Sandisk Corporation (SNDK)

Executive Summary & Action Plan

Verdict

SELL — Price-led NAND peak: 84.6% gross margin, two-thirds of Q4 growth from ASP. At $1,601 against ~$523 intrinsic, we exit into strength.

Price $1600.62 | Market cap $234.4B | Target $1550 | Upside -3.2% 12M market-path target · intrinsic $463–$523 (blended $523) — see Fair Value · Street $2,126 (n=23, reference)

The Take

The tape says supercycle: revenue $1.7B to $9.0B in six quarters, gross margin 22.5% to 84.6%. But two-thirds of June-quarter growth was price, consumer units fell 32% QoQ, and the supply response is funded — Kioxia–Sandisk JV capex +41%, Micron NAND +63%, YMTC at ~14% share. Consensus chased $688 to $2,126 with 122% dispersion. Mid-cycle intrinsic is ~$523. We exit into strength; the NBM floor softens the landing — it does not repeal the cycle.

Action Plan

Sell half at market; sell the rest into strength toward $1,900; exit everything on a break of $1,350.

• R/R 1.1:1 (hold-side gate: bull $2,400 vs bear $900 from $1,600.62 — below 1.5, blocks any long add) | Prob-weighted 12M -3.2% | Confidence: Medium | p_thesis_wrong 35%
• Ladder basis: $1,350 = 8/5 post-guide-miss close ($1,350.50); $1,900 = 3M bull-path point under the $2,354 blow-off high; 50d SMA $1,655.

Forecast Path

The cone starts wide and widens: pricing momentum (EPS revisions +6.6%/30d, breadth 0.8) carries the bull tail near-term, while funded CY27 supply and TrendForce's "gains moderate" call drag the base lower. Probabilities are derived — bull 30% (Q4'26 contract hikes stick, per sentiment odds 65%, discounted for multiple compression), base 45%, bear 25% — blended with p_thesis_wrong 35%; since the call is SELL, p_thesis_wrong (35%) ≥ bull (30%), the 5-point gap absorbing base-band outcomes above $1,600.

By the Charts

Six quarters, 62 points of gross margin — the steepest repricing in NAND history.

COGS flat at ~$1.4B while revenue quintupled — the entire P&L delta is price, not bits.

Datacenter went 11% → 33% of revenue in a year; consumer is being priced out (-32% QoQ).

Intrinsic methods cluster at $463–$523 on mid-cycle economics — roughly 67% below the tape.

Fair Value — Two Layers

Layer 1 intrinsic: log-median of DCF $463 / normalized EV/EBITDA $523 / normalized EV/FCF $523 = $523, built on mid-cycle revenue $26B (E) and a 35% EBITDA margin — the normalization every prior NAND cycle has delivered (Micron GM 61% → 27% in five quarters, 2018–19). Layer 2 12M path: 30% × $2,400 + 45% × $1,350 + 25% × $900 = $1,553, rounded to $1,550. The layers sit ~196% apart because the 12M tape still prices contracted peak economics — $93.9B NBM backlog at ~80% gross margin — as quasi-permanent; we do not promote intrinsic to target, and we do not pay for permanence either. The 12M bear ($900) is deliberately shallower than the full-cycle-trough math ($270–460) because a roll would begin, not complete, within 12 months and NBM floors slow the repricing. Decomposition (arithmetic identity): EPS growth ×3.59 ($73.76 TTM → $264.72 NTM cons.) × re-rate ×0.27 (21.7x → 5.9x trailing) × payout ×1.00 = -3.2%, matching the -3.2% target gap (≤0.1%p).

촉매 캘린더

단기 (0~6개월)

The next two quarters decide whether pricing momentum survives contact with its own supply response — every row below is either a pricing print or a capacity signal, and the stock has already shown (8/5, -5.4%) that it sells off when consensus outruns the company.

중기 (6~12개월)

Risks & Thesis Breakers

• Funded supply response (High × High): JV capex +41% YoY, Micron NAND +63%, YMTC ~14% share building toward 500K wpm — bits land CY27. / Price-led demand air-pocket (High × High): consumer -32% QoQ already; hyperscaler LTAs can convert to CY27 inventory digestion. / Peak-multiple trap (Med-High × High): Street $688 → $2,126 in six months is chasing; 6x forward P/E on 84.6% GM is the classic peak-earnings illusion.
• Breakers (flip us back to hold): ① Q4 CY26 contract prices print another +10%+ QoQ; ② FQ1'27 beats $10.8B with Datacenter >35% of mix and GM ≥75%; ③ hyperscaler CY27 capex guides +30% YoY with explicit storage commentary; ④ YMTC ramp slips past 2027 while JV capex holds ≤$4.5B; ⑤ a quarter of bit growth +30% YoY at flat-to-up ASP (volume-led rotation).

Appendix