Analysis

SanDisk Beat by 372% and Still Fell — Because Guidance, Not Growth, Is What the Record Now Prices

Revenue tripled, and the stock dropped anyway, on forward guidance below estimates — the same pattern the dossier's own price record shows the Korean memory names running through weeks earlier.

SNDK
Coverage intelligence, not investment advice — methodology.

The number, and the number that mattered more

SanDisk’s dossier closes its most recent price data on July 30, 2026, with the stock rallying 26% the same session SK Hynix and Samsung surged in Seoul — a NAND maker moving on the same news as the DRAM makers it doesn’t compete with, evidence the dossier reads as SanDisk’s correlation to the memory complex tightening under stress. The fiscal fourth-quarter print, reported August 5, is the next data point in that same record.

SanDisk reported Q4 revenue up 372% year over year to $8.97 billion, above the roughly $8.48 billion estimate — a beat that, on the revenue line alone, matches the run of triple-digit prints the dossier already tracks through fiscal 2026: 61% growth in Q2, 251% in Q3, now 372% in Q4. Data center revenue came in at $2.97 billion. And shares fell more than 5% after hours, on a first-quarter revenue forecast that came in below estimates.

372%
SanDisk's Q4 fiscal 2026 revenue growth, year over year
to $8.97B, above estimates — shares fell 5%+ anyway, on Q1 guidance below estimates

The pattern the dossier already named

This is the third straight quarter SanDisk has posted a triple-digit-or-near beat and still drawn a mixed-to-negative market reaction. Revenue ran $3.03 billion in the January print, up 61% year over year; $5.95 billion in May, up 251%; now $8.97 billion, up 372% — an accelerating growth rate each quarter, not a decelerating one. The May print flagged a consumer-revenue miss and a pivot to “multiyear customer engagements” language, the same contracted-demand phrasing Micron was using; Q4’s is a clean beat undercut entirely by the forward number instead. SanDisk’s own dossier named this exact print as a watch item before it landed: “Does ‘multiyear customer engagements’ become contracted revenue, or stay a hedge against a demand question nobody has answered.” The August 5 report answers only the first half. Data center revenue of $2.97 billion is a real, disclosed contracted-demand figure — the “multiyear customer engagements” language becoming an actual dollar line rather than staying a phrase. The forecast miss says the hedge language was warranted anyway: the demand question the dossier flagged wasn’t answered by growth accelerating, it was answered by guidance still coming in soft on top of it.

That pattern lines up with the correlation the dossier’s price record already established. SanDisk’s daily-return correlation to Micron ran at 0.75 across its full trading history and rose to 0.92 during the Korean memory selloff and reversal in late June and July — a NAND maker tracking the DRAM/HBM complex tightest exactly when that complex breaks. NAND and HBM are different products serving different parts of the AI stack, and the record does not establish that a NAND maker’s Q1 forecast and a DRAM-side supply constraint are mechanically connected. What the record does establish is that the market has been pricing SanDisk as though they were connected since at least late June, and a below-estimate Q1 forecast gave it another data point to keep doing so.

The same three weeks carried SK Hynix missing its own Q2 estimates on a 257% revenue beat, and Samsung posting a 1,814% operating-profit surge two days later — two very different earnings outcomes from the two Korean DRAM makers, both still read by the market as signals about the memory complex SanDisk’s correlation now tracks at 0.92. A NAND maker’s stock reacting to a DRAM maker’s guidance, and a DRAM maker’s stock reacting to a rival DRAM maker’s miss, is the same mechanism running at two different distances from the product line each company actually sells — the market treating “memory” as one story even where the underlying products, and the companies making them, are not the same.

Sources