For information only. Not investment advice, a recommendation, an offer or a solicitation, and without regard to any recipient’s objectives or circumstances. Figures come from third-party sources believed reliable but not guaranteed, and market data may be delayed. Past performance is not a reliable indicator of future results. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Sandisk Corporation (SNDK) Deep DiveExecutive Summary (The Swiss PM Perspective)
v2.0 Re-underwrite (11/04/2026) — DRAFT for PM validationEverything in this section is an analyst proposal. The probabilities, multiples and levels are judgement calls flagged as such; the verdict stays the PM's call. 1. What the price discounts (reverse DCF)
2. NAND price check (the open item from v1.1)
3. Scenarios (12-month values)
3b. Political-risk overlay (US midterms, 11/03/2026) — added 11/04Event. Democrats won the Senate and the House. Gina Hinojosa unseated Gov. Abbott in Texas (about 5.20M vs 4.50M votes, 99% counted, Texas Secretary of State) and James Talarico won the Senate seat; Democrats also won or kept the governorships of Georgia, Ohio and Pennsylvania. Hinojosa campaigned on an emergency executive order for a statewide data-center moratorium and on halting approvals until the legislature passes rules 13. Texas holds about 27% of planned US data-center gigawatts (Wells Fargo), Georgia 7%, Ohio 5%, Pennsylvania 5%: about 44% together 13. Market reaction (11/04, intraday vs 11/03 close). Builders and equipment makers fell 16–24% (Modine, Hut 8, INNIO, Sterling, Cipher), Vertiv −12%, GE Vernova −11%, Nvidia −1.7%, while Amazon, Microsoft and Alphabet rose 5–6% (less data-center build means less capex to fund) 13. SNDK was −8.0% vs the 11/03 close but only −3.6% vs the 11/02 close, because it had rebounded 4.8% on 11/03. Why this is not simply a panic.
Why it is not nothing.
Model overlay. Weights move from 20/45/35 (draft 1) to 18/42/40; on the draft-1 share count the weighted value fell $1,248 → $1,209 (−3.2%); after the rev. 3 share-count correction it is $1,178. Sensitivity on the base case: FY28 EPS −5% → $1,228, −10% → $1,163 (at 4.5x). The election explains part of the 11/04 drop, not all: Nvidia, with the same capex exposure, fell 1.7% while SNDK fell 8.0%; the rest reads as the NAND-momentum narrative. There is no quantitative decomposition. 11/04 prices are intraday, not closes. 4. Proposed verdict and entry frameworkProposed: Accumulate on weakness, staged entry, Conviction Medium (v1.0: HOLD). No purchase above T1; at the current price the proposal is effectively wait. Levels are pre-split USD; divide by 8 from 11/20/2026.
5. Limits of this re-underwrite
v1.1 Data Refresh & Open Contradictions (11/04/2026)Status in v2.0: items 1, 2 and 4 are addressed in the next section; items 3 and 5 remain open / informational.
The peer table in Valuation Dashboard is not refreshed: it remains as of 08/14/2026 and must not be compared with the 11/03/2026 SNDK figures above. ⚠️ Contradictions raised in v1.1 (resolution notes added in v2.0):
Company Description & Business ModelSandisk Corporation (SNDK) is a global leader in the design and manufacture of vertically integrated NAND flash storage solutions. Officially spun off from Western Digital on February 21, 2025 1, the company operates as a pure-play NAND manufacturer with no DRAM exposure, leaving its top-line highly sensitive to NAND Average Selling Price (ASP) fluctuations 1. SNDK manufactures 100% of its silicon wafers through its long-standing 50/50 joint venture (JV) with Kioxia Corporation at the Yokkaichi and Kitakami fabrication plants in Japan 14. This asset-light manufacturing structure allows SNDK to maintain extremely low direct capital expenditure intensity (capex of 5.8% to 6.3% of revenue, or ~$179M–$562M TTM) 73, leveraging the scale and shared capital burdens of the JV 47. Revenue Breakdown (FY26: $20.25 Billion) 2:
Market Dynamics & Industry PositioningThe global NAND flash memory industry is undergoing a structural paradigm shift. Historically plagued by destructive boom-and-bust cycles due to irrational capacity expansions, the sector in 2026 displays unprecedented supply discipline. Industry-wide NAND capex is projected at just $22.2 Billion in 2026 7, as the major integrated memory producers (Samsung, SK Hynix, Micron) aggressively reallocate capital budgets toward advanced DRAM and High-Bandwidth Memory (HBM) for AI accelerators 7. This structural NAND tightness coincides with a surge in AI-driven storage requirements:
SNDK management indicates that demand significantly exceeds manufacturing capacity, projecting that wafer output will remain under strict contractual allocation through calendar year 2027 3. The Moat: Structural Barriers to Entry & NBMsIn addition to capital intensity and technological manufacturing hurdles, SNDK has established a commercial moat to mitigate historical sector volatility: New Business Models (NBMs) - Contractual StabilizationTo dampen cyclicality, SNDK transitioned its hyperscale and tier-1 OEM customer base to multi-year contractual agreements (NBMs).
Financial Profile & Performance TrackingFY26 Financial Summary (GAAP vs. Non-GAAP) 2:
Valuation Dashboard & Peer Analysis (As of August 14, 2026)
📈 Sell-Side Consensus Reality Check (08/14/2026 snapshot — superseded, see v1.1 refresh)A systematic pull of institutional consensus across S&P Global, TipRanks, StockAnalysis, and MarketBeat on 08/14/2026 showed strong sell-side bullishness 8. It is older than 30 days and must not be used to compute upside; current Street data (CFRA, 11/04/2026): 17 Buy, 4 Buy/Hold, 3 Hold, 1 Hold/Sell, 0 Sell; FY27 EPS $227.80 and FY28 $287.26 (20 estimates) 9:
🛠️ Adjusted 3-Year DCF Valuation Model (WACC 11.5%)Model inputs adjusted to a 11.5% WACC to reflect pure-play NAND commodity beta and mid-cycle margin normalization.
Independent Multi-LLM Review SynthesisA multi-model adversarial validation (incorporating Gemini, DeepSeek, Claude, and Perplexity intelligence) highlights the core tension in SNDK's valuation:
Catalysts & Risk Matrix
Scenarios (Bull, Base, Bear) — v2.0 draftReplaces the v1.0 triple-horizon scenarios. Full table, basis and sensitivities are in the v2.0 Re-underwrite section. 🐂 BULL CASE (Probability: 18%) — $1,980 (+75.8% vs $1,126.52)
😐 BASE CASE (Probability: 42%) — $1,293 (+14.8%)
🐻 BEAR CASE (Probability: 40%) — $696 (−38.2%)
v1.0 scenarios (08/14/2026), kept for the record: Bull $2,200 / $2,800 (30%), Base $1,600 / $1,850 (50%), Bear $780 / $950 (20%). Sourcing Integrity & Audit Ledger
Audit Footnotes:Related Documents
References (15)
|
Sandisk Corporation (SNDK) Deep Dive
Executive Summary (The Swiss PM Perspective)
⚠️ v2.0 DRAFT (11/04/2026) — proposed re-underwrite, awaiting PM validation. SNDK closed at $1,126.52 on 11/03/2026 (−31.9% vs the $1,655.01 v1.0 writing price; −52.2% vs the $2,354.39 high) and traded near $1,036.62 intraday on 11/04 (Yahoo Finance, not a close). Q1 FY27 beat (revenue $11.03B, 10/29/2026) but the group sold off on slowing NAND price momentum and Kioxia's soft guide (11/02). v2.0 replaces the v1.0 HOLD with a proposed staged-accumulation stance; the thesis paragraph below is the v1.0 text, kept for the record. See v2.0 Re-underwrite for the work and the levels. Revised the same day for the US midterm result (Democratic sweep, Texas governor-elect campaigning for a data-center moratorium): bear weight raised to 40%, entry levels lowered (section 3b). Corrected again after a blind review (rev. 3): the share count was wrong (154M diluted, not 135.8M), which lowers the bear value to $696 and the weighted value to $1,178; entry levels are now $1,000 / $820 / $700 pre-split (set by the PM on 11/04).
Investment Thesis: Sandisk Corporation (SNDK) represents the purest, most highly leveraged pure-play exposure to the AI storage supercycle. Following its spinoff from Western Digital in February 2025 1, the stock surged over 3,200% year-over-year 6 before experiencing a healthy ~36% pullback driven by premature cycle-ending fears 6. Trading at $1,655.01 on 08/14/2026 5, SNDK displayed a unique financial profile: 84.6% non-GAAP gross margin (Q4 FY26) 2, zero net debt 1, and structural multi-year revenue visibility via its New Business Model (NBM) contracts guaranteeing a $93.9 Billion revenue floor over 4 years 3.
Recommendation (v2.0 proposal — Accumulate on weakness, staged entry; v1.0 was HOLD, Neutral, Prudent Bias): the v1.0 HOLD rested on a $1,600 base case that sat 3.3% below the $1,655 price, so there was no margin of safety. At $1,126.52 the picture is reversed: the price already discounts FCF falling ~42% from FY27 while Street FY28 EPS still grows ~26%. The NAND data show a sharp deceleration of price gains (not falling prices) and the downside scenario is a 38% drawdown, so a staged entry with explicit thesis-break triggers is proposed rather than a full-size buy. Probability-weighted value is $1,178 (+4.6% vs the 11/03 close, +13.6% vs $1,036.62), a thin edge that only becomes attractive at lower entry prices, so the proposal is effectively wait above the first tranche. The v1.0 text read: "We rate SNDK HOLD" on the basis of Street FY27 EPS of $213.23 and a ~7.6x forward P/E.
| Metric | Target (12-Month) | Return Potential (vs $1,126.52) | Target (3-Year) | Return Potential (vs $1,126.52) | Source |
|---|---|---|---|---|---|
| SNDK Spot Price | $1,126.52 9 | - | - | - | 11/03/2026 close (v1.0: $1,655.01 on 08/14/2026) |
| CFRA 12-Mo Target | $2,070.00 9 | +83.8% | - | - | CFRA Strong Buy, opinion dated 10/30/2026 |
| Bull Case Target | $1,980.00 | +75.8% | n/a | - | v2.0 draft: FY28E EPS $330 × 6.0x (v1.0: $2,200 / $2,800) |
| Base Case Target | $1,293.00 | +14.8% | n/a | - | v2.0 draft: Street FY28E EPS $287.26 × 4.5x (v1.0: $1,600 / $1,850) |
| Bear Case Value | $696.00 | -38.2% | n/a | - | v2.0 draft rev. 3: refreshed 3-year DCF on 154M diluted shares, WACC 11.5% (v1.0: $780 / $950) |
| FY26 Diluted EPS | $70.88 (Non-GAAP) 2 | - | $73.76 (GAAP) 2 | - | SNDK FY26 Earnings Release |
| FY27 Street EPS | $227.80 9 | - | FY28: $287.26 9 | - | CFRA Wall Street consensus, 20 estimates, 11/04/2026 (v1.0: $213.23 on 08/14/2026) |
v2.0 Re-underwrite (11/04/2026) — DRAFT for PM validation
Everything in this section is an analyst proposal. The probabilities, multiples and levels are judgement calls flagged as such; the verdict stays the PM's call.
1. What the price discounts (reverse DCF)
- Setup: 154M diluted shares (Q1 FY27 diluted weighted average, company release 10/29/2026 10; basic shares outstanding were about 142M on 10/26 after buybacks). Rev. 1–2 of this draft used 135.8M, implied from a CFRA market cap, which was wrong. FY27 FCF ≈ $26.3B = Street FY27 EPS $227.80 × 154M × 75% cash conversion (Q1 FY27 adjusted FCF $5.7B ÷ non-GAAP net income $7.6B = 75%; Q4 FY26 was 82%), cross-checked by the Q1 adjusted FCF margin of 51.6% on Street FY27 revenue of about $49–50B (≈ $25–26B). WACC 11.5%, terminal growth 2.5% (unchanged from v1.0). Net cash / debt is ignored.
- At $1,126.52 (market cap ≈ $173B) the price is consistent with FCF settling at ≈ $15.4B/yr from FY28, i.e. −42% vs FY27. At $1,036.62 (11/04 intraday) it is ≈ $14.0B (−47%).
- Street FY28 EPS of $287.26 (+26% vs FY27) maps to ≈ $33B FCF at the same conversion. The market is pricing a cycle top; the Street is not. That gap is the whole debate.
- Sanity check: the same DCF with v1.0 inputs ($22/15/8B) gives $761 on 135.8M shares but $671 on 154M, so the v1.0 $780 carried a low share count as well.
2. NAND price check (the open item from v1.1)
- Momentum is decelerating sharply, levels are not falling. TrendForce (09/30) forecast 4Q26 NAND contract prices +15–20% QoQ and enterprise SSD +23–28%, the only category still accelerating 11. Kioxia, the JV partner producing SNDK's wafers, said its ASP rose ~20–25% QoQ last quarter after ~70% the quarter before, and guided only a low-single-digit gain for Oct–Dec 12.
- Soft spots: NAND wafer contract prices were flat in October across MLC/TLC/QLC, wafer spot prices fell sharply on frozen Chinese smartphone demand, module makers have nearly stopped buying good dies, and client SSD prices are expected to plateau in 1Q27 as seasonality and new capacity arrive 11.
- 2027 enterprise caveat: TrendForce sees enterprise SSD demand strong only in North America, with YMTC's market-share push and high prices as headwinds to further hikes 11.
- Kioxia guide (reported 10/30; the shares fell 15–16% on 11/02): Dec-quarter adjusted operating profit ≈ ¥2.0T vs ¥2.32–2.35T expected; management attributes it to order timing (pull-forward into September, deferrals into March) and says 2027 output is sold out 12. Whether this is timing or the start of a plateau is not resolvable from the sources used.
- NBM trigger (Risk 1): a >40% fall in spot/contract prices is needed to make customer default rational. Nothing above is close: the worst reading is flat wafer contracts. Not verified: the actual NAND spot index level, and SNDK's own Q2 pricing.
- Net read: the stock is trading on the second derivative of NAND pricing. The 11/04 intraday drop (SNDK −8.0% vs 11/03 close, MU −5.7%, WDC −6.3%, STX −7.5%) is sector-wide and coincides with the Democratic sweep in the US midterms (section 3b); the election does not explain all of it.
3. Scenarios (12-month values)
| Scenario | v1.0 prob. | v2.0 prob. | Value | vs $1,126.52 (11/03) | vs $1,036.62 (11/04 intraday) | Basis |
|---|---|---|---|---|---|---|
| Bull | 30% | 18% | $1,980 | +75.8% | +91.0% | FY28E EPS $330 (≈15% above Street) × 6.0x. Cycle-peak fear fades; v1.0 used 10x |
| Base | 50% | 42% | $1,293 | +14.8% | +24.7% | Street FY28E EPS $287.26 × 4.5x. Modest re-rating from 3.9x as earnings are delivered (see plateau cross-check below) |
| Bear | 20% | 40% | $696 | −38.2% | −32.9% | Refreshed DCF: FY27 FCF $26.3B (75% conversion, 154M diluted shares), FY28 $15B, FY29 $8B (v1.0 fade shape), WACC 11.5%, g 2.5% |
| Probability-weighted | $1,178 | +4.6% | +13.6% |
- Why the bear weight moves from 20% to 40% (35% in the first v2.0 draft, +5 points after the 11/03 midterms, see 3b): three independent signals of slowing price momentum arrived within a week (flat wafer contracts, Kioxia guide, TrendForce's 1Q27 plateau call), then a policy shock to the data-center build-out. The weight is a judgement and is frozen until closing prices and the next hard data point (revising it three times in a day on intraday prices was criticised in review). Weighted value by bear weight (bull:base kept at 18:42, bear $696): 30% → $1,258, 35% → $1,218, 40% → $1,178, 45% → $1,138. The first draft's $1,248 used 20/45/35 and the old $773 bear.
- Bear DCF sensitivity (value per share, 154M shares): WACC 10.5% / 11.5% / 12.5% → $766 / $696 / $639 (g 2.5%); FY29 FCF $10B → $812, $6B → $580. At 142M basic shares every value is about 8% higher. The result is dominated by the FY29 terminal FCF assumption, not by FY27.
- Upside/downside skew: bull gain is 2.0x the bear loss from $1,126.52, but the probability-weighted edge is modest. P/E on FY28 Street EPS: 3.9x at $1,126.52, 3.6x at $1,036.62.
- Plateau cross-check (added after review). Q1 non-GAAP EPS $49.51 plus the Q2 guide midpoint of $56 leaves about $122 of the $227.80 Street FY27 for H2, i.e. about $61 per quarter or $245 annualised. Street FY28 of $287.26 is therefore +17% above the H2 FY27 run-rate, so the base case needs continued growth while price gains slow to low single digits. At a flat run-rate × 4.5x the value is about $1,101 (−2.3%) and the weighted value would be about $1,097 (+5.8% vs $1,036.62); a 10% haircut on Street FY28 gives a base of $1,163 and a weighted value of about $1,123. This is why the proposal is conditional on entry levels.
- Not re-derived: 3-year targets (the v1.0 $1,850 / $950 are retired), and the peer table in Valuation Dashboard, which is still as of 08/14/2026.
3b. Political-risk overlay (US midterms, 11/03/2026) — added 11/04
Event. Democrats won the Senate and the House. Gina Hinojosa unseated Gov. Abbott in Texas (about 5.20M vs 4.50M votes, 99% counted, Texas Secretary of State) and James Talarico won the Senate seat; Democrats also won or kept the governorships of Georgia, Ohio and Pennsylvania. Hinojosa campaigned on an emergency executive order for a statewide data-center moratorium and on halting approvals until the legislature passes rules 13. Texas holds about 27% of planned US data-center gigawatts (Wells Fargo), Georgia 7%, Ohio 5%, Pennsylvania 5%: about 44% together 13.
Market reaction (11/04, intraday vs 11/03 close). Builders and equipment makers fell 16–24% (Modine, Hut 8, INNIO, Sterling, Cipher), Vertiv −12%, GE Vernova −11%, Nvidia −1.7%, while Amazon, Microsoft and Alphabet rose 5–6% (less data-center build means less capex to fund) 13. SNDK was −8.0% vs the 11/03 close but only −3.6% vs the 11/02 close, because it had rebounded 4.8% on 11/03.
Why this is not simply a panic.
- Abbott had already paused new grid-connection approvals on 08/03 and air permits; ERCOT tracks more than 400 GW of requests 13. The backlash is bipartisan and partly priced.
- A governor cannot legislate. Before the vote the Texas Tribune expected Republicans to keep both chambers; on 11/04 it reported Democrats likely to flip the Texas House (not final) with the Senate staying Republican, so a split legislature is plausible 13. That cuts both ways: the Senate also held a 15-hour hearing in August at which constituents attacked data-center growth, so some regulation is likely, but a full ban needs both chambers and the governor. The 90th session opens 01/12/2027 (bill pre-filing from 11/09/2026), so any law lands mid-2027 at the earliest. Whether an executive order would reach projects already under construction is unclear.
- Washington is divided: no federal moratorium can pass; the federal risk is oversight.
- No hyperscaler has cut: Amazon about $220B in 2026 and "meaningfully more" in 2027, Alphabet "increase significantly" in 2027, Microsoft trimmed 2026 to about $170B on delivery delays, not demand 14.
- SNDK FY27: supply sold out through 2027 and more than 50% of bits under NBM with $16.5B of deposits. A moratorium starting in January 2027 barely touches FY27 revenue.
Why it is not nothing.
- FY28 is where the Street EPS ($287.26) and the cycle debate sit. Upper-bound direct exposure of SNDK revenue to a Texas pause: about 43.5% datacenter mix (Q1 FY27: $4.8B of $11.0B; 33% in Q4 FY26) × 75–100% US share (assumption; the company cites three U.S. hyperscalers behind the datacenter ramp) × 27% Texas ≈ 9–12%, before any relocation.
- The main channel is price momentum. With supply sold out, a slower build-out shows up first in the pace of price increases, not in volumes. It would ease the memory shortage on top of the Kioxia signal. Amazon named memory prices as a driver of its 2026 capex raise 14, so hyperscalers have an incentive to push back on the memory bill.
- Relocation room is narrower when four states with about 44% of planned gigawatts have Democratic governors.
- Counter-argument (analyst reasoning, unsourced): at power-limited sites, SSD's lower power per terabyte than HDD could favour NAND.
Model overlay. Weights move from 20/45/35 (draft 1) to 18/42/40; on the draft-1 share count the weighted value fell $1,248 → $1,209 (−3.2%); after the rev. 3 share-count correction it is $1,178. Sensitivity on the base case: FY28 EPS −5% → $1,228, −10% → $1,163 (at 4.5x). The election explains part of the 11/04 drop, not all: Nvidia, with the same capex exposure, fell 1.7% while SNDK fell 8.0%; the rest reads as the NAND-momentum narrative. There is no quantitative decomposition. 11/04 prices are intraday, not closes.
4. Proposed verdict and entry framework
Proposed: Accumulate on weakness, staged entry, Conviction Medium (v1.0: HOLD). No purchase above T1; at the current price the proposal is effectively wait. Levels are pre-split USD; divide by 8 from 11/20/2026.
| Tranche | Pre-split price | Post-split (÷8) | Share of target size | Weighted-value upside | Downside to bear ($696) | P/E FY28 Street |
|---|---|---|---|---|---|---|
| T1 starter | ≤ $1,000 | ≤ $125.00 | ⅓ | +18% | −30% | 3.5x |
| T2 add | ≤ $820 | ≤ $102.50 | ⅓ | +44% | −15% | 2.9x |
| T3 reserve | ≤ $700 | ≤ $87.50 | ⅓ (only if triggers below are clean) | +68% | ≈ 0% | 2.4x |
- Logic: T1 is a round $1,000 set by the PM; the weighted edge (to $1,178) there is about +18%, and the v1.0 rule asked for a wider margin of safety than ±0. T2 ($820) is where the downside to the bear value falls to about 15%. T3 ($700) is the bear-DCF value itself, so a purchase there assumes the thesis has not broken. Earlier levels were $1,050 / $900 / $775 (draft 1) and $1,000 / $875 / $775 (rev. 2); rev. 3 keeps T1 and lowers T2 / T3 because the corrected share count cut the bear value from $773 to $696. With the plateau-based base case (about $1,101) the weighted edge at T1 would be about +10%, at T2 about +34%.
- At $1,036.62 (11/04 intraday) the stock is 3.7% above T1. The old watchdog zone ($1,350–1,400) has been replaced by these levels (applied to the watchdog on 11/04).
- Thesis-break triggers (stop adding, reassess size, regardless of price): (a) any TrendForce monthly report showing enterprise SSD contract prices down month on month, or its 1Q27 forecast (expected around late December) showing overall NAND contract prices down QoQ; (b) Q2 FY27 non-GAAP gross margin below 84.5% (low end of the guide), or a Q3 gross-margin guide more than 1 point below the Q2 actual; (c) any report of NBM renegotiation or customer deposit disputes; (d) Kioxia's March-quarter shipments do not recover as management says; (e) evidence of YMTC taking enterprise share in North America; (f) a hyperscaler cuts or defers its 2027 capex plan, or Texas executive orders are extended to projects already under construction.
- Open PM decisions (not defined in this note): the triggers above only stop adding; there is no hard-exit rule, absolute risk budget or position size per tranche.
- Next datapoints: Texas bill pre-filing opens 11/09/2026; NVIDIA results ~11/19 (estimated); TrendForce monthly report ~11/30 (estimated); ERCOT files its data-center audit reports with the PUCT by 12/10/2026 (ERCOT document); Texas 90th Legislature convenes 01/12/2027; governor's inauguration ~01/19/2027 (assumed third Tuesday, unverified); Microsoft and Meta Q4 results ~01/27 and Alphabet and Amazon ~02/03–02/05 with 2027 capex plans (all estimated, unconfirmed); split dates (record 11/09, split-adjusted trading 11/20); SNDK Q2 FY27 results ~01/29/2027 (estimated).
5. Limits of this re-underwrite
- Cash conversion (75%) is calibrated on Q1 FY27 and Q4 FY26 adjusted FCF ÷ non-GAAP net income; adjusted FCF excludes NBM prepayments and deposits and Flash Ventures flows, so reported FCF can be lower in quarters with NBM movements. The 154M share count is the Q1 diluted average; basic shares were about 142M on 10/26 after buybacks.
- NBM protection is partial: the $93.9B floor over about 4 years is roughly $23B a year, under half of FY27E revenue (about $49.9B); deposit mechanics, take-or-pay terms, termination rights and counterparties were not reviewed.
- Not modelled (raised in review): single-JV and Kioxia concentration, an earthquake or fab-outage tail, supply response from Samsung / SK Hynix / Micron / YMTC, tariffs and export controls, AI-capex financing, crowded positioning after a +3,000% run, earnings quality (GAAP EPS above non-GAAP). A fourth tail scenario (gross margin back toward 30–40%) was not built.
- Legacy v1.0 text kept for the record has known slips flagged in review: the "~7.6x" forward P/E at $1,655.01 on $213.23 recomputes to 7.8x; the "~36% pullback" is −29.7% against the stated high; the capex-intensity line (5.8–6.3% of revenue vs $179–562M) is internally inconsistent (Q1 FY27 gross capex was $378M, 3.4% of revenue); the 08/14 consensus low / high attributions are unclear; peer discounts are off by 0.1 point.
- NAND price evidence comes from TrendForce summaries and press; the Kioxia figures are from secondary sources (TradingKey, 24/7 Wall St), not the Kioxia release.
- The H-Score (66) is from
H-score_16.11.2026.csv(down from 95 in the 10/19/2026 snapshot). - Section 3b: regional revenue shares are analyst assumptions; the final Texas House result, the scope of any executive order and the inauguration date were not verified; one Reuters article (on data-center opponents' power) could not be read.
v1.1 Data Refresh & Open Contradictions (11/04/2026)
Status in v2.0: items 1, 2 and 4 are addressed in the next section; items 3 and 5 remain open / informational.
| Item | v1.0 (08/14/2026) | v1.1 (11/03–11/04/2026) |
|---|---|---|
| Price | $1,655.01 | $1,126.52 (11/03/2026 close) 9 |
| FY27 P/E | ~7.6x on Street EPS $213.23 8 | 4.85x on CFRA FY27E operating EPS $232.40; 4.95x on Street EPS $227.80 9 |
| FY28 P/E | n/a | 3.9x on Street EPS $287.26 9 |
| FY27 Street EPS | $213.23 (23 analysts) | $227.80 (20 estimates, range $219.30–$241.23; +6.8% vs v1.0) 9 |
| Gross margin (non-GAAP) | 84.6% (Q4 FY26) | 85.1% (Q1 FY27); Q2 guided 84.5%–85.5% 10 |
| Quarterly EPS vs guide | Q1 guide $44–46 | $49.51 non-GAAP, above the guided range ($52.60 GAAP) 910 |
| NBM customers / coverage | 8 customers, >50% FY27 bits | 8 customers, >50% of FY27 and ~2/3 of FY28 bit shipments, pipeline exceeds supply through 2030 9 |
| Contract backlog | $93.9B minimum floor 3 | $89B RPO exiting Q1 FY27 ($60B exiting FY26) 9 |
| Street target | $2,094 (S&P consensus) | $2,070 (CFRA 12-mo target, 10/30/2026) — a different source, not a like-for-like change 9 |
The peer table in Valuation Dashboard is not refreshed: it remains as of 08/14/2026 and must not be compared with the 11/03/2026 SNDK figures above.
⚠️ Contradictions raised in v1.1 (resolution notes added in v2.0):
- Verdict vs scenarios: the v1.0 HOLD rests on a $1,600 base case that was −3.3% from $1,655.01. At $1,126.52 the same base case implies +42.0%, and even the 3-year bear case ($950) is only 15.7% below spot. The v1.0 rule "fresh size above $1,600 needs a wider margin of safety" is now moot. The HOLD, the 20/50/30 probabilities and the DCF need a fresh underwriting pass. → v2.0: done, see below (proposal, pending PM validation).
- Bear-case DCF ($780) assumed FY27 FCF of $22.0B and Street FY27 EPS of $213.23; Street EPS has since risen 6.8% and Q1 FY27 beat. The DCF was not recomputed. → v2.0 rev. 3: recomputed, FY27 FCF $26.3B on 154M diluted shares, value $696.
- Contract backlog: the $93.9B "minimum contracted floor" from the 08/12/2026 call and CFRA's $89B RPO exiting Q1 FY27 are not reconciled (different definitions or timing are possible). Do not use both in one calculation.
- NBM default trigger: Risk 1 (spot drop >40% makes customer default rational) refers to NAND spot/contract prices, not to the equity. The ~56% equity drawdown since the 06/25/2026 peak (11/04 intraday; 52.2% at the 11/03 close) is not evidence that the trigger is hit, and this refresh did not verify NAND spot or contract price trends. → v2.0: contract-price trend checked (TrendForce, Kioxia), trigger far from being hit; NAND spot index level still not verified.
- Stock split: SNDK announced an 8-for-1 forward split (record 11/09/2026, distribution after the 11/19/2026 close, split-adjusted trading from 11/20/2026). Every price, target and level in this document is pre-split; divide by 8 from 11/20/2026 (e.g. $1,126.52 → $140.82).
Company Description & Business Model
Sandisk Corporation (SNDK) is a global leader in the design and manufacture of vertically integrated NAND flash storage solutions. Officially spun off from Western Digital on February 21, 2025 1, the company operates as a pure-play NAND manufacturer with no DRAM exposure, leaving its top-line highly sensitive to NAND Average Selling Price (ASP) fluctuations 1.
SNDK manufactures 100% of its silicon wafers through its long-standing 50/50 joint venture (JV) with Kioxia Corporation at the Yokkaichi and Kitakami fabrication plants in Japan 14. This asset-light manufacturing structure allows SNDK to maintain extremely low direct capital expenditure intensity (capex of 5.8% to 6.3% of revenue, or ~$179M–$562M TTM) 73, leveraging the scale and shared capital burdens of the JV 47.
Revenue Breakdown (FY26: $20.25 Billion) 2:
- Edge (PC, Smartphones, IoT): $12.16 Billion (60.0% of revenue, +195% Y/Y) 2
- Datacenter (Enterprise SSDs): $5.15 Billion (25.4% of revenue, +437% Y/Y) 2
- Consumer (Memory cards, USB drives): $2.94 Billion (14.5% of revenue, +29% Y/Y) 2
Market Dynamics & Industry Positioning
The global NAND flash memory industry is undergoing a structural paradigm shift. Historically plagued by destructive boom-and-bust cycles due to irrational capacity expansions, the sector in 2026 displays unprecedented supply discipline. Industry-wide NAND capex is projected at just $22.2 Billion in 2026 7, as the major integrated memory producers (Samsung, SK Hynix, Micron) aggressively reallocate capital budgets toward advanced DRAM and High-Bandwidth Memory (HBM) for AI accelerators 7.
This structural NAND tightness coincides with a surge in AI-driven storage requirements:
- AI Inference & KV NAND: Generative AI inference requires massive context-window storage. The "KV NAND" architecture (Key-Value caching offloaded to ultra-fast NAND) introduced by Nvidia at CES 2026 created a new storage category 1. SNDK estimates this will generate an incremental demand of 75 to 100 exabytes in FY27 and 150 to 200 exabytes in FY28 1.
- Enterprise QLC SSDs: Hyperscalers are aggressively replacing legacy Hard Disk Drives (HDDs) with high-capacity Quad-Level Cell (QLC) SSDs for training and inference data lakes. Enterprise SSDs now account for 50% of the Datacenter NAND total addressable market, up from 30% in 2025 3.
SNDK management indicates that demand significantly exceeds manufacturing capacity, projecting that wafer output will remain under strict contractual allocation through calendar year 2027 3.
The Moat: Structural Barriers to Entry & NBMs
In addition to capital intensity and technological manufacturing hurdles, SNDK has established a commercial moat to mitigate historical sector volatility:
New Business Models (NBMs) - Contractual Stabilization
To dampen cyclicality, SNDK transitioned its hyperscale and tier-1 OEM customer base to multi-year contractual agreements (NBMs).
- Coverage: SNDK has secured 10 NBM agreements with 8 major datacenter and edge customers, featuring average contract durations of 4 to 5 years 23.
- Visibility: These agreements represent a minimum contracted revenue floor of $93.9 Billion based on contractual floor pricing 3.
- Financial Collateral: Unlike legacy non-binding MOUs, NBMs are secured by $16.5 Billion in cash deposits and escrowed financial prepayments 4.
- Mix Shift: NBMs will account for over 50% of total bit shipments in FY27 and ~66% in FY28 3.
Financial Profile & Performance Tracking
FY26 Financial Summary (GAAP vs. Non-GAAP) 2:
- Revenue: $20.25 Billion in FY26 (+175% vs $7.36 Billion in FY25) 2.
- Non-GAAP Gross Margin: 71.6% in FY26 (+41.4 percentage points vs 30.3% in FY25). Q4 FY26 gross margin reached an all-time peak of 84.6% 2.
- Non-GAAP Operating Margin: 62.7% for the full year ($12.7B operating income). Q4 FY26 operating margin hit 79.2% ($7.1B operating income) due to OPEX compression to 5.4% of revenue 3.
- Non-GAAP Net Income: $10.99 Billion ($70.88 per diluted share) in FY26 compared to $440 Million ($2.99 per share) in FY25 2.
Valuation Dashboard & Peer Analysis (As of August 14, 2026)
| Company | Ticker | Trailing EV/EBITDA | Fwd NTM EV/EBITDA | Period End | As of Date | Premium/Discount vs Target | TTM FCF Yield | Next-FY Fwd P/E | Accounting Regime |
|---|---|---|---|---|---|---|---|---|---|
| Sandisk Corp. | SNDK | 23.3x 2 | 22.0x 4 | Jul 31, 2027 | Aug 14, 2026 | - | ~3.2% 7 | ~7.6x 8 | US GAAP |
| Kioxia Holdings | KIOX | N/A | 8.0x 4 | Jun 30, 2027 | Aug 14, 2026 | -63.6% (Discount) | N/A | ~9.0x | IFRS |
| Micron Technology | MU | 19.4x 6 | 14.0x 6 | Aug 31, 2027 | Aug 14, 2026 | -36.3% (Discount) | ~1.8% | ~8.5x | US GAAP |
| Western Digital | WDC | 29.0x 4 | 5.0x 4 | Jun 30, 2027 | Aug 14, 2026 | -77.2% (Discount) | N/A | ~12.0x | US GAAP |
Sell-Side Consensus Reality Check (08/14/2026 snapshot — superseded, see v1.1 refresh)
A systematic pull of institutional consensus across S&P Global, TipRanks, StockAnalysis, and MarketBeat on 08/14/2026 showed strong sell-side bullishness 8. It is older than 30 days and must not be used to compute upside; current Street data (CFRA, 11/04/2026): 17 Buy, 4 Buy/Hold, 3 Hold, 1 Hold/Sell, 0 Sell; FY27 EPS $227.80 and FY28 $287.26 (20 estimates) 9:
- Consensus Rating: Buy / Strong Buy (23 Analysts: 16 Strong Buy, 3 Buy, 3 Hold, 1 Sell) 8.
- Average 12M Price Target: $2,094.00 (+26.5% upside relative to $1,655.01 spot) 8.
- Price Target Range: Low $1,000.00 (Wells Fargo / RBC at $1,550-$1,600) | High $3,600.00 (Cantor Fitzgerald at $2,900, BofA at $2,500) 8.
- Forward Estimates: Consensus projects FY27 revenue of $48.92 Billion (+141% Y/Y) and Non-GAAP EPS of $213.23 (+200% Y/Y) 8.
- Implied Forward P/E: At $1,655, SNDK trades at ~7.6x FY27 consensus EPS, reflecting an attractive forward valuation that partially underpins institutional buying support 8.
Adjusted 3-Year DCF Valuation Model (WACC 11.5%)
Model inputs adjusted to a 11.5% WACC to reflect pure-play NAND commodity beta and mid-cycle margin normalization.
- Model Parameters: WACC 11.5%, Terminal Growth Rate 2.5%.
- Explicit FCF Projections: FY27(E) $22.0B (supported by eSSD expansion); FY28(E) $15.0B; FY29(E) $8.0B (cycle normalization).
- DCF Intrinsic Fair Value (v1.0): $780.00 (−52.9% vs the $1,655.01 spot of 08/14/2026). v2.0 rev. 3 refresh: FY27 FCF $26.3B (was $22.0B), 154M diluted shares, same WACC / growth / FY28-29 fade → $696 (−38.2% vs $1,126.52). Sensitivities in the v2.0 section.
Independent Multi-LLM Review Synthesis
A multi-model adversarial validation (incorporating Gemini, DeepSeek, Claude, and Perplexity intelligence) highlights the core tension in SNDK's valuation:
- The Bullish View (Street & Sell-Side): Supported by a forward P/E of ~7.6x on explosive FY27 earnings ($213 EPS), locked-in multi-year NBM cash flows, and secular AI datacenter storage demand.
- The Bearish / Prudent View (Internal Risk Framework & Independent Reviewers): Highlighted by potential NBM counterparty renegotiation risks (where a 17.5% deposit default threshold could be breached in a steep spot downturn) and WACC sensitivity.
- Synthesis (v1.0, 08/14/2026 — superseded by v2.0): This analytical tension justified our HOLD stance at $1,655. The review was not re-run for v2.0.
Catalysts & Risk Matrix
- Delivered (10/29/2026): Q1 FY27 results — revenue $11.03B, non-GAAP EPS $49.51 vs the $44.00–46.00 guide, 85.1% gross margin 10.
- Next catalysts: Q2 FY27 results (guided revenue $12.2–12.6B, non-GAAP EPS $55–57 10; date not confirmed in the sources used, likely late January 2027); 8-for-1 stock split (record 11/09/2026, trading split-adjusted from 11/20/2026); additional NBM signings (FY28 coverage expected to tick up 9); first HBF customer samples in CY27.
- Risk 1 (NBM Defection Math): $16.5B in collateral covers 17.5% of total NBM value. If spot prices drop >40%, financial arbitrage favors customer default/re-negotiation 4. v2.0 check: no contract-price series is anywhere near −40%; worst reading is flat wafer contracts in October 11.
- Risk 2 (Supply Normalization): Potential ASP normalization in 2028 as BiCS 8/10 wafer capacity ramps up from Kioxia, Samsung, and SK Hynix 4.
- Risk 3 (Momentum / second derivative, added v2.0): NAND price gains are slowing (Kioxia ASP +70% → +20–25% → low single digit; TrendForce sees a 1Q27 plateau in client SSD and YMTC pressure in 2027 enterprise) 1112. The equity can keep falling on deceleration even if prices stay high.
- Risk 4 (Data-center policy, added v2.0 rev. 2): Democratic governors in Texas, Georgia, Ohio and Pennsylvania (about 44% of planned US data-center gigawatts) and a Texas campaign promise of a moratorium raise the odds that the 2027–28 build-out slips or relocates 13. First-order effect on SNDK FY27 is small (sold-out supply, NBM coverage); the risk is to FY28 volumes and above all to price momentum.
Scenarios (Bull, Base, Bear) — v2.0 draft
Replaces the v1.0 triple-horizon scenarios. Full table, basis and sensitivities are in the v2.0 Re-underwrite section.
BULL CASE (Probability: 18%) — $1,980 (+75.8% vs $1,126.52)
- Thesis: FY28E EPS reaches ~$330 (≈15% above Street) as NBM-covered bits (>50% of FY27, ~2/3 of FY28) and enterprise QLC demand hold; the cycle-peak fear fades and the multiple recovers to 6.0x.
BASE CASE (Probability: 42%) — $1,293 (+14.8%)
- Thesis: Street FY28E EPS ($287.26) is delivered, NAND price gains plateau but do not reverse, and the multiple recovers modestly from 3.9x to 4.5x. Range-bound trading around $1,000–1,400 until Q2 FY27 results.
BEAR CASE (Probability: 40%) — $696 (−38.2%)
- Thesis: Price momentum turns into a plateau then decline in 2027 (client SSD first, then enterprise as YMTC and new capacity arrive). FCF fades $26.3B → $15B → $8B; the market capitalises that fade at WACC 11.5%. A breach of NBM defence (>40% price fall) is not required for this case. A slower or relocated US data-center build-out (section 3b) feeds this path through weaker price momentum.
v1.0 scenarios (08/14/2026), kept for the record: Bull $2,200 / $2,800 (30%), Base $1,600 / $1,850 (50%), Bear $780 / $950 (20%).
Sourcing Integrity & Audit Ledger
| Source Category | Found / Coverage | Last Update Date | Verified / Checked | Reference File/URL |
|---|---|---|---|---|
| 1. Regulatory Filings | YES | 2026-08-05 | YES | SNDK Q4 FY26 Form 10-K/Earnings Release 2 |
| 2. Earnings Call Transcripts | YES | 2026-08-12 | YES | SNDK Q4 FY26 Earnings Call Transcript 3 |
| 3. Real-Time News Flow | YES | 2026-08-11 | YES | Reuters secondary share sale & Trefis Analysis 14 |
| 4. Sell-Side Broker Research | YES | 2026-08-14 | YES | TipRanks, StockAnalysis, S&P Global Consensus 8 |
| 5. Live Street Consensus (MANDATORY) | YES | 2026-11-04 | YES | CFRA Wall Street consensus (20 estimates) 9; 08/14/2026 TipRanks / StockAnalysis / MarketBeat snapshot 8 superseded |
| 6. Market Consensus Data | YES | 2026-08-14 | YES | MarketBeat & S&P Analyst Forecasts 8 |
| 7. Independent Multi-LLM Review | YES | 2026-08-14 | YES | Gemini 3.1 Pro, DeepSeek V4, Claude CLI, Perplexity 8 — not refreshed in v1.1 |
| 8. CFRA Stock Report (v1.1) | YES | 2026-11-04 | YES | CFRA analyst view dated 10/30/2026, price as of 11/03/2026 9 |
| 9. Q1 FY27 results (press + company release) | YES | 2026-11-04 | YES | Sandisk press release 10/29/2026 (revenue, EPS, margin, guide, diluted shares, adjusted FCF, split) and Yahoo Finance / Motley Fool coverage 10 |
| 10. NAND price research (v2.0) | YES | 2026-11-04 | PARTIAL | TrendForce 4Q26 forecast (09/30/2026) and monthly price report via Capital Times (10/30/2026) 11; summaries, not the full TrendForce reports |
| 11. Kioxia guidance (v2.0) | YES | 2026-11-02 | PARTIAL | TradingKey and 24/7 Wall St, 11/02/2026 12; secondary sources, Kioxia release not read |
| 12. Policy / election (v2.0 rev. 2) | YES | 2026-11-04 | PARTIAL | Texas Secretary of State results, CNBC, Reuters, Texas Tribune, StockTitan (11/04/2026) 13; hyperscaler capex from earnings coverage 14; Reuters piece on data-center opponents not readable |
Audit Footnotes:
References(15)
- 1
Sandisk Corporation (SNDK) Company Profile: Sourced from
/repo/wiki/companies/Sandisk Corporation.md. - 2
Sandisk Q4 FY26 Earnings Release (August 5, 2026): Q4 Revenue $8.97B, Gross margin 84.6%, FY26 Revenue $20.25B, Non-GAAP EPS $70.88.
- 3
Sandisk Q4 FY26 Earnings Call Transcript (August 12, 2026): $93.9B NBM backlog, $16.5B prepayments/deposits, Datacenter mix at 38% as stated on the call (the Q4 FY26 press release shows 33.2%: $2,977M of $8,965M).
- 4
Trefis High Quality Portfolio Analysis: Kioxia JV partner EV/EBITDA multiple at 8x vs SNDK at 22x.
- 5
StockAnalysis.com Historical Share Price Records (Aug 14, 2026): Current market quote ~$1,655.01.
- 6
AInvest & Seeking Alpha Market Reports (August 2026): Comparison multiples and market pullbacks.
- 7
Eightgen AI Insights & AInvest Analysis: Capex intensity ($179M TTM) and FCF margins.
- 8
S&P Global, TipRanks & StockAnalysis Consensus (August 14, 2026): 23 Analysts consensus Buy, Avg Price Target $2,094. FY27 Consensus Revenue $48.92B (+141% Y/Y), FY27 Consensus EPS $213.23 (+200% Y/Y), Forward P/E ~7.6x.
- 9
CFRA Stock Report, Sandisk Corporation (SNDK) —
raw/research/stocks/2026-11-04-sandisk-corporation.md: Strong Buy, 12-mo target $2,070 (analyst view 10/30/2026), price $1,126.52 (close 11/03/2026), Oper. EPS FY27E $232.40 / FY28E $276.00, Street FY27 $227.80 / FY28 $287.26 (20 estimates), RPO $89B, 8 NBM customers. - 10
Press coverage of Q1 FY27 results (10/29–10/30/2026): Yahoo Finance and Motley Fool — revenue $11.03B, adjusted EPS $49.51, adjusted gross margin 85.1%, Q2 guide $12.2–12.6B / EPS $55–57, 8-for-1 split dates. Checked in rev. 3 against the company release (10/29/2026): revenue $11,031M, non-GAAP EPS $49.51, GAAP EPS $52.60, non-GAAP gross margin 85.1%, Q2 guide, diluted weighted shares 154M, Q1 adjusted FCF $5,696M, datacenter revenue $4,802M (Q4 FY26: $2,977M of $8,965M), 8-for-1 split effective 11/19/2026.
- 11
TrendForce (09/30/2026 press release; 10/28–10/30/2026 monthly bulletins via Capital Times): 4Q26 NAND contract +15–20% QoQ, enterprise SSD +23–28%; October NAND wafer contracts flat; wafer spot down sharply; client SSD plateau expected 1Q27; YMTC pressure on 2027 enterprise pricing.
- 12
Kioxia FQ3 guidance (reported 10/30/2026; shares fell 11/02; TradingKey, 24/7 Wall St): Oct–Dec adjusted operating profit ≈ ¥2.0T vs ¥2.32–2.35T expected; ASP +20–25% QoQ last quarter after ~+70%, guided low single digit; 2027 output sold out per CFO. Secondary sources.
- 13
US midterms, 11/03–11/04/2026: Texas Secretary of State results (Hinojosa ≈5.20M vs Abbott ≈4.50M, 99% counted); CNBC 11/04 "AI data center stocks … decline as Texas governor result raises risks of a slowdown"; Reuters 11/04 (data-center stocks slide, Texas data centers face uncertain future); Texas Tribune 10/30 (divided government) and 11/04 (Texas House likely to flip, Senate stays Republican); Reuters / DLA Piper / ERCOT documents for the 08/03 pause and the 12/10 audit report date; StockTitan 11/04 (intraday prices vs 11/03 close). Wells Fargo shares of planned US data-center GW: Texas ~27%, Georgia 7%, Ohio 5%, Pennsylvania 5%.
- 14
Hyperscaler Q3 2026 capex (10/28–10/30/2026 coverage: Business Insider, Yahoo/StockTwits, Luminix): Amazon ~$220B 2026 ("meaningfully more" in 2027), Alphabet $195–205B (2027 "increase significantly"), Microsoft ~$170B calendar 2026 (cut from ~$175B on delivery delays), Meta $135–145B. Amazon cited higher memory prices in raising its guide.
- C
Live Street Consensus (Aug 14, 2026): 23 Analysts (16 Strong Buy, 3 Buy, 3 Hold, 1 Sell). Average 12M Price Target $2,094.00 (+26.5% upside). FY27 Revenue $48.92B (+141%), FY27 EPS $213.23 (+200%). Forward P/E ~7.6x.
Document facts
- H-Score
- 66
- Conviction
- Medium
- Sector
- Tech Semiconductors
- Geography
- US
- Accounting
- US GAAP
- Created
- 2026-08-14