Sandisk earned $6.9B. Pricing explains the record quarter

On 5 August, Sandisk reported a quarter that would be difficult to describe as bad. Revenue rose 372% against the same quarter a year earlier. Gross margin went from 26.2% to 84.6%. The company earned $6.9 billion in three months, having lost money over the prior full year.

The explanation that circulated was that guidance disappointed. That is a reasonable-sounding sentence, and it is worth checking against the document, because the filing does not support it in the way the phrasing implies.

The quarter

For the full fiscal year ended 3 July 2026, revenue was $20,248M against $7,355M, and net income was $11,433M against a loss of $1,641M.

An 84.6% gross margin in NAND flash is not a normal number. It is not a normal number in most industries. It is what happens when a commodity is scarce and the seller has capacity.

The guidance was not weak

Here is the part that does not fit the headline. Sandisk guided fiscal Q1 2027 revenue to $10.30–10.80 billion. The quarter it had just reported was $8.965 billion.

Guidance midpoint        $10,550M
Q4 FY2026 actual          $8,965M
──────────
Implied sequential change   +17.7%

Guided non-GAAP diluted EPS of $44.00–46.00 is likewise above the $39.25 the company just delivered. In absolute terms the company told the market that next quarter would be bigger than the record it had just set.

Guidance can still disappoint if analysts expected more, and by several accounts that is what happened. But “guidance was weak” and “guidance was below what people hoped for” are different claims, and only the second one survives contact with the document.

Fabrinet provides another example of why a record year needs a closer reading: its largest customer’s revenue shrank in dollars while the company grew.

What the filing does say about durability

There is one sentence in the release that is more consequential than the guidance range, and it is the company’s own description of where the growth came from:

Sequential revenue growth came approximately one-third from higher volumes and two-thirds from higher pricing.

Two-thirds of the sequential gain was price. Prices in memory are cyclical in a way that volumes are not, and an 84.6% gross margin exists because cost of revenue barely moved while revenue multiplied: $1,383M this quarter against $1,403M a year ago, when revenue was one-fifth the size.

That is the actual question in front of anyone valuing this business, and it is not a question about next quarter. Guidance answers next quarter. Nothing in the release answers the one after that.

The $43 million that is easy to misread

The cash flow statement contains a figure that will distort any automated screen run over this company.

Capital expenditure was 0.48% of revenue in the quarter. For a semiconductor manufacturer that figure is not merely low, it is implausible — and it is not what it appears to be.

Sandisk’s fabrication capacity sits in Flash Ventures, its joint venture with Kioxia, which is not consolidated. The funding shows up elsewhere in the same statement: Notes receivable issuances to Flash Ventures of $123M in the quarter and $462M for the year, plus equity-method losses of $102M and $160M respectively.

A screen that computes free cash flow as operating cash flow minus capex will classify Sandisk as asset-light. It manufactures flash memory. This is the same category of error as a free cash flow figure that changes sign depending on which definition you use — the arithmetic is right and the conclusion is wrong.

Two other items in the quarter are worth separating from operations before drawing conclusions about earnings quality:

• $804M of pre-tax income was a gain on equity securities — a mark-to-market item, not operations. Pre-tax income was $7,849M, of which operating income was $7,037M. These items cut both ways and can dwarf the operating result: Intel booked a $12.5bn mark-to-market loss on its own shares in the June 2026 quarter, turning a $1.8bn operating profit into an $11bn net loss.
• $4,524M of stock was repurchased in the quarter, and the board authorised a further $14B, bringing remaining authorisation to $15.5B.

Limits of this analysis

• These are unaudited results in an 8-K, not a 10-K. The audited annual report will follow and may restate figures. Everything here is as-reported on 5 August 2026.
• The Flash Ventures adjustment is directional, not precise. Notes receivable issuances and equity-method losses are not a drop-in substitute for consolidated capex. The point is that reported capex understates capital intensity, not that these specific figures replace it.
• Consensus estimates are not verified. Guidance is compared with the company’s own prior quarter.

Check it yourself

The earnings release is linked below. The three statements are in it: income statement, balance sheet, and cash flow. Every figure in this article comes from those pages, other than the daily closes, which are labelled where they appear.

If a number here does not match the filing, tell us and it will be corrected on the article and on the corrections log, with the date.

A related guidance comparison is DICK’S 0.9% sales revision and 16.9% EPS revision.