-margins never last.

"In the short run, the market is a voting machine but in the long run, it is a weighing machine." The memory industry weighs about $25 billion of mid-cycle revenue, 15-20% operating margins, and anemic free cash flow after maintenance capex. The market is currently voting for $550 billion. The weighing machine will eventually correct this.

Executive Summary

Metric Value Assessment
Current Price ~$487 Near all-time highs
52-Week Range $73.32 - $491.98 Trading near top
Market Cap $550B Mega-cap
P/E (TTM) 23.0x Moderate at peak earnings
P/E (Forward) 7.7x Cheap if earnings sustain
Book Value/Share $64.24 7.6x P/B - rich
Dividend Yield 0.1% Token dividend
ROE (TTM) 39.8% Peak cycle - unsustainable
EPS (TTM) $21.21 Peak cycle earnings
Beta 1.61 Highly volatile

VERDICT: WAIT. Micron is executing brilliantly at the peak of a historic HBM/AI upcycle, but the stock at $487 prices in perfection. The memory industry's brutal cyclicality means buying at peak margins (75% gross!) carries extreme risk of 50-70% drawdown when the cycle turns. Accumulate aggressively below $250; Strong Buy below $175.

Phase 1: Risk Assessment

1.1 Memory Cycle Risk (CRITICAL)

Memory semiconductors are the most cyclical segment in all of technology. Micron's operating history proves this:

Fiscal Year Revenue ($B) Gross Margin Net Income ($B) EPS
FY2018 (Aug) $30.4 58.9% $14.1 $11.95
FY2019 $23.4 45.7% $6.3 $6.29
FY2020 $21.4 30.6% $2.7 $2.83
FY2021 $27.7 37.6% $5.9 $6.06
FY2022 $30.8 45.2% $8.7 $8.34
FY2023 $15.5 -9.1% -$5.8 -$4.45
FY2024 $25.1 22.4% $0.8 $1.27
FY2025 $37.4 39.8% $8.5 $8.29
FQ1 2026 $13.6 (Q) 56.8% $5.2 (Q) $4.60
FQ2 2026 $23.9 (Q) 75.0% $13.8 (Q) $12.07

The pattern is unmistakable: Revenue swings from $15B to $30B+ to $15B again. Gross margins oscillate from 59% to negative 9% and back. Net income moves from +$14B to -$6B. This is NOT a stable business -- it is a deeply cyclical commodity producer experiencing its best quarter in history.

Current cycle position: We are at or near peak. FQ2 2026 gross margin of 75% is the highest in Micron's history. HBM demand from AI is extending the upcycle, but history says margins this elevated ALWAYS revert.

1.2 HBM Competition Risk (HIGH)

Micron's current outperformance is driven by HBM (High Bandwidth Memory) for AI accelerators. Key concerns:

  • Samsung is catching up. Samsung's HBM3E has been qualified by NVIDIA and is shipping in volume. Samsung's HBM4 is in development.
  • SK Hynix was first to HBM. SK Hynix remains the dominant HBM supplier with ~50% share. Micron has gained to ~20-25% but SK Hynix has a structural lead.
  • 3-to-1 cannibalization. Each HBM die consumes 3x the DRAM wafer capacity of standard DDR5. This tightens conventional DRAM supply now, but when HBM demand plateaus, the reversion will flood the DDR5 market.
  • HBM margins are extraordinary but temporary. Reportedly 3-5x higher ASP than standard DRAM. As Samsung fully qualifies and competition normalizes, these margins will compress.

1.3 China Risk (MODERATE-HIGH)

  • China banned Micron products from critical infrastructure in May 2023
  • Micron is reportedly exiting China server chip sales entirely
  • China was ~12% of revenue in FY2024, declining to mid-single digits
  • Chinese domestic memory makers (CXMT, YMTC) are expanding but still trail in advanced nodes
  • Loss of China market is partially offset by HBM/AI demand elsewhere, but it is a permanent structural revenue loss

1.4 Capital Intensity Risk (HIGH)

  • FY2026 capex guided to $20-25 billion (up from $13.8B in FY2025)
  • New Idaho fab, New York fab, Singapore HBM facility, Japan upgrades
  • Memory fabs cost $15-20B each and take 3-4 years to build
  • This capex must be spent regardless of where we are in the cycle
  • Depreciation will rise dramatically in FY2027-2029 as new fabs come online, compressing margins even if revenue holds

1.5 NAND Oversupply Risk (MODERATE)

  • NAND is structurally less profitable than DRAM (more competitors: Samsung, SK Hynix, Kioxia/WD, Micron, YMTC)
  • Micron exited mobile managed NAND (low-margin) in FY2025 to focus on data center SSDs
  • NAND has historically been a drag on profitability, with negative margins in downturns
  • Currently benefiting from HDD supply shortages and AI data storage demand

1.6 Commodity Pricing / Zero Switching Costs (STRUCTURAL)

  • DRAM and NAND are fungible commodities. A Samsung DDR5 module and a Micron DDR5 module are interchangeable.
  • Customers (hyperscalers, OEMs) routinely dual- or triple-source memory
  • No meaningful switching costs exist except in HBM (temporary differentiation)
  • Pricing is determined by supply/demand balance, not brand loyalty

Phase 2: Financial Analysis

2.1 Income Statement Through the Cycle

Metric FY2018 FY2019 FY2020 FY2021 FY2022 FY2023 FY2024 FY2025 FQ2'26 Ann.
Revenue ($B) 30.4 23.4 21.4 27.7 30.8 15.5 25.1 37.4 ~95*
Gross Margin 58.9% 45.7% 30.6% 37.6% 45.2% -9.1% 22.4% 39.8% 75.0%*
Op Margin 49.3% 31.5% 14.0% 22.7% 31.5% -37.0% 5.2% 26.4% ~68%*
Net Margin 46.5% 27.0% 12.5% 21.2% 28.2% -37.5% 3.1% 22.8% ~58%*
R&D ($B) 2.3 2.4 2.6 2.7 3.1 3.1 3.4 3.8 ~4.5*

*FQ2'26 annualized is misleading -- it reflects the absolute peak quarter. Do not extrapolate.

10-Year Average Revenue: ~$25B 10-Year Average Operating Margin: ~15-20% (including negative years) Cycle-Normalized EPS (FY2017-FY2025 average): ~$4.50/share Peak-to-Trough Revenue Swing: 2x (from $30.8B to $15.5B)

2.2 Revenue by Segment (Current)

  • DRAM: ~75-78% of revenue (dominant). HBM is a subset of DRAM.
  • NAND: ~22-25% of revenue. Data center SSDs growing rapidly.
  • HBM specifically: ~$5-6B quarterly run rate (implied from $2B in FQ4 2025 growing rapidly)
  • Data Center: 56%+ of total revenue (up from ~30% historically)

2.3 Balance Sheet Strength

Metric FY2021 FY2022 FY2023 FY2024 FQ2'26
Total Assets ($B) 58.8 66.3 64.3 69.4 ~83
Total Equity ($B) 43.9 49.9 44.1 45.1 ~54
Long-Term Debt ($B) 6.6 6.8 13.1 11.2 11.5
Cash & ST Inv ($B) 7.8 8.3 8.6 7.0 9.6
Net Debt ($B) -1.2 -1.5 4.5 4.2 1.9
Debt/Equity 0.17 0.15 0.32 0.31 0.28
Book Value/Share $38.5 $44.5 $40.4 $40.3 $64.24

Assessment: Balance sheet is solid but not fortress-grade. Net debt is modest relative to cash flows at peak. However, during the FY2023 trough, Micron had to issue $5B in new debt while burning cash. The balance sheet strength is cyclical -- it deteriorates when the cycle turns.

2.4 Cash Flow Analysis

Metric FY2019 FY2020 FY2021 FY2022 FY2023 FY2024 FY2025
OCF ($B) 13.2 8.3 12.5 15.2 1.6 8.5 17.5
CapEx ($B) 9.9 8.2 10.0 12.1 7.7 8.4 15.9
FCF ($B) 3.3 0.1 2.4 3.1 -6.1 0.1 1.7
Dividends ($B) 0.2 0.2 0.5 0.5 0.5 0.5 0.5
Buybacks ($B) 2.7 0.2 1.2 2.4 0.4 0.3 0

Critical insight: Free cash flow is anemic even in good years because capex intensity is enormous. Over FY2019-FY2025 (7 years), cumulative OCF was ~$77B but cumulative capex was ~$72B, leaving only ~$5B of cumulative FCF -- less than 1% of current market cap per year.

Capex/Revenue ratio: Typically 30-45% of revenue. This is among the highest in all of semis. Memory is a capex treadmill.

2.5 Dividend History

  • Micron initiated dividends in Q4 FY2021 ($0.10/quarter)
  • Raised to $0.115/quarter in 2022
  • Raised to $0.15/quarter in Q1 FY2026
  • Annual dividend: ~$0.60/share = 0.12% yield at $487
  • Payout ratio: ~3% of earnings -- token dividend, not a capital return story
  • The dividend is insignificant to the investment thesis

2.6 EPS History and Cyclicality (Full 20+ Year View)

Year EPS Year EPS
FY1998 -$0.59 FY2012 -$0.98
FY1999 -$0.14 FY2013 -$0.14
FY2000 $2.56 FY2014 $3.23
FY2001 -$0.21 FY2015 $2.69
FY2002 -$0.80 FY2016 $0.06
FY2003 -$1.90 FY2017 $4.86
FY2004 $0.20 FY2018 $11.95
FY2005 $0.27 FY2019 $6.29
FY2006 $0.27 FY2020 $2.83
FY2007 -$0.32 FY2021 $6.06
FY2008 -$1.22 FY2022 $8.34
FY2009 -$2.00 FY2023 -$4.45
FY2010 $1.29 FY2024 $1.27
FY2011 $0.15 FY2025 $8.29
TTM $21.21

Over 28 years: 11 years of negative EPS. Average EPS ~$2.50/share (including losses). This is a business that destroys capital as often as it creates it. The current $21 EPS is 8x the long-term average.

Phase 3: Moat Assessment

3.1 Moat Type: NARROW (Oligopoly + Technology, Not Structural)

The bull case for a moat:

  • DRAM is a 3-player oligopoly: Samsung (40%), SK Hynix (30%), Micron (~25%)
  • Barriers to entry are enormous ($15-20B per fab, 5+ years to ramp)
  • No new DRAM entrant has succeeded in 25+ years (CXMT in China is trying)
  • Technology leadership in HBM provides temporary pricing power
  • Micron leads in 1-gamma DRAM node (4 consecutive node leadership)

Why the moat is NARROW, not WIDE:

  • Oligopoly provides supply discipline... sometimes. But all three players aggressively add capacity during upcycles, causing the next downturn
  • Products are fungible commodities -- zero switching costs for customers
  • Technology leadership rotates -- Samsung led in the 2010s, Micron leads now in DRAM, SK Hynix leads in HBM
  • HBM differentiation is TEMPORARY -- Samsung has qualified HBM3E with NVIDIA, closing the gap
  • Margins revert to commodity levels over full cycles (10-year average operating margin ~15-20%)
  • No recurring revenue, subscription, or lock-in mechanism
  • Customer concentration in hyperscalers gives buyers enormous bargaining power

3.2 HBM: Structural Shift or Cyclical Superpeak?

Arguments for structural shift:

  • AI training/inference requires exponentially more memory bandwidth
  • HBM TAM growing from $35B (2025) to $100B (2028) -- 40% CAGR
  • HBM has higher barriers (advanced packaging, thermal management, CMOS logic die)
  • Customized HBM4E creates some switching costs with hyperscalers
  • 3:1 trade ratio with DDR5 constrains conventional DRAM supply

Arguments for cyclical superpeak:

  • Samsung and SK Hynix are rapidly expanding HBM capacity
  • When all three players are at full HBM production, pricing will normalize
  • HBM margins of 3-5x regular DRAM are unsustainable as competition catches up
  • AI capex cycles are not immune to downturns (see 2000-2002 telecom bust analogy)
  • Micron's own history: margins ALWAYS revert to mean

My assessment: BOTH. HBM is a genuine structural shift that elevates Micron's mid-cycle earnings. But the current margin levels (75% gross) are a cyclical superpeak that will not persist. Mid-cycle HBM margins will likely settle at 1.5-2x regular DRAM, not 3-5x.

3.3 Competitive Position

Metric Samsung SK Hynix Micron
DRAM Share ~40% ~30% ~25%
NAND Share ~33% ~18% ~12%
HBM Share ~25% ~50% ~25%
Technology Node 1b 1b 1-gamma (leading)
Vertical Integration Highest High Moderate
Diversification Phones, displays, foundry SK Group Pure-play memory

Micron's pure-play nature is both a strength (focused execution) and weakness (no diversification buffer in downturns).

Phase 4: Synthesis and Valuation

4.1 Cycle-Adjusted Valuation

The fundamental problem: You cannot value Micron on TTM earnings. A P/E of 23x on $21 EPS assumes peak earnings persist. They will not.

Approach 1: Cycle-Normalized Earnings

  • 10-year average EPS (FY2016-FY2025): ~$4.30
  • Generous mid-cycle EPS (assuming HBM structurally elevates earnings): ~$8-10
  • At $487: P/E on 10-year average = 113x. On generous mid-cycle = 49-61x
  • This is extremely expensive for a cyclical commodity producer

Approach 2: Book Value Floor

  • Current book value: $64.24/share
  • P/B ratio: 7.6x -- near all-time highs
  • Historical P/B range: 1.0x (trough 2023) to 3.5x (normal peak) to 7.6x (current)
  • At 2x book (reasonable upswing value): $128
  • At 3x book (generous): $193
  • At 1x book (trough): ~$64

Approach 3: Revenue Multiple

  • TTM revenue: $58B; Forward FY2026E revenue: ~$74B
  • P/S on TTM: 9.5x; P/S on forward: 7.4x
  • Historical average P/S: 2-4x
  • At 3x normalized revenue ($25B mid-cycle): $67/share
  • At 4x peak revenue ($74B): $263/share

Approach 4: EV/EBITDA

  • Current EV/EBITDA: 13.6x on TTM EBITDA of $36.8B
  • Mid-cycle EBITDA estimate: $12-15B
  • At 8x mid-cycle EBITDA: EV = $96-120B, equity value = ~$75-95/share

4.2 What Price Would I Pay?

I need to buy Micron cheaply enough that even a cyclical downturn cannot permanently impair the investment.

Strong Buy: $175 (~2.7x book, ~20x generous mid-cycle EPS of $8-10)

  • This requires a 64% decline from current levels
  • Historically, MU declines 50-70% peak-to-trough every cycle
  • At $175, you would own a technology leader at a price that provides margin of safety through the cycle
  • Downside to trough book (~$65) would be ~63%, but book value is growing

Accumulate: $250 (~3.9x book, ~28x mid-cycle EPS)

  • This requires a 49% decline from current
  • Reasonable if HBM structurally elevates mid-cycle earnings to $10+
  • Still significant downside risk in a severe downturn

Current Price ($487): WAIT

  • At $487, you are paying 7.6x book for a cyclical commodity producer at peak margins
  • The last time MU traded at 7x+ book was... never before this cycle
  • Forward P/E of 7.7x looks cheap, but that assumes FY2027 earnings of ~$63, which requires 75%+ gross margins to persist -- they will not

4.3 When Will the Cycle Turn?

Predicting exact timing is impossible, but the ingredients for a downturn are assembling:

  • Massive capacity additions: $20-25B capex in FY2026 alone, plus Samsung and SK Hynix investing similarly
  • New fabs coming online: Idaho (2027), New York (2030), Singapore HBM (2027)
  • AI capex concentration risk: If hyperscaler AI spending slows even 10-20%, HBM demand drops sharply
  • HBM trade ratio works in reverse: When HBM demand stabilizes, the freed DDR5 capacity floods the market
  • Historical pattern: Upcycles last 2-3 years. This one started in H2 FY2024 (1.5 years ago)

Best guess: The cycle likely has 6-18 months of strength remaining (through early 2027), followed by margin compression as Samsung HBM4 volumes normalize pricing and AI capex growth decelerates. A meaningful downturn by FY2028 (H2 2027 to H1 2028) would be consistent with historical patterns.

4.4 Scenario Analysis

Scenario Probability FY2027 EPS Fair Value Return from $487
Bull: AI supercycle extends to 2028 25% $40+ $500-600 +3% to +23%
Base: Normal cycle downturn 2027-28 50% $15-20 $200-300 -38% to -59%
Bear: Sharp downturn + HBM oversupply 25% $0-5 $80-150 -69% to -84%
Expected Value ~$18 ~$250 -49%

Conclusion

Micron Technology is executing brilliantly. The company has legitimate technology leadership in 1-gamma DRAM, a growing HBM franchise, and is riding the most powerful demand wave in memory history. Sanjay Mehrotra's team deserves credit for positioning Micron as a tier-1 AI enabler.

But execution and valuation are different questions. At $487, Micron is priced for a permanent structural shift that has never occurred in the memory industry. Every prior "this time is different" moment in memory -- the PC boom (late 1990s), the smartphone boom (2017-2018), the data center buildout (2021-2022) -- was followed by a savage downturn.

HBM does represent a genuine structural improvement to Micron's mid-cycle earnings power. But 75% gross margins are not the new normal. They are the peak of a cycle that will, as it always has, revert.

The correct strategy is patience. Wait for the inevitable downturn, then buy aggressively when the market gives up on Micron (as it did at $49 in December 2022 and $73 in April 2025). The downside risk of buying at $487 vastly outweighs the upside potential.

Key Data Sources

  • AlphaVantage: Financial statements, company overview, earnings transcripts, dividends
  • Micron Investor Relations: FQ1 and FQ2 2026 earnings releases and prepared remarks
  • Company filings: 10-K annual reports (SEC EDGAR)

Analysis completed April 15, 2026. All financial data verified against primary sources.

=== VERDICT: MU | WAIT | SB:$175 | Acc:$250 | Current:$487 ===