SanDisk CorporationNASDAQ:SNDK
Published 2026-08-06, with market data as of the 2026-08-05 close. Financials are sourced from SanDisk's FY2026 Q4 report (issued after market close on 2026-08-05) and public filings.
SanDisk shattered the market's prior 'memory-cycle stock' bias with a 'record earnings plus 10 NBM long-term deals locking in $93.9B of minimum revenue': Q4'26 revenue of $8.97B (+51% QoQ / +372% YoY), Non-GAAP gross margin of 84.6%, Non-GAAP EPS of $39.25 (well above the $34.96 consensus and the $30-33 guide); FY2026 revenue of $20.25B (+175%). After the close on August 5 the stock briefly dipped -12% (the Q1'27 guide midpoint of $10.55B was only marginally below the $10.62B consensus — a 'sell-the-news' move), but the NBM model (~8 customers, weighted average >4 years, $93.9B minimum, $59.8B RPO, $16.5B financial guarantees) stretches revenue visibility from '3 months' to '4+ years'.
But we caution: after the surge volatility is extreme — in July alone it drew down as much as -53%, and over 52 weeks it ran from $40.53 to $2,354.39 and back to $1,350.50; this is a name priced to perfection on the AI-storage narrative. We rate it Buy with Low-Medium conviction and a 12-month base target of $1,650 (+22% vs spot): the NBM minimum revenue plus data-center mix rising from 12% to 38% are real, but memory remains cyclical — this cycle is just longer — and the guide below consensus, the CXMT (China) supply threat and NAND cyclicality cap the pace and certainty of upside. We apply an explicit cycle discount to valuation (Forward P/E ~9x, on a FY2027E EPS assumption of $150); the market already doubts earnings sustainability, so the margin of safety is thin.
§01Investment Summary
SanDisk's Q4'26 report settled the debate over whether NAND is 'just another cycle': revenue +372%, gross margin 84.6%, EPS $39.25 all beat, and the NBM model stretches revenue visibility from 3 months to 4+ years. The -12% after the close on August 5 was 'good print, guide below consensus' — sell-the-news, not a fundamental reversal. Memory remains cyclical, this cycle is just longer — we cross-check three Forward P/E scenarios (8/11/14x) on a normalized FY2027E EPS assumption, arriving at a probability-weighted 12-month base target of $1,650.
Three reasons to Buy: (1) Q4'26 beat across the board (revenue / gross margin / EPS all beat) and the NBM model stretches visibility to 4+ years; (2) data-center mix rose from 12% to 38% and the NAND shortage extends past 2027 (Citi sees full-year ASP +186%); (3) $15.5B remaining buyback plus ~$6.5B net cash (zero interest-bearing debt) provide downside protection.
Three key risks: (1) the guide was only marginally below consensus (Q1'27 mid $10.55B vs $10.62B, just -0.7%), and the -12% after-hours was mostly sell-the-news; (2) CXMT (ChangXin) IPO +466% raising capacity, reshaping supply mid-to-long term; (3) NAND cyclicality plus the extreme -53% July drawdown. This is why we set the $1,650 base target below the $2,250 consensus median — the market treats it as a cyclical name, and so do we, rather than pricing it like a euphoric AI-software stock.
§02Price and Street View
On August 5, SNDK closed at $1,350.50 (-5.40%, volume 16.4M shares, 11.05% turnover) and briefly fell to ~$1,220 (-12%) after hours — earnings beat but the Q1'27 guide midpoint missed consensus, triggering a sell-the-news selloff. From the year open (2025-12-31) $237.38 to the June 22 high $2,354.39 to August 5 $1,350.50, the intra-year range is 882% and YTD +469%. The -53% July drawdown (from ~$2,354 to ~$1,220) was driven mainly by a ~$1.1B forced liquidation by the Situational Awareness fund plus the CXMT IPO (+466%) and semiconductor deleveraging — not purely fundamental deterioration.
| Metric | Value | Source / Note |
|---|---|---|
| Current price | $1,350.50 | 2026-08-05 收盘 (-5.40%)；盘后 ~$1,220 |
| Market cap | ~$200B | 1,350.50 × 148.09M shares (行情源) |
| 52W range | $40.53-$2,354.39 | 低点近分拆 / 高点 2026-06-22 |
| YTD | +469% | 2025-12-31 $237.38 → 8/5 $1,350.50 |
| Volume (8/5) | 16.37M | 换手率约 11.05% |
| Analyst PT | $1,000 / $2,381 / $3,169 | Low / Avg / High · 30 家 (Aug 5) |
| Recommendation split | 77% / 20% / 3% | Buy / Hold / Sell · 30 家 (Aug 2026) |
| TTM 指标 | EPS $70.88 (Non-GAAP) | FY2026 |
| Shares outstanding | 148.09M | 行情源 |
§03Bull/Bear View
Bull: the 'picks-and-shovels' of AI storage + NBM re-architects the NAND model + data-center mix surges
Q4'26 revenue of $8.97B (+51% QoQ / +372% YoY), Non-GAAP gross margin 84.6%, EPS $39.25 (beat $34.96, +12.3%); FY2026 revenue $20.25B (+175%). The NBM model (10 deals / ~8 customers, weighted average >4 years, $93.9B minimum, $59.8B RPO, $16.5B guarantees) stretches revenue visibility from 3 months to 4+ years, covering >50% of FY2027 bits and ~2/3 of FY2028 bits — the first 'price floor + ceiling' structure in 30 years of NAND cycles. Data-center revenue was $2,977M in Q4 (+103% QoQ, 38% of bits, vs 12% a year earlier), Edge $5,432M (+48% QoQ); the NAND shortage persists (TrendForce Q2'26 contract prices +70-75% QoQ, Citi sees full-year +186%). Net cash ~$6.5B (zero interest-bearing debt) plus $15.5B remaining buyback. The CEO says it is 'still early'.
Bear: guide below consensus + CXMT China supply threat + extreme volatility and cyclicality
The core bear concern: the Q1'27 guide midpoint of $10.55B was only marginally below the $10.62B consensus (-0.7%), and the EPS midpoint of $45 was slightly under $45.58 — 'a good print is not enough; the market wants consistent upward revisions.' Structurally, CXMT (ChangXin Memory) IPO +466% raising capacity could, mid-to-long term, reshape NAND supply and depress the price floor; the -53% July drawdown (from ~$2,354 to ~$1,220) shows how extremely the name is priced on narrative and how sensitive it is to flows (a ~$1.1B forced liquidation by the Situational Awareness fund was enough to trigger it). Every NAND super-cycle in history has collapsed on oversupply, and the NBM floor does not remove cycle risk. Forward P/E ~9x implies the market doubts earnings sustainability. Concentration: NBM has only ~8 large customers, and a default by any one (even with the $16.5B guarantee) would still hit.
§04Financial Truth
Q4'26 revenue of $8,965M (+51% QoQ, +372% YoY, above the $7.75-8.25B guide top); Non-GAAP gross margin 84.6% (vs Q3 78.4%, vs 26.4% a year earlier); Non-GAAP operating margin 79.2%; Non-GAAP EPS $39.25 (vs guide $30-33, vs consensus $34.96, Beat +12.3%); GAAP EPS $43.97; Adj. FCF $5,035M (56% margin); cash $4,762M + marketable securities $1,777M, zero interest-bearing debt (net cash ~$6.5B). Roughly one-third of the sequential increase came from volume, two-thirds from price (NAND price increases).
| Metric (Q4'26) | Actual | Consensus | Diff | Note |
|---|---|---|---|---|
| 总营收 | $8,965M | $8,250M (指引上沿) | 超上限 | +51% QoQ / +372% YoY |
| Non-GAAP EPS | $39.25 | $34.96 | +$4.29 (+12.3%) | beat |
| Non-GAAP 毛利 | 84.6% | 81.5% | +3.1pp | vs Q3 78.4% |
| Q1'27 营收指引 | $10.3-10.8B | $10.62B | 中值 -0.7% | 仅微低于共识 |
| Q1'27 EPS 指引 | $44-46 | $45.58 | 中值 -1.3% | 略低 |
By segment: Data Center $2,977M (+103% QoQ, +1298% YoY, 38% of bits), Edge $5,432M (+48% QoQ), Consumer $556M (-32% QoQ). FY2026: DC $5,153M (+437% YoY), Edge $12,160M (+195% YoY), Consumer ~$2,935M. The rise in NAND average selling price is the main driver of the gross-margin jump.
| Metric ($M) | Q4'25 | Q3'26 | Q4'26 |
|---|---|---|---|
| Revenue | ~1,900* | 5,945 | 8,965 |
| Non-GAAP GM | 26.4% | 78.4% | 84.6% |
| Non-GAAP EPS | $0.29 | $23.41 | $39.25 |
*Q4'25 revenue derived as $8,965M / 4.72 (~$1,900M, marked DERIVED); Q4'25 GM/EPS per the company's YoY disclosure. NBM: this report discloses 10 deals cumulatively (5 in April + 5 new this quarter including 3 new customers); the $93.9B minimum, $59.8B RPO and $16.5B guarantees are from the prior 10-K (the press release did not update the amounts), covering >50% of FY2027 bits and ~2/3 of FY2028. Buyback: $14B added, $15.5B remaining. Management expects the NAND market >$300B in 2026 and ~$500B in 2027, with data-center mix 30%→50%.
| Guide & Capital Return | 数值 | Note |
|---|---|---|
| Q1'27 营收指引 | $10.3-10.8B | 中值 $10.55B，略低于共识 $10.62B（仅 -0.7%） |
| Q1'27 EPS 指引 | $44-46 | 中值 $45 |
| Q1'27 GM 指引 | 83-85% | 维持高位 |
| 剩余回购授权 | $15.5B | 新增 $14B |
| 现金 / 净现金 | $4,762M / ≈$6.5B | 现金 $4.76B + 有价证券 $1.78B，零有息债 |
Cash flow (official 8-K consolidated statement of cash flows, $M): Q4'26 operating cash flow (OCF) $7,126, capex $43, FCF $7,083, adjusted FCF (Adj FCF, excluding Flash Ventures and NBM prepayment effects) $5,035 (56% of revenue); FY2026 OCF $11,671, FCF $11,494, Adj FCF $8,743 (43% of revenue). Light capex (PP&E only $177M/year) makes free-cash conversion extremely high, supporting the $15.5B buyback and NBM fulfillment capacity.
| 现金流量 ($M) | Q4'26 | FY2026 |
|---|---|---|
| Operating Cash Flow (OCF) | 7,126 | 11,671 |
| Capex (PP&E, net) | 43 | 177 |
| Free Cash Flow (FCF) | 7,083 | 11,494 |
| Adjusted FCF (Adj FCF) | 5,035 | 8,743 |
| Adj FCF margin | 56% | 43% |
§05Peer Comparables
SanDisk competes in 'NAND flash + AI storage' against Micron (MU), SK hynix, Kioxia and Samsung. On valuation: TTM P/E {TTM_PE:.1f}x (Non-GAAP EPS $70.88), Forward ~9x (FY2027E EPS assumption $150, under cycle-discount multiples) — well below pure-play software-AI names like PLTR, but above the trough of traditional memory cyclicals; this is why the market treats SNDK as a cyclical and we discount the multiple. On growth: FY2026 revenue +175% and 84.6% gross margin show pricing power. The market prices SNDK as an option on 'AI-storage picks-and-shovels + NBM partial de-cycling', not as 'already fully de-cycled'.
| SNDK's own multiples | Value | Note |
|---|---|---|
| P/E (TTM, Non-GAAP $70.88) | 19.1x | computed |
| Forward P/E (FY2027E EPS $150 假设) | 9.0x | MODEL · 周期折价 |
| P/S (TTM, $20.25B) | 9.9x | computed |
| P/S (FY2027E, ~$43B 假设) | ~4.7x | MODEL |
| EV / Sales (TTM) | ~9.6x | 净现金 ≈$6.5B（现金 $4.76B + 有价证券 $1.78B，零有息债） |
| Net cash / Volatility | ≈$6.5B / 极高 | FY2026 零有息债务；7 月 -53% |
§06Ownership
Institutions are concentrated (13F); the top three are FMR 8.92% (13.20M), BlackRock 7.03% (10.40M) and Vanguard 6.43% (9.53M); State Street 3.76%, Geode 2.84%. Former parent Western Digital still holds 0.70% (trimming). Insider/founder-linked stakes are small. The July drawdown was mainly driven by a ~$1.1B forced liquidation by Situational Awareness LP (holds 0.77%). No material congressional-disclosure trades.
§07Valuation and Price Target
SNDK is profitable (FY2026 Non-GAAP EPS $70.88), but the TTM {TTM_PE:.1f}x anchors to peak-cycle earnings, which is unreliable for a memory company. We re-do valuation under a 'memory remains cyclical, this cycle is just longer' framework: a normalized FY2027E EPS assumption (Base $150, Bull $195, Bear $85, anchored to the $157 Q4'26 annualized run-rate with conservative haircut / hold / modest uplift) plus cycle-discount multiples (Bear 8x, Base 11x, Bull 14x — well below the prior euphoric 10/14/19x, reflecting the market's pricing of memory as a cyclical) gives three scenarios: Bear $680 = 8x (cycle rolls over, CXMT expands, ASP compresses — ~$101B mcap); Base $1,650 = 11x (NBM delivers, data-center mix to 50%, shortage persists but gets a cyclical multiple — ~$244B mcap); Bull $2,730 = 14x (AI-storage re-rating, continued beats, near but below the $3,169 sell-side high — ~$404B mcap). Probabilities 25/45/30 (cycle downside lifts the Bear weight), weighted $1,730, with a 12-month base target of $1,650 (+22% vs spot). Rating and scenarios are self-consistent: bull $2,730 > spot $1,350.50 > bear $680; weighted $1,730 > spot. We do not adopt the $2,250 consensus median / $2,381 average — after the -12% after-hours they will likely be cut in bulk, and they under-price NAND cyclicality; our Base already applies an explicit cycle discount. FY2027E EPS is a MODEL assumption (based on annualizing the $45 Q1'27 guide midpoint with a conservative haircut), not consensus.
| Scenario | FY2027E EPS x P/E | PT | Weight |
|---|---|---|---|
| Bear | $85 × 8x / 148.09M 股 | $680 | 25% |
| Base | $150 × 11x / 148.09M 股 | $1,650 | 45% |
| Bull | $195 × 14x / 148.09M 股 | $2,730 | 30% |
| Weighted PT | 0.25×680 + 0.45×1650 + 0.30×2730 | $1,730 | 100% |
![图 9：估值足球场——Street 共识 [$1,000, $3,169] 与 VM 三档 $680/$1,650/$2,730，现价 $1,350.50。](/api/research-chart/sndk-real-data-202608/charts/zh/ch09_football_field.svg)
![Figure 9: Valuation football field — Street consensus [$1,000, $3,169] vs VM tiers $680/$1,650/$2,730; spot $1,350.50.](/api/research-chart/sndk-real-data-202608/charts/en/ch09_football_field.svg)
§08Catalysts and Risks
| Date / Window | Event | Direction |
|---|---|---|
| From Aug 2026 | Sell-side repricing after the Q1'27 guide below consensus (-12% after-hours) | Near-term over? |
| Ongoing | New NBM customer signings / CXMT capacity competition | Both |
| 2026-09-16 | FOMC (rate-sensitive) | A hike pressures high multiples |
| Oct/Nov 2026 | Q1'27 earnings (tests the $10.55B guide and 83-85% margin) | Key test |
| Ongoing | NAND contract-price trend (TrendForce / Citi +186% full-year) | Positive if persists |
| 2027 | CXMT capacity release / data-center mix 50% | Bear/bull divide |
(1) Sell-the-news: the Q1'27 mid $10.55B was only marginally below the $10.62B consensus (-0.7%); the -12% after-hours was mostly a valuation give-back after the huge run into the print (52W high $2,354), not a guide blow-up;
(2) CXMT (ChangXin) IPO +466% raising capacity, reshaping supply and depressing the price floor mid-to-long term;
(3) Extreme volatility: -53% in a single July month, 882% 52-week range, highly flow-sensitive;
(4) NAND cyclicality: the NBM floor does not remove the cycle; every historical super-cycle has collapsed on oversupply;
(5) Customer concentration: NBM has only 10 deals (~8 large customers); a default by any one would hit (even with the $16.5B guarantee);
(6) Rates: high multiples are sensitive to the 10Y;
(7) Valuation-assumption risk: FY2027E EPS is a MODEL assumption; if actual comes in well below $150 the Base PT does not hold.
Upside risk: if Q1'27 beats, the NAND shortage persists and the CXMT threat eases, the market could re-rate to 14x forward — the $2,730 bull case (+102% vs spot).
§09Sources, Limits and Disclosure
- SanDisk FY2026 Q4 earnings (official SEC 8-K, accession 000162828026053346, 2026-08-05; full Ex-99.1 release; same source as IR page investor.sandisk.com, used as primary here
- Price and valuation: Nasdaq / quote feed / moomoo / stockanalysis (2026-08-05/06)
- Analyst ratings and consensus: S&P Global / moomoo (30 brokers, 8/1) / Citi ($2,500, 6/25)
- Ownership: quote feed / 13F (2026-08-06 snapshot)
- NBM and industry: Tencent News / 163 / TechTimes / TrendForce / Citi
- Macro and market: Jin10 / Wall Street CN
Limitations: spot is the 2026-08-05 regular close of $1,350.50 (after-hours briefly ~$1,220, not used); Q4'25 revenue ($1,901M) is from the company's YoY disclosure (the Q4'26 base), not derived by division; FY2027E EPS ($85/$150/$195) is a MODEL assumption, not consensus, anchored to the $157 Q4'26 annualized run-rate with conservative haircut / hold / modest uplift, and the multiples (8/11/14x) are subjective assignments under a cycle-discount framework, not market consensus; analyst target Low/Avg/High ($1,000 / $2,381 / $3,169) are from Benzinga / TECHi broker aggregation (30 brokers, Aug 5 snapshot); Citi's single $2,500 (6/25) is illustrative only, not the market high (high is Susquehanna $3,169); Beta is not from a realtime source (volatility proxied by the -53% July drawdown and 882% 52-week range); EV approximates market cap (net cash ~$6.5B: cash $4.76B + marketable securities $1.78B, zero FY2026 interest-bearing debt); one Xueqiu post citing $8.17B / 80.4% GM contradicts the IR official $8.97B / 84.6% and was discarded in favor of the IR release. This report is not investment advice.