I'll research SNDK (Sandisk) before writing the preview. Let me start by gathering recent earnings materials, filings, and news.Let me read the most recent earnings materials and check recent news around the report date.Now let me get recent news and stock price data leading into the report.Let me quickly gather external context on consensus estimates and the July selloff via web search.Let me read a couple of these to understand consensus estimates and the July selloff narrative.# Sandisk (SNDK) — Fiscal Q4 2026 Earnings Preview
Reports after the close, Wednesday, August 5, 2026 (FY2026Q4 + full fiscal year)
The one-line setup
Sandisk goes into its fiscal year-end print as both one of the great AI-memory momentum stories of the past year and, more recently, the S&P 500's single worst-performing stock of July 2026. The report is less about whether the quarter was huge — guidance and consensus both imply a blowout — and more about whether management can defend the durability of ~80% gross margins and the "no-more-cyclicality" narrative against a violent late-July memory de-rating driven by fresh Chinese competition. The options market is pricing a move on the order of ±25%.
What the company guided (fiscal Q4 2026)
On the Q3 call (April 30), management guided FQ4 to:
| Metric (Non-GAAP) |
Guidance |
| Revenue |
$7.75B – $8.25B |
| Gross margin |
79.0% – 81.0% |
| Operating expenses |
$480M – $500M |
| Tax expense |
$775M – $875M |
| Diluted EPS |
$30.00 – $33.00 |
| Diluted shares |
~158M |
For context, that revenue midpoint (~$8.0B) is up roughly 300%+ year over year, and the EPS range compares to just $0.29 in the year-ago quarter.
What the Street expects
Consensus sits at or slightly above the top of guidance — the Zacks revenue consensus is ~$8.3B (~+337% YoY) with EPS consensus in the $33–$34 range (Zacks ~$34.24; other trackers cite ~$33.38). Two things matter here:
- The bar is above the guide. Analysts are effectively assuming another beat-and-raise, so an in-line-to-guidance print could read as a disappointment even though it would be a record quarter.
- The beat history is extreme. SNDK has topped consensus in each of the last four quarters, and last quarter was a genuine blowout: revenue of $5.95B vs. a $4.4–4.8B guide, gross margin of 78.4% vs. 65–67% guided, and EPS of $23.41 vs. $12–14 guided. Expectations for outperformance are baked in.
The stock: from AI darling to July's biggest loser
This is the crux of the setup. SNDK ran to an all-time high above $2,300 in late June, then collapsed — down ~54% in July, the worst stock in the S&P 500 for the month — before a ~26% bounce on July 30 (its best day since January). It closed August 4 around $1,428, roughly 40% below the June peak.
The sell-off was largely sector, not company-specific:
- CXMT's blockbuster Shanghai IPO (shares reportedly surged ~466% on debut after raising ~$8.6B — the largest mainland semiconductor offering) reignited fears of Chinese NAND/DRAM oversupply and margin normalization.
- Reports that China has made progress on domestic DUV lithography and that Apple is testing CXMT chips stoked worries that Chinese memory could reach top-tier customers.
- Peer weakness fed the derating: SK Hynix fell ~47% from its June peak; Micron and Western Digital also sold off.
- After a ~500% YTD gain, SNDK was simply a crowded, profit-taking target as the "AI memory trade" unwound.
Net: the setup is asymmetric and emotional. A confident guide could snap the stock back hard; any hint of ASP softness or margin give-back plays directly into the bear thesis.
The bull case going in
- New Business Models (NBMs) — the "de-cyclicalization" story. As of the Q3 call, Sandisk had signed 5 multi-year supply agreements, with the three closed in-quarter carrying ~$42B of minimum contractual revenue (RPO) and the five together backed by >$11B in financial guarantees (prepayments + third-party instruments). Management said these cover over one-third of FY2027 bits and expects that share to climb toward/above 50%. A reported multi-year NAND supply deal with Meta underscores hyperscaler demand and gives the framework credibility.
- Data center inflection. Datacenter revenue grew 233% sequentially to $1.47B in Q3 (25% of the portfolio), driven by TLC enterprise SSDs. FQ4 is when QLC "Stargate" solutions begin shipping for revenue — a potential incremental leg.
- Structural NAND tightness. Industry commentary frames memory supply growing ~20%/yr against demand spiking far faster (KV-cache/inference dynamics), keeping prices elevated. Sandisk grows bits via BiCS8/BiCS10 nodal transitions rather than expensive greenfield, so incremental capex-to-revenue is falling.
- Balance sheet & capital return. The company is now net cash / zero-debt (paid off the remaining TLB, ~$3.7B cash at Q3) and authorized a $6B buyback — a lever that looks far more attractive after a 40%+ drawdown.
The bear case / risks
- Peak-cycle skepticism. Bears (e.g., Morningstar) argue the NAND pricing cycle peaks around 2028 and turns down by 2029 — i.e., today's ~80% margins are the top, not a new baseline. Bernstein has cited an NBM floor near $0.29/GB as downside protection, but the market is clearly pricing in eventual normalization.
- China competition (CXMT, domestic DUV progress) threatening long-term supply growth and pricing power.
- NBM concentration & "teeth." The Meta-type deals concentrate demand in a few hyperscalers, and skeptics still question how binding the guarantees are — management pushed back hard on this last quarter.
- Bit-shipment optics. Bits were down high-teens sequentially in Q3 as the company built inventory for the Stargate ramp and NBMs; revenue growth has been price-led, so the durability of ASPs is the whole ballgame.
What to watch on the print / call
- Beat magnitude & the FY2027 setup. Does revenue clear ~$8.3B consensus, and — more important — what's the first FQ1'27 guide? Given fiscal-year-end, expect initial framing of FY2027, where NBMs already cover >1/3 of bits.
- Gross-margin trajectory. Can Sandisk hold 79–81%, and how much of it is locked in via NBMs vs. exposed to spot pricing? Any deceleration signal is the key bear trigger.
- NAND ASP commentary for 2H CY2026 — the single most market-moving data point.
- NBM progress: count beyond 5, additional RPO/guarantee disclosure, whether any edge (PC/phone) NBMs get signed, and updated FY27 bit-coverage %.
- Data center / enterprise SSD: Stargate QLC revenue ramp, TLC demand, hyperscaler order signals.
- China response: management's framing of CXMT and Chinese capacity — the market wants reassurance.
- Capital return pace: how aggressively the $6B buyback is deployed given the depressed share price.
- BiCS10 ramp / capex: confirmation that bit growth stays mid-to-high-teens via nodal transitions, with capex "up modestly in dollars."
Bottom line
Consensus already assumes another record, beat-and-raise quarter, so the numbers themselves are almost secondary. With the stock having round-tripped ~40% off its highs on China-supply fears, the report is really a referendum on margin durability and the credibility of the NBM "no-more-cyclicality" thesis. Watch gross-margin guidance, ASP commentary, and any expansion of the contracted-bit base — those, not the headline EPS, will decide which of the two narratives (structural AI memory winner vs. late-cycle commodity) the market runs with. Expect an outsized move either way.
Sources: Sandisk FQ3'26 earnings release and conference call transcript (April 30, 2026); company FQ4'26 guidance; Zacks/Trefis/Morningstar/Hudson Labs previews; and July–August 2026 market news (Barron's, Benzinga, 24/7 Wall St., Seeking Alpha, Simply Wall St). Figures are Non-GAAP unless noted. This is not investment advice.