The cheapest-looking number on SanDisk’s page is the one that should worry you most. At the close on 2 September 2026 SNDK traded at $1,553.40, which is roughly 8.6 times the earnings its own guidance implies for the year ahead — a discount to the broad market, on a share that has risen more than 2,800% in twelve months. In memory, that combination is not a bargain signal. It is the late-cycle signature, where the multiple compresses precisely because the market distrusts the denominator. Sandisk Corporation earned $73.76 a share in the fiscal year to 3 July 2026 against a $11.32 loss the year before, and guided to $44.00–$46.00 for the December quarter alone. This SanDisk stock prediction therefore brackets the spot price rather than chasing it: a $2,650 bull case, an $1,850 base case and a $780 bear case.
Here is the part a target price on its own cannot tell you, and it is the reason this piece is built the way it is. Both scenarios use broadly the same valuation multiple — around 13 times in the bull case, around 20 times in the bear. The entire 3.4-fold spread between $780 and $2,650 comes from one variable: where NAND contract pricing sits when Sandisk converts its contracted order book into revenue. The company itself quantified the point in its fiscal fourth-quarter release: sequential revenue growth came “approximately one-third from higher volumes and two-thirds from higher pricing”. So the argument that SNDK is cheap on 8.6 times is really an argument about price per bit eighteen months out. Having tracked this cycle since the Western Digital separation completed on 21 February 2025, that is the only question I would spend time on.
Key facts
- SNDK closed at $1,553.40 on 2 September 2026, up 1.08% on the day — stockanalysis.com
- FY2026 revenue $20.25bn, up 175%; GAAP net income $11.43bn; diluted EPS $73.76 — Sandisk FY2026 Form 10-K, 17 August 2026
- FQ4 2026 gross margin 84.6%, against 26.2% in the same quarter a year earlier — Sandisk FQ4 release, 5 August 2026
- Remaining performance obligations at 3 July 2026: $59.8bn, entirely long-term customer agreements; a further $31.3bn of such contracts signed after the balance-sheet date — Form 10-K, Notes 4 and 17
- Long-term debt zero at 3 July 2026, from $1.83bn a year earlier; cash $4.76bn — Form 10-K balance sheet
- FY2026 capital expenditure $177m on $20.25bn of revenue — 0.87% of sales — Form 10-K, investing activities
- Board approved an additional $14.0bn buyback on 5 August 2026, taking remaining authorisation to $15.5bn — Form 8-K, Item 8.01
What is actually happening, and why the numbers look implausible
Sandisk does not look like a company that grew revenue 175% in a year. It looks like one that repriced. Revenue went from $7.36bn in fiscal 2025 to $20.25bn in fiscal 2026, but gross margin went from 30.1% to 71.5% over the same period — and inside the year the ladder was steeper still: 50.9% in the December quarter, 78.4% in the March quarter, 84.6% in the June quarter. Costs did not fall by 54 percentage points. Prices rose.
The mechanism is straightforward once you see the structure. Sandisk buys every flash wafer it sells from Flash Ventures, a set of three joint ventures with Kioxia in which it holds a 49.9% interest, spanning eight fabrication facilities in Japan — six at Yokkaichi and two at Kitakami. Wafers transfer at cost plus a small mark-up, and output is split roughly 50/50. Because the fabs sit inside the joint venture rather than on Sandisk’s balance sheet, the company’s own capital expenditure in fiscal 2026 was $177m against $11.67bn of operating cash flow. That is a memory business with the capital intensity of a software company: 0.87% of revenue.
Think of it as a toll road built with someone else’s money. When traffic and toll rates both rise, almost all of the increment falls through to cash, which is why Sandisk generated roughly $11.5bn of free cash flow in a single fiscal year and retired $1.83bn of long-term debt to zero. The same structure runs in reverse. The 10-K is blunt about the clause that does it: Sandisk is “contractually obligated to pay for 50% of the fixed costs of Flash Ventures regardless of whether we order any flash-based memory”. Asset-light on the way up; operationally geared on the way down. That single sentence is the bear case, and it is written by the company.
The demand side has genuinely changed shape rather than merely inflating. Datacenter revenue reached $5.15bn in fiscal 2026, up 437% year on year, and $2.98bn in the June quarter alone — a third of group revenue from a segment that produced $213m in the comparable quarter of 2025. That shift is the reason the pricing move has lasted longer than sceptics expected, and it echoes what we have seen across the memory complex, from DDR5 contract prices rising almost 500% as hyperscalers book 2027 supply to the NAND forecasts tracked in our TrendForce coverage. “This quarter marks a fundamental inflection point for Sandisk — where our technology leadership is enabling a deliberate shift in our mix toward the highest-value end markets, led by Datacenter,” said David Goeckeler, Chairman and Chief Executive of Sandisk, in the company’s fiscal third-quarter release on 30 April 2026.
What Sandisk and its peers are actually doing about it
The most consequential corporate response is not a product. It is a contract type. Sandisk has been converting spot customers into what it calls New Business Models — multi-year agreements that commit the customer to a stated volume and Sandisk to deliver it, with pricing split between fixed and variable components and backed by financial guarantees. Five were signed by the April earnings call; five more followed, three of them with new customers. Then, after the fiscal year closed, two further agreements with an aggregate transaction price of $31.3bn.
The scale of that book is easy to miss because it sits in a revenue note rather than a headline. At 3 July 2026, the transaction price allocated to remaining performance obligations was $59.8bn, of which $58.7bn had not yet been billed, and approximately 19% is expected to be recognised as revenue over the following twelve months. Add the post-quarter agreements and the contracted book is around $91.1bn — some 4.5 times fiscal 2026 revenue. Sandisk also holds $1.5bn of customer security deposits on its balance sheet as refund liabilities, and reports a further $5.0bn of collateral lodged with third-party financial institutions that it could draw on in the event of customer default.
That is a materially different disclosure from anything in the memory sector’s recent past, and it is the strongest single argument against treating SNDK as a pure commodity cycle. It is also not a hedge against price. The agreements de-risk volume and counterparty credit; the 10-K states the pricing mechanisms consist of “fixed and variable components”, and management’s own framing is that the model “reduce[s] certain elements of industry cyclicality” — not that it removes them.
On capital, the response has been unusually direct. Sandisk spent $4.5bn buying back 3.0 million shares during fiscal 2026, an average of roughly $1,512 each — within 3% of where the stock trades today. On 5 August the board authorised a further $14.0bn, lifting remaining authorisation to $15.5bn, or about 6.8% of the current market capitalisation, and said it expects to fund repurchases from operating cash flow. Peers have moved the same way: SK Hynix announced a $28.6bn buyback after its shares fell 9.8%. When several cyclicals simultaneously decide their own equity is the best asset available, it usually says more about their view of replacement cost than about their view of the next four quarters.
Management frames the same capital decision around durability rather than the cycle. “We closed fiscal 2026 with a leading technology portfolio, established datacenter as a key growth pillar, and deepened our customer partnerships,” Goeckeler said on 5 August 2026. “Our technology and products are well positioned to create value for our customers and generate growing and durable free cash flow.” The word doing the work in that sentence is “durable”, and it is exactly the claim the next four quarters will test.
The valuation maths, and where the $2,650 and $780 come from
Start with what is fixed. There were 146,419,001 shares outstanding at 7 August 2026, so at $1,553.40 the market capitalisation is about $227.4bn. Net cash and marketable securities of roughly $6.5bn — $4.76bn of cash plus a $1.78bn holding in Nanya, listed in Taipei — put enterprise value near $220.9bn. Trailing GAAP earnings of $73.76 give a P/E of 21.1. Annualising the mid-point of December-quarter guidance of $45.00 gives 8.6 times. Free cash flow of about $11.5bn is a 5.1% yield on the equity. Book value is $107.47 a share, so the stock changes hands at 14.5 times book.
Now the scenarios. Neither is a forecast for a particular date; both are twelve-month levels.
| Bear — $780 | Base — $1,850 | Bull — $2,650 | |
|---|---|---|---|
| Move from $1,553.40 | −49.8% | +19.1% | +70.6% |
| Earnings assumption | ~$39 normalised | ~$180 (guidance run-rate) | ~$200 FY2027 |
| Multiple applied | ~20× | ~10× | ~13× |
| What has to happen | NAND pricing reverts, gross margin falls to roughly 40% on ~$22bn of revenue | Pricing plateaus near current levels; the contracted book converts as scheduled | Datacenter mix keeps lifting margin while the $91.1bn book converts at contracted prices |
| Balance-sheet support | Zero debt, $4.76bn cash, $15.5bn buyback authorisation | ~$11.5bn annual free cash flow | 19% of $59.8bn recognised within twelve months |
Both levels bracket the spot price deliberately, and both are drawn from levels this stock has already visited. The bear at $780 is a 49.8% fall — smaller than the 56.5% drawdown SNDK delivered between its highest close of $2,335 on 25 June 2026 and $1,015.89 on 29 July, a round trip that took five weeks. The bull at $2,650 is 13.5% above that June record. For context, the consensus of 24 analysts polled by S&P Global carries an average target of $2,125, a low of $1,000 and a high of $3,600, so our bear sits below the street’s most cautious analyst and our bull well inside its most optimistic. That is intentional: sell-side targets in memory rarely model the fixed-cost clause.
The comparison that matters most is with the fab owners. Micron and Western Digital, Sandisk’s former parent, both carry their own capacity. Sandisk does not, which is why its incremental cash conversion in an up-cycle is extreme and why its downside is a fixed-cost obligation rather than an impairment.
The structural tensions that could break either case
Three risks sit outside the price of NAND, and all three are disclosed rather than speculative.
The first is the joint venture itself. Sandisk cannot unilaterally direct Flash Ventures’ activities, must fund 49.9% to 50.0% of capital investments the ventures decide to make where their own cash flow is insufficient, and is restricted from manufacturing flash outside the ventures. The three entities were extended in January 2026 and now co-terminate on 31 December 2034. Any misalignment with Kioxia on the technology roadmap or on capacity additions transmits directly into Sandisk’s cost base, and the 10-K explicitly flags that Kioxia’s ownership and capital structure “have changed in recent years and could continue to change”.
The second is trade policy. The company lists “evolving trade policies, tariff regimes and trade wars” among its principal risks, which for a business whose entire wafer supply is manufactured in Japan and whose demand is concentrated in US and Chinese datacentres is not boilerplate. A tariff or export-control change that alters where AI storage can be shipped would hit volume and price simultaneously.
The third is the accounting question buried in the good news. A $91.1bn contracted book with $1.5bn of customer deposits and $5.0bn of third-party collateral is a genuine asset, but it also means a growing share of reported revenue depends on customers honouring multi-year commitments through a downturn they did not forecast. Sandisk notes that no customer accounted for more than 10% of net revenue in fiscal 2026, 2025 or 2024, which limits single-name exposure — but the agreements are concentrated in datacentre and edge buyers whose capital budgets move together. The precedent is on file: in fiscal 2025 Sandisk took $75m of underutilisation charges and $24m of inventory write-downs when supply and demand fell out of balance, and it recorded $91m of inventory write-downs in fiscal 2026 even at peak margin.
Trading structure adds a fourth, softer pressure. SNDK has become one of the most heavily traded names in the offshore equity-perpetual market, where open interest reached $1.73bn — the largest of any single-stock perpetual. Leveraged positioning of that size does not change the fundamentals, but it does explain why a 25% single-day move in either direction has stopped being remarkable in this ticker.
What happens next
Three things are worth watching, in order of how quickly they resolve.
First, the December-quarter print, due in early November 2026 on the cadence Sandisk has kept since its 6 November 2025 release. Guidance is $10.30bn–$10.80bn of revenue and $44.00–$46.00 of non-GAAP earnings per share. The company beat the top of its June-quarter revenue guidance by 8.7% and the top of its earnings guidance by 18.9%, so the market is unlikely to reward an in-line result. What matters is the split management gives between volume and price: another quarter weighted two-thirds to price would support the bull case, while a volume-led beat at flat pricing would argue the base case is already the ceiling.
Second, the pace of the buyback. A $15.5bn authorisation funded from operating cash flow is roughly 6.8% of the shares at today’s price. If Sandisk executes it quickly, management is signalling that it regards $1,553 as below intrinsic value; if it slows, the fiscal 2026 average purchase price of $1,512 starts to look like a ceiling management set for itself.
Third, the conversion schedule of the contracted book. Approximately 19% of the $59.8bn is expected to land within twelve months, or about $11.4bn. Watch contract liabilities and refund liabilities in the quarterly balance sheet: rising deposits mean new agreements are still being signed, while a plateau would be the first sign that customers have stopped pre-committing supply.
Our base case is $1,850, reached by applying a low-double-digit multiple to guidance run-rate earnings. It is a modest premium to spot precisely because we do not think the market is mispricing Sandisk so much as pricing a wide distribution of outcomes. That distribution — not a point estimate — is what the $780 and $2,650 levels are for.
Frequently asked questions
What is the SanDisk stock prediction for the next twelve months?
Our bull case is $2,650 and our bear case is $780, against a spot price of $1,553.40 on 2 September 2026, with a base case of $1,850. The bull assumes datacentre mix keeps gross margin near current levels while the contracted order book converts; the bear assumes NAND pricing reverts and gross margin falls to roughly 40%. This is analysis, not investment advice.
Why does SNDK trade on such a low forward P/E?
Because the earnings are cyclical. Annualising the mid-point of December-quarter guidance of $45.00 a share gives about 8.6 times at $1,553.40, but that assumes NAND pricing holds. Trailing GAAP earnings of $73.76 put the stock on 21.1 times. In memory, a single-digit forward multiple at a cycle peak usually reflects doubt about the earnings, not a valuation opportunity.
How much of SanDisk’s revenue is already contracted?
At 3 July 2026 the transaction price allocated to remaining performance obligations was $59.8bn, entirely under long-term customer agreements, with roughly 19% expected to be recognised within twelve months. Two further agreements worth $31.3bn were signed after the balance-sheet date, taking the contracted book to about $91.1bn, or 4.5 times fiscal 2026 revenue.
Does SanDisk own its own fabs?
No. All of its flash wafers come from Flash Ventures, three joint ventures with Kioxia in which Sandisk holds 49.9%, spanning eight facilities in Japan. That is why fiscal 2026 capital expenditure was only $177m. The trade-off is that Sandisk must pay half of Flash Ventures’ fixed costs regardless of how much output it takes.
What is the biggest risk to the bull case?
A reversal in NAND contract pricing. Sandisk disclosed that roughly two-thirds of its June-quarter sequential revenue growth came from price rather than volume, so gross margin at 84.6% is doing most of the work in the earnings. Combined with the obligation to cover half of Flash Ventures’ fixed costs, a pricing reversal compresses margin far faster than volume alone would.
When does SanDisk next report results?
Sandisk has filed its quarterly results 8-K on 6 November 2025, 29 January 2026, 30 April 2026 and 5 August 2026. On that cadence, fiscal first-quarter 2027 results are due in early November 2026.
This article is analysis and information only. It is not investment advice, and it is not a recommendation to buy or sell any security. Figures are drawn from Sandisk Corporation’s Form 10-K for the fiscal year ended 3 July 2026, its Form 8-K filings of 30 April 2026 and 5 August 2026, and closing price data from stockanalysis.com as of 2 September 2026. Prices and estimates change; verify before acting.