SanDisk Fiscal Q4 2026: Why $1.38 Billion in Costs Matters More Than $8.97 Billion in Revenue
SanDisk closed its fiscal year with June quarter revenue of $8.965 billion, up 51 percent from March and 372 percent from a year ago. Non-GAAP earnings came in at $39.25 a share against company guidance of $30 to $33. The stock fell after ho urs anyway.
The figure worth sitting with is buried further down the income statement. Cost of revenue for the quarter was $1.383 billion. A year earlier, on revenue of $1.901 billion, it was $1.403 billion. SanDisk sold roughly five times as much product and spent slightly less doing it. Almost every incremental dollar of revenue over the past twelve months fell straight through to gross profit. That is the entire story of this fiscal year, and it explains why the argument about this company has very little to do with how well it is run.
Gross margin reached 84.6 percent, up from 78.4 percent in March and 26.2 percent a year ago. Full year revenue was $20.248 billion against $7.355 billion, with net income of $11.433 billion versus a loss of $1.641 billion the year before. That prior-year loss included a goodwill write-down taken when the market was valuing the newly separated business at a fraction of its carrying value. Eighteen months later the company earned more in one quarter than the write-down was worth.
Datacenter revenue hit $2.977 billion in the quarter, roughly double the March figure and up from $213 million a year ago. Edge came in at $5.432 billion. Consumer fell 32 percent sequentially to $556 million, which is the number the bears had circled going in. It reads worse than it is. Consumer grew 29 percent for the full year. What happened in June is that bits got pulled out of retail channels and pointed at hyperscalers, because that is where the pricing was. It is an allocation decision, not a demand problem, though it does mean the consumer business is now a rounding error in a company that used to be defined by it.
On competitive position, the strongest thing SanDisk has is structural rather than technical. Capital expenditure for the entire fiscal year was $177 million on $20.2 billion of revenue. Net property, plant and equipment sits at $674 million. The fabs live inside Flash Ventures, the joint venture with Kioxia, and SanDisk funds them through notes receivable rather than carrying them on balance sheet. Compared with every other memory manufacturer on earth, this is an extraordinarily capital-light way to sell bits, and it is why the free cash flow conversion in an up cycle looks the way it does. The technical position is real too. The BiCS10 generation is sampling at 1 terabit TLC, the QLC bit density work with Kioxia leads the industry, and the High Bandwidth Flash specification developed with SK hynix just cleared its first OCP release.
The catch is that the same structure is a dependency. The partner supplying the wafers is also the competitor whose capacity decisions set the price of the product. SanDisk does not control the supply side of its own market.
And an 84.6 percent gross margin in NAND is not evidence of a moat. It is evidence of a shortage. The honest test of whether this company has built something durable comes down entirely to the New Business Model agreements, of which there are now ten, five signed since the April call including three with new customers. The balance sheet is where they show up. Contract liabilities went from $25 million a year ago to $1.242 billion across current and non-current. Refund liabilities went from $126 million to $1.5 billion. Customers are sending money in advance to lock supply, which is not a thing that has ever happened in this industry at this scale. It also flatters the cash flow statement, and to management’s credit they say so: reported free cash flow of $7.083 billion for the quarter falls to $5.035 billion once you remove $1.938 billion of prepayments and deposits.
Whether those contracts survive contact with a down cycle is unknown. Nobody has run the experiment. Contracts get renegotiated when spot pricing goes to a third of contract pricing, and every memory executive alive knows it.
Which brings us to the September quarter guidance, and to the reason the shares sold off into the evening. Revenue is guided to $10.3 to $10.8 billion with non-GAAP earnings of $44 to $46 and gross margin of 83 to 85 percent. Revenue up about 18 percent sequentially at the midpoint, earnings up about 15, margin flat to marginally lower. This is the first quarter of the cycle in which the company is not guiding margin expansion. It is not a warning and the absolute numbers are enormous. But for a stock whose entire re-rating rests on the claim that these margins are structural rather than cyclical, the first flat outlook is precisely the data the skeptics have been waiting for.
The shares closed Wednesday at $1,427.62, up 10.84 percent on the session, for a market value near $190 billion. They are up roughly 500 percent in 2026 and remain the best performer in the S&P 500 this year, which badly understates how violent the ride has been. At the end of June the stock was up 858 percent and trading near $2,350. July removed 47 percent and more than $150 billion of market value, the worst month since the separation from Western Digital in February 2025. Then it rallied about 25 percent across five sessions going into this report, with options pricing a move of roughly 15 percent in either direction. Everyone knew the quarter would be enormous. That was the problem.
Street targets tell you how unresolved this is. Consensus sits near $2,030, with the average of the buy-rated coverage closer to $2,200. Susquehanna carries $3,050. Bank of America is at $2,500. Wells Fargo is Equal Weight at $1,620. A spread that wide on a company this large is not a disagreement about the model. It is a disagreement about whether the model applies.
The base case is that pricing holds through December, contracted volume grows as a share of shipments, and fiscal 2027 lands somewhere around $170 to $185 in earnings per share. At the current price that is roughly eight times. The stock grinds toward $1,800 to $2,200 as the market gradually concedes the earnings are real. The catalyst is the August 13 investor day, specifically whether management is willing to put a number on how much of fiscal 2027 bit supply is under contract and for how long.
The bull case needs that disclosure to be better than expected. If contracted coverage clears half of fiscal 2027 volume with fixed pricing components, and datacenter continues compounding at anything close to the current pace, the market can stop applying a cyclical multiple and start applying a contracted one. That is the path to $2,500 to $3,000, and it is entirely about disclosure rather than operations.
The bear case has almost nothing to do with SanDisk. New NAND capacity arrives, spot pricing rolls over, and the entire memory cohort derates together regardless of what any individual contract book says. Micron, Kioxia, the Korean producers and SanDisk all fall as one, because that is what has happened in every prior cycle and the market has a long memory for it. The catalyst is a capacity announcement from a competitor, not an earnings report from this company.
One number to carry into August 13. The board just authorized another $14 billion of buyback, taking the remaining total to $15.5 billion, against a market value near $190 billion. SanDisk already spent $4.524 billion repurchasing shares during the June quarter, at prices well above where the stock trades tonight. If the guided margins hold, the September quarter alone produces enough cash to fund a third of the new authorization. If they do not, management will have spent its peak-cycle cash buying its own stock at peak-cycle prices, and the treasury line will sit on the balance sheet as the permanent record of the call.