Back to Portfolio
SNDK

SanDisk Corporation

$603.17 USD 100.1B market cap March 27, 2026
SanDisk Corporation SNDK BUFFETT / MUNGER / KLARMAN SUMMARY
1 SNAPSHOT
Price$603.17
Market CapUSD 100.1B
EVUSD 99.1B
Net DebtUSD -0.96B
Shares147.6M
2 BUSINESS

SanDisk develops, manufactures, and sells NAND flash memory storage products including SSDs, embedded storage, memory cards, and USB drives. Spun off from Western Digital in February 2025, it serves three segments: Edge (PC/Mobile, 56%), Consumer (30%), and Datacenter (15%). Revenue is driven by the NAND flash cycle, currently in an AI-fueled supercycle with explosive pricing and demand growth.

Revenue: USD 7.4B (FY2025); Q2 FY26 annualized ~$12.1B Organic Growth: 61.2% YoY (Q2 FY26)
3 MOAT NARROW

Limited moat in a commodity business. Sources include: (1) Scale as #3-4 global NAND producer, (2) Kioxia Flash Ventures JV providing shared R&D and fab costs, (3) BiCS 3D NAND technology architecture, (4) Enterprise SSD Stargate controller platform. However, NAND is fundamentally a commodity with no pricing power, minimal switching costs, and technology leads measured in months, not years. The SanDisk consumer brand provides some retail differentiation but cannot overcome commodity dynamics in enterprise. This is NOT a Buffett business.

4 MANAGEMENT
CEO: David V. Goeckeler (since Feb 2025)

Excellent capital allocation in post-spinoff period. Paid down $1,246M of long-term debt in just 6 months (from $1,829M to $583M). No dividends, no share buybacks yet. CapEx disciplined at $39M in Q2 FY26 (1.3% of revenue) while leveraging Kioxia JV for manufacturing. Maintaining R&D investment for BiCS8/BiCS10 transitions. Clean balance sheet transformation from leveraged spinoff to near-net-cash in under a year.

5 ECONOMICS
35.2% (Q2 FY26 peak cycle; mid-cycle est. 5-8%) Op Margin
~30% (Q2 FY26 annualized; mid-cycle est. 3-6%) ROIC
USD 0.98B (Q2 FY26); FY2025 full year ($0.12B) FCF
Net cash position Debt/EBITDA
6 VALUATION
FCF/ShareUSD 6.64 (Q2 FY26 quarterly; mid-cycle ~$3-5 annual)
FCF Yield1.1% (on Q2 FY26 quarterly; mid-cycle ~0.5-0.8% annual)
DCF RangeUSD 140 - 520

Base case DCF: $310. Assumes mid-cycle revenue $14B declining to $10B through cycle, average operating margin 15% across full cycle, 12% discount rate, 3% terminal growth. Bull case ($520) assumes sustained AI demand keeps mid-cycle revenue at $18B with 25% margins. Bear case ($140) reflects severe downcycle with $10B revenue and 8% margins. All scenarios reflect NAND's extreme cyclicality -- straight-line peak extrapolation is the most dangerous assumption possible.

7 MUNGER INVERSION -45%
Kill Event Severity P() E[Loss]
NAND cycle turns: contract prices collapse 50%+ within 12-18 months -65% 70% -45.5%
Chinese NAND (YMTC) capacity floods global market -40% 45% -18.0%
Customer demand destruction at current elevated prices -35% 55% -19.3%
Samsung/SK Hynix announce major capacity expansions -45% 40% -18.0%
AI demand growth slows or training efficiency reduces storage needs -50% 25% -12.5%

Tail Risk: In a severe downcycle (comparable to 2022-2023), SNDK could see revenue fall 40%+ and return to operating losses of $1-2B annually. The $5B goodwill would face impairment risk. Combined with multiple compression from 11x to 3x P/S, the stock could revisit $100-150 (75-80% downside from current levels). This is not a tail risk but the most likely outcome within 2-3 years based on historical NAND cycle patterns. Every prior supercycle has ended this way.

8 KLARMAN LENS
Downside Case

In a bear scenario, NAND contract prices peak in mid-2026 and begin declining by Q4 2026 as Samsung and SK Hynix bring deferred capacity online. Revenue falls from a peak of ~$18B annualized to $8-10B within 18 months. Gross margins collapse from 65% to 15%. Net income swings from +$2B/quarter to -$500M/quarter. The stock drops 70%+ as multiples compress and earnings evaporate simultaneously.

Why Market Wrong

At the spinoff ($48), the market was clearly wrong -- mispricing a viable business due to spinoff mechanics, sector hatred, and zero earnings history. That mispricing has been corrected 12x over. Today, the market MAY be wrong if AI fundamentally changes NAND demand to be less cyclical (secular growth vs. cyclical). Long-term supply agreements (LTAs) could also smooth revenue and margins. But this is an unproven thesis in a 40-year commodity cycle history.

Why Market Right

The market may be right that this is peak cycle. NAND contract prices up 85-90% in a single quarter is historically unprecedented and unsustainable. No commodity has ever maintained 65% gross margins for more than a few quarters. The AI demand thesis, while real, does not repeal the laws of supply and demand -- capacity will eventually catch up. Current $100B market cap implies the market already values SNDK at 2-3x its peers, and peer multiples already reflect the same AI tailwinds.

Catalysts

Downside catalysts (most likely): NAND price declines, capacity announcements, AI spending slowdown, inventory buildup at customers. Upside catalysts: Kioxia merger approval, sustained NAND shortage through 2027, AI demand acceleration beyond current forecasts, BiCS10 technology lead.

9 VERDICT REJECT
B- Rejected
Strong Buy$175
Buy$300
Sell$750

SanDisk is a well-managed commodity NAND business riding an extraordinary AI-driven supercycle. At $603, the stock prices in years of peak-cycle earnings that NAND history suggests are unsustainable. Mid-cycle normalized earnings support a fair value of ~$310. Value investors should wait for the inevitable cyclical downturn, targeting entry at $150-250 during the next NAND trough (likely 2027-2028). The Aschenbrenner/momentum thesis was correct at $48-237 but does not justify buying at $603 with a value investing framework. Watchlist candidate for cyclical entry.

🧠 ULTRATHINK Deep Philosophical Analysis

SNDK - Ultrathink Analysis

The Real Question

The real question is not whether AI will drive massive NAND demand -- it will. The real question is whether this time is truly different for memory semiconductors, or whether we are watching the same movie we have seen a dozen times before, just with a new protagonist called "AI."

Every NAND supercycle in history has been accompanied by a compelling narrative for why margins will stay elevated: smartphones in 2017, cloud computing in 2018, 5G in 2021. Each time, the narrative was correct about demand but wrong about sustainability. Demand did grow. Capacity grew faster. Margins collapsed.

The question for SanDisk is not about the next two quarters -- Q3 FY26 guidance of $12-14 EPS with 65-67% gross margins is almost certainly achievable. The question is about 2027, 2028, and 2029. Will SNDK still be earning $40+ per share, or will it be losing money again as it was just 18 months ago?

Hidden Assumptions

The market at $603 is making several assumptions that deserve scrutiny:

Assumption 1: "No new NAND capacity outside China through 2027." This is stated as fact, but it is actually a decision that can be reversed. Samsung alone has the ability to bring significant capacity online with 12-18 months notice. When NAND prices are up 85-90% in a single quarter, the economic incentive to add capacity becomes overwhelming. The question is not if but when.

Assumption 2: "AI demand fundamentally changes NAND economics." The implicit belief is that AI inference storage needs will create a permanent demand floor that prevents the brutal downturns of past cycles. But AI workloads are currently <15% of SNDK's revenue (Datacenter segment). The Edge and Consumer segments -- 85% of revenue -- are classic cyclical markets with no structural change.

Assumption 3: "Long-term agreements will smooth the cycle." SNDK has signed its first LTA with prepayment. This is positive but unproven. LTAs in commodity markets often include price adjustment mechanisms that still expose producers to spot market dynamics. The natural gas industry learned this painfully -- LTAs did not prevent the 2023 price collapse.

Assumption 4: "$100B is justified because forward earnings are $40+ per share." This assumes peak-cycle earnings are a valid basis for valuation. By this logic, oil companies should have been valued at 2022 peak earnings forever. They weren't. Every commodity cycle teaches the same lesson: buy when earnings are terrible and the narrative is bearish; sell when earnings are spectacular and the narrative is bullish.

The Contrarian View

For the bears to be right, only one thing needs to happen: history needs to repeat. That is the base case, not the contrarian view. The truly contrarian position here would be that NAND has become a permanently tight market -- that would require accepting that all four major producers have simultaneously and permanently abandoned the expansion behavior that has defined their industry for 40 years.

Here is what would make the bears right (and what I believe is most likely):

  • Samsung announces fab expansion by Q4 2026
  • NAND contract prices peak in mid-2026
  • By early 2027, spot prices are declining 5-10% per quarter
  • By late 2027, SNDK gross margins are back to 25-30%
  • By 2028, operating losses return as the cycle fully reverses
  • The stock trades at $150-250 during the trough

This is not pessimism. This is the NAND cycle, observed consistently across four decades.

Simplest Thesis

SanDisk is a commodity business priced for secular growth, and that mispricing will correct when the NAND cycle inevitably turns.

Why This Opportunity Exists

The opportunity does not exist today at $603. It existed at $48 in February 2025, when the spinoff mechanics, institutional dumping, and sector hatred created a genuine mispricing of a viable business. That opportunity is gone.

The future opportunity will exist when the NAND cycle turns and the stock falls 60-80% from its peak. That is when the narrative will be maximally bearish: "NAND is a terrible business," "AI storage demand was overhyped," "SanDisk can't compete with Samsung." All of those statements will be wrong at trough prices, just as all the bullish statements are partially wrong at peak prices.

The deeper truth is that Mr. Market treats NAND companies as growth stocks during upturns and value traps during downturns. Neither characterization is correct. NAND companies are cyclical cash flow generators that should be bought at trough and sold at peak. The challenge is that human psychology makes this almost impossible -- buying feels terrible at the trough (because the company is losing money) and selling feels terrible at the peak (because the company is making record profits).

A value investor's edge here is purely behavioral: the willingness to buy when headlines scream disaster and sell when headlines scream opportunity.

What Would Change My Mind

I would consider buying SNDK at current prices only if ALL of the following were true:

  1. LTA contracts covering 50%+ of revenue with fixed prices and 3+ year terms, demonstrably smoothing revenue through a cycle. (Currently: one LTA signed, covering a small fraction of revenue.)

  2. Datacenter becoming 40%+ of revenue with demonstrated pricing power distinct from commodity NAND. (Currently: 15% of revenue.)

  3. Kioxia merger completed with regulatory approval, creating clear #1 global NAND position with cost synergies. (Currently: uncertain.)

  4. Two consecutive down-quarter earnings beats during a NAND price decline, proving the business model has structural resilience. (Currently: untested -- the company has only existed during an upcycle.)

  5. Valuation at or below 15x mid-cycle earnings (not peak earnings), implying a stock price of $45-75. (Currently: $603, or 120-200x mid-cycle earnings.)

None of these conditions exist today.

The Soul of This Business

At its core, SanDisk makes a commodity that the world increasingly needs. That is both its salvation and its curse. The world will always need more storage -- this is as close to a guaranteed secular trend as exists in technology. But "the world needs more of something" has never been sufficient to create durable competitive advantage. The world needs more wheat, more aluminum, more DRAM, and more NAND. The producers of these commodities earn their cost of capital only at the peaks.

What makes SanDisk's position fragile is not its technology, its management, or its markets. It is the fundamental economics of commodity production: when prices rise enough, supply will follow, and when supply arrives, prices will fall. This is not a bug in the system. It is the system. No management team, however brilliant, can repeal the economics of commodity cycles. They can manage through them -- and Goeckeler's debt paydown suggests he understands this -- but they cannot prevent them.

The soul of SanDisk is a company that will make extraordinary money for 1-2 years, then lose money for 1-2 years, then repeat. The art of investing in it is recognizing where you are in the cycle and acting accordingly. Right now, we are at the peak. The time to act is not now. It will come.

Executive Summary

Three-Sentence Thesis

SanDisk is a pure-play NAND flash memory company riding an unprecedented AI-driven supercycle that has transformed it from a loss-making WDC spinoff into a $100B profit machine in just 13 months. The company benefits from a structural supply shortage (no new NAND capacity outside China through 2027), explosive AI inference storage demand, and a potential Kioxia merger that would create the world's largest NAND manufacturer. However, at $603 with a trailing 12-month loss and forward P/E of ~15x on peak-cycle earnings guidance, the stock already prices in perfection, and NAND's brutal cyclicality means current margins are unsustainable, making this a classic late-cycle momentum trap for value investors.

Key Metrics Dashboard

Metric Value Assessment
Price $603.17 Up 1,141% from $48.60 spinoff price
Market Cap $100.1B
Forward P/E (Q3 FY26 guide) ~15x On peak-cycle $40+ EPS
P/B 10.15x Elevated
P/S (TTM) 11.2x Very high for hardware
Net Debt Net cash (~$960M) Strong
Gross Margin (Q2 FY26) 50.9% Peak cycle; guide 65-67% Q3
Operating Margin (Q2 FY26) 35.2% Extraordinary for NAND
FCF (Q2 FY26 annualized) ~$3.9B Strong but cyclical
ROE (TTM) -9.4% Distorted by loss quarters
Insider Ownership 5.5% Moderate

Decision

REJECT at current price. WAIT for cyclical downturn.

  • This is a deeply cyclical commodity business experiencing peak pricing
  • At $603, the stock prices in 2+ years of supercycle earnings
  • NAND cycles historically mean 50-80% drawdowns from peak
  • Fair value range: $200-350 (mid-cycle normalized earnings)
  • Strong Buy: $150-200 (trough/early recovery)
  • Position size: 0% now; 3-5% at strong buy levels

Phase 0: Why Does This Opportunity Supposedly Exist?

The Aschenbrenner Thesis

Leopold Aschenbrenner's Situational Awareness LP built a $250M position (5.9% of portfolio) in Q4 2025, increasing holdings by 816%. His thesis centers on SNDK as critical AI infrastructure -- NAND flash storage is the "last mile" for AI inference, where models must be stored on fast, high-density SSDs. This is part of his broader bet on the physical layer of AGI: data centers, power, networking, and now storage.

Why The Market Was "Wrong" (Past Tense)

The market mispriced SNDK for several valid reasons that have now largely corrected:

  1. Spinoff orphan: At $48.60, institutional holders of WDC received SNDK shares and dumped them (spinoff discount)
  2. No track record: New company, no earnings history, NAND downcycle losses
  3. Sector hatred: Memory/storage was deeply out of favor in 2024-early 2025
  4. Index exclusion: Not in S&P 500 initially, limiting passive buying

All four of these have since reversed -- SNDK is now in the S&P 500, posting record margins, and the NAND cycle has turned violently upward. The 1,141% move has largely closed the mispricing gap.

Phase 1: Risk Analysis (Inversion -- What Could Destroy This Investment?)

The Fundamental Problem: NAND Is a Commodity

Before analyzing SNDK as an investment, we must confront the most important fact: NAND flash memory is a commodity. Unlike software or branded consumer goods, one company's NAND bit is functionally interchangeable with another's. This is the single most important fact about this business.

Top 10 Risks

# Risk Event Severity Likelihood (3yr) Expected Loss
1 NAND cycle turns: prices collapse 50%+ -65% 70% -45.5%
2 China capacity flooding market (YMTC expansion) -40% 45% -18.0%
3 Customer pushback/demand destruction at high prices -35% 55% -19.3%
4 Samsung/SK Hynix aggressive capacity expansion -45% 40% -18.0%
5 AI demand slowdown / training efficiency gains -50% 25% -12.5%
6 Technology disruption (CXL memory, new architectures) -30% 15% -4.5%
7 Kioxia merger blocked by regulators -20% 35% -7.0%
8 Execution failure on BiCS10 node transition -25% 20% -5.0%
9 Tariff/trade war disrupting supply chains -20% 30% -6.0%
10 Loss of key customer concentration -15% 15% -2.3%

Total Expected Downside: Sum of top 4 non-overlapping risks = ~45-55%

Critical Risk Deep Dive: The NAND Cycle

NAND flash is the most violently cyclical industry in technology. Historical evidence:

  • 2018-2019 downcycle: NAND prices fell 50%+, Samsung margins went from 50%+ to single digits
  • 2022-2023 downcycle: WDC/SNDK flash segment went from $3.2B gross profit to $430M (87% decline)
  • 2024 recovery: Prices recovered but from deeply depressed levels
  • 2025-2026 supercycle: Current environment with 85-90% price increases

The pattern is ALWAYS the same:

  1. Shortage develops (demand > supply)
  2. Prices spike, margins explode
  3. Producers invest in new capacity
  4. Supply catches up/overshoots
  5. Prices collapse, margins evaporate
  6. Repeat every 3-5 years

Current cycle position: We are firmly in stages 2-3. SNDK guided Q3 FY26 gross margins of 65-67% -- these are extraordinary, unsustainable margins for a commodity business.

Tail Risk: The Chinese NAND Threat

YMTC (Yangtze Memory Technologies Corp) is aggressively expanding NAND capacity in China, currently at 128-layer and developing 232-layer technology. If Western sanctions are loosened, Chinese NAND could flood global markets. This is a non-trivial risk that could permanently impair Western NAND makers' pricing power.

Phase 2: Financial Analysis

Revenue Trajectory

Period Revenue QoQ YoY Gross Margin
FY2022 $9,754M - - 33.3%
FY2023 $6,086M - -37.6% 7.1%
FY2024 $6,663M - +9.5% 16.1%
FY2025 $7,355M - +10.4% 30.1%
Q1 FY26 (Sep25) $2,308M +21.4% +22.6% 29.8%
Q2 FY26 (Dec25) $3,025M +31.1% +61.2% 50.9%
Q3 FY26 Guide $4,600M* +52.1% +142%* 66.0%*

*Midpoint of guidance range

Profitability Analysis

SNDK's current profitability is extraordinary but must be viewed through the cycle lens:

Q2 FY26 (Dec 2025) -- Current Quarter:

  • Revenue: $3,025M
  • Gross Margin: 50.9% (vs. 7.1% in FY2023)
  • Operating Margin: 35.2% (vs. -33.4% in FY2023)
  • Net Income: $803M (vs. ($2,143M) in FY2023)
  • Non-GAAP EPS: $6.20

Q3 FY26 Guidance (stunning):

  • Revenue: $4.4-4.8B
  • Gross Margin: 65-67% (historically unheard of for NAND)
  • Non-GAAP EPS: $12-14

Normalized Mid-Cycle Estimate: Using average of FY2022 (peak) and FY2024 (trough) as guide:

  • Mid-cycle revenue: ~$8-9B
  • Mid-cycle gross margin: ~25%
  • Mid-cycle operating margin: ~5-8%
  • Mid-cycle net income: ~$400-700M
  • Mid-cycle EPS: ~$3-5

This means current earnings are 10-15x above mid-cycle levels.

Balance Sheet Assessment

Strengths:

  • Net cash position (~$960M net cash as of Dec 2025)
  • Rapid debt paydown: LT debt from $1,829M to $583M in 6 months
  • $1,539M cash on hand
  • Book value: $69.19/share ($10.2B equity)

Concerns:

  • $5.0B goodwill (48.9% of equity) -- from WDC's original SanDisk acquisition
  • Goodwill is at risk of impairment in any downcycle
  • Previously wrote down ~$2.2B of goodwill (FY2024 vs FY2022 levels)

Cash Flow Analysis

Q2 FY26 operating cash flow of $1,019M is exceptional. Annualized FCF of ~$3.9B at current margins. However:

  • FY2025 full-year OCF was only $84M
  • FY2024 OCF was negative ($309M)
  • FY2023 OCF was negative ($713M)
  • Only FY2022 ($1,151M) showed strong OCF in recent history

This is a business that generates zero to negative FCF for 2-3 years of every cycle, then generates massive FCF for 1-2 years.

Owner Earnings Calculation

Using Q2 FY26 annualized (peak cycle):

  • Net Income: $3,212M (annualized)
  • Add back: D&A $152M, SBC $232M
  • Less: Maintenance CapEx ~$200M
  • Owner Earnings: ~$3,400M
  • Owner Earnings Yield: 3.4% at $100B market cap

Using mid-cycle normalized:

  • Net Income: ~$500M
  • Add back: D&A ~$200M, SBC ~$180M
  • Less: Maintenance CapEx ~$200M
  • Owner Earnings: ~$680M
  • Owner Earnings Yield: 0.68% at $100B market cap

The stock is priced for perpetual peak earnings. Mid-cycle returns are unacceptable.

Valuation

DCF (10-year, three scenarios):

Assumption Bull Base Bear
FY27 Revenue $18B $14B $10B
Terminal Growth 4% 3% 2%
Avg Operating Margin 25% 15% 8%
Discount Rate 12% 12% 12%
Fair Value/Share $520 $310 $140

Key assumption: NAND is cyclical. No straight-line extrapolation of peak margins is valid.

Relative Valuation:

Metric SNDK Micron (MU) Samsung SK Hynix
Forward P/E ~15x ~12x ~10x ~8x
P/B 10.2x 3.2x 1.5x 2.8x
P/S (TTM) 11.2x 4.5x 1.8x 3.5x

SNDK trades at a 2-3x premium to peers on every valuation metric. This premium reflects the "pure-play NAND" story and AI narrative momentum, but is not justified by fundamentals.

Phase 3: Moat Analysis

Moat Rating: NARROW (at best)

The uncomfortable truth: NAND flash is a commodity business with no durable competitive advantage.

Moat Sources Assessment

Moat Source Present? Strength Evidence
Brand Weak Low Consumer SanDisk brand has value but limited pricing power
Switching Costs Minimal Low SSDs are largely standardized; customers can switch suppliers
Network Effects None None Not applicable
Cost Advantage Moderate Medium Scale + Kioxia JV provides some cost advantages
IP/Technology Moderate Medium BiCS technology, 3D NAND patents, but competitors have similar tech
Regulatory None None No regulatory barriers to entry

Detailed Assessment

What SNDK has:

  1. Scale: #3 or #4 NAND producer globally (behind Samsung, SK Hynix, and level with Micron)
  2. Kioxia Partnership: Flash Ventures JV provides shared fab costs, technology development
  3. BiCS Technology: Proprietary 3D NAND architecture (BiCS8 ramping, BiCS10 in development)
  4. Enterprise SSD Platform: Stargate controller for AI/datacenter applications
  5. Consumer Brand: SanDisk brand recognition in retail storage

What SNDK lacks:

  1. Pricing power: Cannot set prices above market rates for commodity NAND
  2. Differentiation: Enterprise SSDs are increasingly standardized
  3. Vertical integration: Unlike Samsung (which makes its own controllers, DRAM, and foundry), SNDK is pure NAND
  4. Customer lock-in: Hyperscalers qualify multiple suppliers; no single-source dependency

Moat Duration: 5-10 years

The technology moat (BiCS architecture) provides some differentiation, but competitors are within 6-12 months on each node generation. The Kioxia partnership is the most durable advantage but could be disrupted by merger failure or partner conflict.

The Buffett Test: Would Buffett Own This?

Almost certainly not. Buffett has explicitly avoided commodity technology businesses throughout his career. He would note:

  1. No pricing power (commodity)
  2. Massive cyclicality (earnings swing from +$1B to -$2B)
  3. Capital-intensive ($200-500M+ annual CapEx)
  4. Technology risk (must keep up with node transitions)
  5. No dividends

This is the antithesis of a Buffett business.

Phase 4: Decision Synthesis

The Core Dilemma

SNDK is a well-run company in a terrible industry at a terrible price. The AI demand thesis has merit, but:

  1. The cycle always turns. No NAND supercycle has lasted more than 18-24 months historically.
  2. Capacity will come. Samsung, SK Hynix, and YMTC will expand. "No new capacity" is a temporary condition.
  3. Prices will normalize. 65-67% gross margins in NAND are as unsustainable as oil at $150/barrel.
  4. The stock has discounted the future. At $603 and ~$100B market cap, the market expects 2-3+ more years of supercycle earnings.

Entry Price Analysis

Scenario Price P/E Basis Probability
Strong Buy $150-200 30-40x mid-cycle ($5 EPS) Wait for downcycle
Accumulate $250-350 15-20x mid-cycle normalized Early recovery
Fair Value $310 Base case DCF
Current Price $603 15x peak EPS Overvalued
Sell Above $750+ Extreme momentum

Position Sizing

  • Current recommendation: 0% allocation
  • At strong buy ($150-200): 2-3% allocation (cyclical risk limits position)
  • Maximum position: 3-5% (commodity business ceiling)

Monitoring Triggers

Metric Threshold Action
NAND contract prices QoQ decline >15% Begin watching for entry
Gross margin Below 35% Getting closer to mid-cycle
Stock price Below $300 Start building position
Stock price Below $200 Aggressive accumulate
China NAND capacity >15% global share Reassess thesis entirely
Kioxia merger Approved/blocked Adjust fair value

Leopold Aschenbrenner / Situational Awareness Context

Aschenbrenner's 816% increase in SNDK position (Q4 2025) was well-timed from a momentum perspective -- the stock subsequently rose from ~$237 to $603+. His thesis that NAND storage is critical AI infrastructure is directionally correct. However:

  1. His time horizon may be shorter than a value investor's (hedge fund vs. long-term hold)
  2. His thesis is about demand, not competitive advantage or valuation
  3. Momentum investing works until it doesn't -- the NAND cycle will turn
  4. $250M at $237 average is very different from buying at $603

His position validates the AI storage demand narrative but does not make SNDK a value investment at current prices.

Conclusion

SanDisk is experiencing a once-in-a-decade supercycle driven by AI demand and constrained NAND supply. The company is executing well, generating enormous cash flow, and paying down debt rapidly. However, this is a deeply cyclical commodity business trading at peak margins with a $100B market cap that implies years of continued supercycle conditions.

For value investors: REJECT at current prices. Add to watchlist for the inevitable cyclical downturn. Target entry: $150-250 range during the next NAND downcycle (likely 2027-2028).

For momentum/growth investors: May have further upside if Q3 FY26 guidance is met/beaten, but risk/reward is increasingly unfavorable at $600+.

Key insight: The best time to buy NAND stocks is when everyone hates them and they're losing money. That was February 2025 at $48. The worst time to buy is when margins are at records and the narrative is most compelling. That is now.

Analysis based on: AlphaVantage financial data, SEC EDGAR filings, SanDisk IR press releases, web research on NAND market dynamics, and Situational Awareness LP 13F filings.