Sandisk Corporation (SNDK) Stock Analysis — October 2026

Sandisk Corporation turns NAND flash, memory that keeps data when power is off, into storage for devices, people and data centers. Its business is a chain: customers pay for products, Sandisk pays for memory and the work of making and selling them, and the difference depends on prices, contracts and supply.

1.1 — What it is and what it sells

Sandisk sells NAND flash memory in forms suited to different jobs, from tiny embedded components to complete storage drives.

NAND flash keeps information without electricity. An SSD, or solid-state drive, combines that memory with electronics, software, a casing and a connection so a computer can use it like a hard drive, but without moving parts. A wafer is earlier in the process: a silicon disc carrying many memory chips before they are separated and packaged.

Sandisk’s largest reported end market in FY2026 was Edge, storage used near the person or machine creating the data. It produced $12,160 million of revenue. Within this market, client SSDs go into PCs, phones, consoles, cars and other devices. Sandisk sells them to device makers, distributors and people upgrading their own machines.

Embedded flash is built permanently into a finished product. A phone or car manufacturer, known as an OEM, buys the component and designs it into the device. Sandisk’s iNAND automotive products are made for temperatures from -40˚C to +105˚C. Because an OEM designs around a particular component, changing suppliers later can require testing and redesign.

Enterprise SSDs serve servers and data centers. Sandisk sells them to cloud operators, server makers and companies running their own infrastructure. Its SN655, for example, is designed for workloads that read large amounts of stored data. Data-center revenue rose 437% in FY2026, helped by higher sales and prices.

Consumers buy the same underlying memory in more familiar forms: portable SSDs, removable memory cards and USB flash drives. A photographer might use a fast card to record and transfer images, or a portable SSD to carry large files between machines. Retailers, online stores and distributors bring these products to buyers.

Sandisk also has an upstream layer involving NAND wafers and components. But its single reported operating segment combines these activities rather than showing separate product businesses. The central business choice is therefore where to place limited memory supply: into data-center drives, embedded components, client SSDs or retail products.

The segment table shows the company’s reported revenue, operating income or loss, net income or loss, and total revenue across the listed years.

Results by segment, as the company reports them

US$ millions

That range of products explains what Sandisk sells; the next question is how sales become profit and cash.

1.2 — How it makes money

Sandisk makes money when the price it receives for storage exceeds the cost of producing, developing and delivering it.

Imagine a photographer buying a portable SSD in a shop. The customer pays the retailer, and perhaps a distributor also takes a share. Sandisk records the wholesale amount it receives, not the full shop price. From that revenue it pays for flash memory, electronics, assembly, testing, delivery, engineering, sales and support.

The unusual cost is the memory itself. Kioxia, a Japanese memory company, works with Sandisk through Flash Ventures, their manufacturing joint ventures. The ventures own or lease production equipment, and Sandisk buys its share of wafers while sharing variable production costs and half the fixed costs. Those factory obligations continue even when customers postpone orders, and Sandisk also helps fund equipment investment.

Gross margin shows what remains after costs assigned to making and selling products. Sandisk’s FY2026 gross margin was 71.5%. After operating expenses such as research, development, sales and administration, operating income was $12,389 million on $20,248 million of revenue. Operating income is therefore not the same as gross profit, and neither measure says when cash arrived.

Sandisk is also replacing some spot-market uncertainty with New Business Model agreements, or NBMs. These multi-year contracts commit customers to product volumes. Their prices contain fixed and variable elements, with floors that limit declines and ceilings that limit increases. Customers get greater confidence about supply, while Sandisk gets a firmer basis for planning production. In August 2026, Sandisk said eight customers had signed agreements backed by minimum financial guarantees.

Cash can arrive before Sandisk has earned the related revenue. An advance is payment for products still owed, so it remains a contract liability until Sandisk performs. A refundable deposit is security that may have to be returned. At July 3, 2026, contract liabilities were $1,242 million and refund liabilities were $1,500 million. Sandisk excluded NBM prepayments and deposits from adjusted free cash flow because they are not ordinary earned sales.

So the dollar flow is more complicated than a sale: customers pay, Sandisk pays its suppliers and workers, and the remainder becomes profit only after the required work and accounting are complete. Those customers, and the routes through which they buy, determine the business’s exposure.

1.3 — Who pays

Sandisk’s immediate customers are mostly businesses, even when the final product is bought by an individual.

Cloud operators, server manufacturers and enterprise customers buy drives for data centers and other infrastructure. They care about capacity, performance, reliability and whether the drive fits a particular system. Once storage becomes part of an infrastructure design, replacing it requires testing. An NBM adds a financial commitment to that technical relationship.

Device makers also pay Sandisk. A PC manufacturer buys client SSDs, while a phone or automotive manufacturer buys embedded flash for a product it will sell under its own name. Distributors buy and resell client products, creating the channel between Sandisk and smaller buyers. At the consumer end, retailers and online stores sell portable SSDs, cards and USB drives to individuals and organizations.

Reported geography follows the ship-to location, not necessarily the final user. FY2026 revenue attributed to Hong Kong was $5,126 million, while revenue attributed to the United States was $3,714 million. A shipment to Hong Kong may enter a factory, distributor or regional supply chain before travelling elsewhere, so these figures do not measure household demand directly.

No customer supplied more than 10% of FY2026 revenue, but the ten largest customers together supplied 44%. No single buyer dominates, yet the largest group still matters. The eight NBM customers announced in August 2026 may not be the same group as the ten largest customers, but their commitments make future delivery and payment obligations important.

Behind this customer network sits a concentrated manufacturing relationship. Sandisk buys substantially all the flash wafers used in its products from the ventures with Kioxia. The ventures operate the Yokkaichi and Kitakami plants in Japan, and their framework runs through December 2034. They sell wafers to Sandisk and Kioxia under commercial arrangements, after which each company makes and sells its own products.

This resembles sharing a bakery: Sandisk sells its own cakes, but depends on shared ovens and still owes part of the costs of keeping them available. A disruption could affect SSDs, cards, embedded products and data-center drives at once. Sandisk’s business therefore rests on two linked promises: customers must buy the storage they commit to, and its manufacturing network must deliver it.

The regional table shows Sandisk’s reported revenue by the listed shipment regions and years.

Revenue by region, as the company reports it

US$ millions

Sandisk Corporation sells NAND flash, a form of digital storage that keeps data when power stops, in chips, drives, cards and other products. Its market is large and growing, but its fortunes depend on where storage demand grows, which rivals can supply it, and whether Sandisk can protect its place when too many bits arrive.

2.1 — The market it sells into

Sandisk’s market is one storage technology sold at several stages, so it must be counted in both dollars and bits without adding those stages together.

NAND flash stores photographs on memory cards, files on solid-state drives (SSDs), and data in cloud servers. A supplier may sell a wafer or chip, or turn that NAND into a finished SSD, card or USB drive. Counting both chip sales and finished-product sales as separate markets would count some storage twice, like counting both the flour and the cake as separate meals.

Dollars show what customers spend; bits show how much storage they receive. A bit is a zero or one, and eight bits make a byte. Revenue can rise because suppliers ship more bits, charge higher prices, or sell a richer mix of enterprise SSDs instead of cheaper cards.

One 2026 forecast valued the global NAND market at $270.6 billion. Sandisk later expected more than $300 billion for the same year, showing how quickly prices and forecasts were moving.

The physical market is also enormous. Sandisk projects 1.2 zettabytes of total flash shipments in 2026; a zettabyte is one trillion gigabytes. But an enterprise data-center forecast of 1.2 zettabytes by 2030 describes capacity demand, not Sandisk revenue. A cloud operator buying an enterprise SSD pays for the drive, which contains NAND, a controller and firmware; servers, networking and cloud services belong to other suppliers. The size of the field is therefore clear, while Sandisk’s addressable share depends on what it actually sells.

2.2 — Where the market is heading

Enterprise data centers are taking a larger share of NAND because AI is creating more storage work, but that share alone does not prove durable growth.

Inference means running an already trained AI model to answer a question or perform a task. Each use can require retrieving documents, loading model information and preserving intermediate results. As cloud providers deploy AI more widely, they are ordering more enterprise SSD capacity.

Capacity and speed do not serve every workload equally. QLC, or quad-level-cell NAND, stores four bits in each memory cell. It suits systems that hold very large collections of information and read them more often than they rewrite them. Sandisk’s SN861 targets demanding mixed workloads, while its SN670 is designed for capacity-intensive storage and reaches 122.88 terabytes.

Enterprise SSDs accounted for 48% of global NAND bits shipped in the second quarter of 2026, up from 26% a year earlier. That is a mix measure, not a growth rate. Enterprise shipments may have increased, consumer shipments may have fallen, or suppliers may have redirected scarce output to enterprise customers. Consumer shortages accompanied the shift.

The important question is therefore whether enterprise customers are consuming more absolute bits and placing repeat orders. Sandisk benefits most when data-center demand expands, rather than merely when consumer supply becomes weaker.

2.3 — The rivals

Sandisk faces different rivals at the chip factory, the enterprise-drive customer and the retail shelf, so no single comparison captures the whole contest.

Samsung Electronics, SK hynix and Micron sell NAND alongside DRAM and high-bandwidth memory, which are working memories used by processors. Their total corporate revenue shows scale, but not the strength of their NAND operations. NAND-specific revenue and bit shares provide a fairer comparison.

In the second quarter of 2026, Samsung held 28% of NAND revenue, SK hynix 19% and Sandisk 11%. By bits shipped, Samsung held 25% and SK hynix 22%. Revenue rewards price and product mix, while bits show physical output. SK hynix’s Solidigm subsidiary is especially relevant because it sells high-capacity QLC enterprise SSDs.

YMTC, a Chinese NAND producer, reached 14% of NAND bits shipped in that quarter. It offers enterprise drives, including the PE310 and PE522, but a product launch does not prove that major customers have qualified those drives for production use.

Kioxia competes for finished-drive customers while sharing manufacturing and development with Sandisk. Kingston, PNY, ADATA, Transcend and Netac compete mainly through branded products and distribution. Phison and Silicon Motion supply controllers and storage engineering to other brands. These are different business models and must be judged by different measures.

The table sets out the companies competing across NAND manufacturing, finished storage, retail brands and controller supply.

The rivals

2.4 — What protects it

Sandisk’s strongest protection is the accumulated ability to manufacture advanced NAND and turn it into storage products that customers trust.

Sandisk and Kioxia operate Flash Ventures, a shared factory system that develops and manufactures NAND wafers. A wafer is a silicon disc containing many future memory chips. Sandisk owns 49.9% of the ventures and Kioxia 50.1%; each generally receives about half the output. Sharing development and factory investment spreads the cost across large production volumes.

The arrangement is also a continuing financial commitment. Flash Ventures supplies wafers to both partners at cost plus a small markup, while Sandisk pays half the ventures’ fixed costs even if it takes less output. The framework runs through December 2034, and Sandisk agreed to pay Kioxia $1.2 billion from 2026 through 2029 for manufacturing services and supply availability.

A complete enterprise SSD is harder to reproduce than a cheap memory chip. Firmware, the software inside the drive, manages errors, wear, temperature, encryption and sudden power loss. Sandisk’s SN655 also offers dual-port redundancy, so storage can remain accessible if one connection fails. Qualification means a customer has tested and approved the drive in its own system.

That approval creates a practical barrier: a cheaper rival still must prove that millions of drives will behave predictably for years. But Kioxia also competes with Sandisk, and the partnership restricts alternatives. The protection is real, but it depends on cost, reliability and continuing customer approval.

2.5 — Consumer brand and portfolio value

Sandisk’s consumer brand helps buyers choose storage, while its broad portfolio gives it several routes into devices and retail channels.

Consider a family expanding a Nintendo Switch 2. Sandisk sells an officially licensed microSD Express card tested and approved for the console. That branding reassures the buyer about compatibility without requiring technical knowledge.

Brand familiarity can support distribution, but consumers can compare Sandisk with Samsung, Lexar and PNY on capacity, compatibility and price. A discount from a rival can quickly weaken any premium for the name. Retail placement matters, but repeat purchases and margins after promotions matter more.

Sandisk’s consumer revenue rose 29% in fiscal 2026. Fiscal fourth-quarter consumer revenue was $556 million, down 32% sequentially and 5% from the year-earlier quarter. Those changes may reflect prices, supply allocation, timing or units sold, so they do not by themselves prove a loss of loyalty or attractive margins.

Sandisk also sells to OEMs, or original equipment manufacturers that put storage inside laptops and other devices. Serving several markets may let it direct products toward the best opportunities, but breadth also brings forecasting, inventory and support costs. The brand and portfolio help Sandisk reach customers; they do not cancel the NAND cycle.

2.6 — The NAND supply and demand cycle

NAND prices can swing sharply because expensive factories are slow to adjust, while new technology can suddenly produce many more bits.

In 3D NAND, memory cells are stacked vertically like floors in a building. Sandisk’s BiCS10 design is moving toward 332 layers. More layers can increase the bits produced from each wafer and reduce cost per bit, but if several suppliers make the same advance, industry supply can expand without many new factories.

A 2026 estimate put the NAND supply deficit at 4–5%, reflecting AI demand and limited capacity expansion. Sandisk’s latest quarter showed that roughly one-third of sequential revenue growth came from higher volume and two-thirds from pricing. That pricing strength indicates tight conditions, but it may not last.

Supply can return through process upgrades, better yields and new equipment in existing buildings. Sandisk and Kioxia began BiCS10 production at Kitakami Fab2, while suppliers continued targeting higher bit growth. If those extra bits arrive before demand catches up, prices can fall even while long-term storage use rises.

Inventories make the cycle harder to read. Manufacturers can hold lean stocks while module makers hold excess finished products because consumer demand is weak. A specialist therefore watches supply and demand in bits, by end market, alongside wafer starts, usable yields, inventories and contract prices.

Sandisk’s customer agreements may soften the downturn, but its lasting returns depend on whether data-center demand absorbs the next wave of NAND before suppliers compete away today’s shortage prices.

Sandisk makes flash storage, the electronic memory that keeps data when power is off. Its history is a record of trading scale and control for flexibility: what it spent bought factories, technology and supply, but the returns have not all been proved.

3.1 — How it got here

Sandisk became a storage company by making an expensive technology affordable and sharing the cost of manufacturing it.

Founded as SunDisk in 1988, the company sold its first 20-megabyte solid-state drive, or SSD, for $1,000 in 1991. An SSD stores data electronically instead of on a spinning disk. The opportunity was that flash memory could keep information without power; the challenge was cost. From the 1990s to 2022, the industry reduced the cost of storing a bit by 50,000 times.

Manufacturing scale changed the economics. In 2000, Sandisk and Toshiba created FlashVision, a joint manufacturing company, so Sandisk could obtain production without financing an entire semiconductor factory alone. During the 2008 downturn, Samsung’s offer of $26 a share was rejected. In 2009, Sandisk sold slightly more than one-fifth of its captive wafer capacity to Toshiba, reducing its exposure just as unused capacity was expensive. NAND prices then recovered sharply, helping the business alongside better industry supply and demand.

Western Digital, a storage company best known for hard drives, completed its purchase of Sandisk in 2016 for an announced equity value of roughly $19 billion. The flash business then shared capital decisions with a larger storage group.

The arrangement ended in 2025. Western Digital shareholders received one Sandisk share for every three Western Digital shares, without paying for those shares or surrendering their existing stock. Sandisk borrowed $2 billion and paid about $1.5 billion to Western Digital for transferred assets and liabilities. It repaid the term loan in 2026. Independence restored control over capital, but began with a cash transfer and a new financing burden—setting up the investment choices that followed.

3.2 — The big decisions

Sandisk’s major investments bought useful capabilities, but they did not all earn a clear return.

The enterprise expansion was mixed. Sandisk bought Pliant and SMART Storage Systems, then completed the purchase of Fusion-io for about $1.1 billion. Fusion-io supplied flash products and software that helped servers reach stored data quickly. Enterprise-solutions revenue rose 10% in 2015, partly because of Fusion-io. But Sandisk also cancelled an acquired development project and wrote off its entire $61 million value. The acquisitions added capability, yet public results do not establish whether they earned enough.

Sandisk also changed what it owned in manufacturing. In 2024, it sold 80% of SanDisk Semiconductor Shanghai to JCET, a chip assembly and testing company, while retaining 20%. The sale produced $659 million before tax and a $34 million accounting gain. It reduced equipment and working-capital demands, but contract manufacturing could raise product costs.

A goodwill impairment exposed a different weakness. Goodwill is the part of an acquisition price left after assigning value to identifiable assets and liabilities. In fiscal year 2025, or FY2025, Sandisk recorded a $1.8 billion impairment because projected cash flows and comparable-company values no longer supported the inherited book value. No new cash left the company, but the charge admitted that part of earlier acquisition spending had lost economic support.

The newest decision protects supply. Sandisk paid $970 million for shares in Nanya Technology, a memory company, alongside a multiyear DRAM supply agreement. NAND stores data; DRAM is temporary working space that helps an SSD organize and retrieve it quickly. The shares were locked up for three years, and their higher quoted value produced an $807 million unrealized gain. That was not cash received. The investment works only if reliable DRAM supply and purchasing costs justify it, so the next question is how Sandisk funds all these commitments.

3.3 — Where the cash went

Sandisk’s cash story moved from funding shortfalls to large returns of capital, but reported cash does not remove every future obligation.

The table follows free cash flow through distributions, dividends, buybacks, acquisitions, asset sales, investments, share issues and debt.

Follow the money: where the free cash flow went

US$ millions

In FY2026, the picture changed: Sandisk generated abundant free cash flow, repaid debt and bought back shares. A debt-free balance sheet removes interest and refinancing risk. Buybacks increase the remaining owners’ claim on future earnings, but they create value only if Sandisk buys shares below what the business will ultimately earn.

Investment continued beside those returns. Sandisk’s share of gross capital expenditure was $562 million, while net cash capital spending was $153 million. External funding, leasing and equipment costs recovered through wafer purchases explain much of the difference. The equipment was not free; its cost moved through the manufacturing ventures and future purchases.

Timing also matters. Fourth-quarter operating cash included $1,938 million of customer prepayments and deposits that were excluded from adjusted free cash flow. That cash arrived early, but Sandisk still owed customers products. The company’s future plan therefore depends on surviving weaker NAND prices while preserving enough cash for factories, components and delivery commitments. That pressure is clearest inside Flash Ventures.

3.4 — Flash Ventures with Kioxia

Flash Ventures gives Sandisk shared access to advanced factories, but it also makes fixed costs difficult to escape.

Kioxia is a Japanese memory company and the successor to Toshiba’s memory business. Its partnerships with Sandisk operate through Flash Partners, Flash Alliance and Flash Forward. Sandisk owns 49.9% of each venture and generally buys half the output. Kioxia operates the Japanese factories, while the ventures own or lease much of the equipment. This arrangement gives Sandisk supply and technology at a scale it would struggle to reproduce alone.

The downside appears when demand falls. Sandisk pays variable production costs and its share of fixed costs, including buildings and equipment, even if it buys fewer wafers. Underutilization charges were $249 million in FY2024. They fell to $75 million in FY2025 and $11 million in FY2026. Fewer purchases can therefore leave the same factory cost spread across fewer chips.

The partnership also commits Sandisk to future supply. The Yokkaichi agreements now run through 2034. Sandisk agreed to pay Kioxia $1.165 billion during 2026–2029 for manufacturing services and continued availability. The companies also announced a combined investment plan exceeding $31 billion in Japan through 2032, dependent on government support. That is not a $31 billion Sandisk bill, but it shows how early capital decisions shape future capacity.

The technology can justify the commitment. Sandisk’s BiCS10 samples had 332 memory layers and 59% greater bit density than BiCS8. More density can lower storage cost, but only if factories produce enough working chips at acceptable cost. Even without corporate debt, Sandisk supported $923 million of venture equipment-lease guarantees and $402 million of future building-depreciation prepayments. Shared scale brings shared obligations, leading directly to the costs of independence.

3.5 — Standalone costs after the spin-off

Sandisk’s results after the spin-off cannot be compared fairly with earlier profits until the old shared costs and new financing are separated.

Before 2025, Sandisk’s accounts were carved out of Western Digital’s records. Direct costs belonged to the flash business, but shared finance, legal, human-resources and information-technology costs were allocated by measures such as revenue or headcount. Those allocations need not equal the cost of running an independent public company.

Pro forma accounts then showed how the separation and financing might have affected earlier periods. For FY2024, pro forma interest expense was $166 million, including a $126 million financing adjustment. The financing structure therefore changed the earnings base before operations changed.

The estimated overhead increase was smaller but still incomplete. Compensation and benefits rose $11 million, while service fees, facilities and other specified changes produced an $8 million net autonomous operating-cost increase. That estimate did not fully quantify permanent reporting, compliance and hiring costs. Sandisk must prove that operating improvements cover these costs and its factory commitments through a full memory downturn.

The central verdict is therefore mixed. Sandisk used capital to gain technology, supply and independence, and FY2026 showed strong cash generation. But impairments, underutilization charges and unproven long-term returns mean that spending has created potential more clearly than it has yet demonstrated durable value.

Sandisk Corporation is a public company that turns NAND flash memory—the semiconductor memory that keeps data when power is off—into storage products. To decide whether its leaders are on shareholders’ side, we need to examine its managers, board, owners and rewards.

4.1 — Who runs it

Sandisk is run by experienced technology executives whose record as leaders of a standalone company is still short.

Sandisk became separate from Western Digital, its former parent, on February 21, 2025. It received the flash-memory business and related operations, while Western Digital kept its own business. That separation gave Sandisk clearer responsibility for its investment, borrowing, cash and results.

David Goeckeler has been chairman and chief executive since 2025. He ran Western Digital from March 2020 until the separation, after leading major networking and security operations at Cisco. His most important decision was therefore to lead the separation itself. His next test is to make the independent company perform through the sharp price swings of the memory market.

Sandisk is trying to make customer relationships more predictable. Its multiyear agreements include firm financial commitments and can bring cash forward, although advance payments remain a liability until Sandisk delivers the products. The arrangement resembles a restaurant taking a deposit for future catering: it improves planning, but it does not guarantee a profitable meal. NAND contract prices fell more than 50% during the latest completed downturn, so these agreements reduce uncertainty without removing market risk. Sandisk reported three such agreements in April 2026. By the August results call, eight customers had signed.

Luis Visoso, the finance chief, must turn those commitments into cash, financing and disciplined spending. His experience includes cloud-economics work at Amazon Web Services and finance leadership at technology companies. By April 2026, Sandisk reported a debt-free balance sheet. That is a concrete result, though the longer test is whether the company can preserve it when conditions weaken.

The officers table shows the named executives and the roles they hold.

Who runs it

The team now has a clear standalone mandate; the board must show that it can supervise that mandate.

4.2 — The board

Sandisk’s board has formal independence and relevant experience, but Goeckeler also chairs the board that oversees him.

The board appoints and evaluates management, approves major decisions and monitors risk for shareholders. Seven of eight directors are independent, with Goeckeler the exception. Independence limits disqualifying financial or employment ties, but it does not guarantee courage, expertise or effective challenge.

The directors bring experience in semiconductor manufacturing, international technology leadership, operations and people management. Rick Cassidy, Tom Caulfield, Devinder Kumar and Necip Sayiner add industry knowledge, while Ellyn Shook and Miyuki Suzuki broaden the board’s leadership perspective. That range matters because Sandisk must manage factories, engineers, global customers and long investment cycles.

Rick Cassidy is the lead independent director. He leads executive sessions in which directors meet without management, creating a place to review strategy, performance and management effectiveness. This is the board’s main counterweight to the chief executive’s influence over meetings and agendas.

The board also added Alexander R. Bradley, effective December 30, 2025, to its audit committee. His industrial-project finance experience is useful where investment decisions can outlast the market conditions that justified them.

The board table shows each director’s tenure and independence status.

The board

The structure can work, but its credibility depends on whether directors challenge generous targets and unfavorable arrangements rather than merely satisfying independence rules.

4.3 — Who owns it

Sandisk has no disclosed majority owner, so influence is distributed among investment managers, insiders and smaller shareholders.

Investment managers such as Vanguard Capital Management, BlackRock, FMR and Jane Street hold influential positions, but a reported institutional holding does not necessarily mean the manager supplied the money. Pension funds and mutual funds may own the economic interest while the manager decides how to vote or whether to sell. That power is called beneficial ownership.

The Fidelity-related positions illustrate the point. Abigail Johnson’s reported position overlaps with FMR’s because voting and selling authority can be divided between related parties. The holdings should not be treated as separate piles of shares. Large managers can influence elections, pay votes and shareholder campaigns without controlling the company.

Sandisk has one class of common stock, and each share carries one vote. A shareholder can vote for directors, participate in advisory votes on executive pay and support proposals, but cannot personally order a factory investment or replace an executive. Those decisions pass through the board and voting process.

The ownership table shows principal holders, insiders, their roles, share classes, values and recent transactions.

Who owns it: insiders and principal holders

Western Digital’s exit created a different concern. It exchanged retained Sandisk shares to reduce debt, and those shares were then sold publicly in secondary offerings. Sandisk paid all the offering expenses but received no sale proceeds. The arrangement created a possible overhang while the former parent’s block entered the market, and Sandisk’s obligation for later exit expenses ended on March 19, 2026. Ownership is dispersed, but that does not mean every separation decision favored continuing shareholders.

4.4 — Insider buying and selling

Sandisk’s insiders have reduced their holdings through substantial sales, with no reported open-market purchases.

During the latest twelve months, officers and directors sold 88,895 shares for US$133,769,706.28 in 76 open-market trades. David Goeckeler accounted for 67,679 shares and US$104,972,834.41. Those sales reduce insiders’ exposure, so they deserve attention, but a sale can reflect diversification, taxes or spending rather than a judgment that the business is failing.

The timing gives Goeckeler’s sales a more precise meaning. He adopted a Rule 10b5-1 plan on May 29, 2026. Such a plan sets trading instructions in advance, often through a broker, before the insider knows material nonpublic information. It resembles arranging automatic monthly sales instead of deciding each month whether to sell. September’s transactions therefore do not by themselves show a fresh bearish reaction, although the earlier choice to adopt the plan still matters.

Luis Visoso and several other directors also sold. Reported insider transactions can include stock awards, shares surrendered to cover withholding and genuine market sales. Those actions have different meanings. The useful question is how much exposure each insider retains after awards and sales, not whether any single transaction looks alarming.

That leaves pay as the next test: does Sandisk reward results while preserving shareholders’ claim on the business?

4.5 — Pay, and promises kept

Sandisk links executive rewards to performance and share ownership, but payout ceilings and dilution make the alignment worth watching.

David Goeckeler’s fiscal 2026 compensation was US$15,181,160, while Luis Visoso’s was US$8,430,409. Those totals include the accounting value of stock awards, not just cash received. Restricted stock units promise shares after continued service; performance stock units also depend on results. The design gives executives a stake in the company, but issuing shares can give existing shareholders a smaller percentage of the same business.

The first year of the fiscal 2026–2028 performance program reached its 200% ceiling for both revenue and adjusted earnings per share. The final award averages three annual scores, so one excellent year cannot determine the entire result. But once a year reaches its ceiling, further performance earns no additional reward for that year. The remaining targets will show whether the program encourages durable execution or mainly rewards a favorable part of the memory cycle.

The table brings together stock pay, buybacks and the company’s share-count measures.

Stock pay and the share count

US$ millions

Sandisk has delivered tangible results, including debt reduction and actual repurchases. At Investor Day on August 13, 2026, management promised to return all excess cash after investing in the business. That promise leaves judgment over what counts as necessary investment, but it gives shareholders a clear principle to test.

The overall answer is guardedly positive: experienced managers have delivered early results and meaningful ownership exposure, while the board and owners provide real but imperfect checks. Whether they remain on shareholders’ side will depend on disciplined investment, tougher performance targets and restraint when the memory market becomes less generous.

Sandisk Corporation makes NAND flash, a form of memory that keeps data when power is off. Its accounts show a company emerging from a severe downturn, with extraordinary growth that still needs to be tested for durability.

5.1 — Growth

Sandisk’s sales growth reflects both more storage sold and more money earned for each unit of storage.

An exabyte is one billion gigabytes. Sandisk’s “exabytes sold” measure therefore tracks physical storage, while ASP, or average selling price, tracks revenue per gigabyte. Revenue can rise because either measure improves, or because sales shift toward more valuable products.

Revenue fell from US$9,754 million in FY2022 to US$6,086 million in FY2023. It recovered by 9.5% in FY2024 and 10.4% in FY2025, when exabytes sold rose 6% and ASP rose 4%. Both volume and price helped.

FY2026 was different. Revenue reached US$20,248 million, up 175.3%, while total exabytes grew only in the mid-teens. Datacenter revenue rose from US$960 million to US$5,153 million. Edge revenue rose from US$4,127 million to US$12,160 million. The result came heavily from pricing and mix, not just physical output.

The year also contained 53 weeks, including a 14-week first quarter. Sandisk began selling some customer testing units instead of giving them away, which changed where their costs appeared. A specialist therefore tests shipments, realized price and customer use before treating the headline growth as durable.

These rows show revenue, margins, profit, EPS and dividends across five years.

Results

US$ millions

The next rows show segment sales, working-capital days and customer concentration.

The company's own measures

US$ millions

5.2 — Profit

Sandisk’s operating economics improved sharply, but the comparison is distorted by a large goodwill impairment in FY2025.

Gross profit is sales left after product costs; gross margin expresses it as a share of sales. Gross margin rose from 7.1% in FY2023 to 71.5% in FY2026. NAND factories have high fixed costs, so weak prices and idle capacity can damage margins together. Underutilization charges fell from US$249 million in FY2024 to US$11 million in FY2026, showing that factory economics improved.

Operating income also includes research, selling and corporate costs. FY2025 reported an operating loss of US$1,377 million, including a US$1,830 million goodwill impairment. Goodwill is the premium paid above identifiable net assets in an acquisition; writing it down reduces accounting profit without using cash at that moment. Removing that single charge would have produced operating profit of US$453 million.

FY2026 operating profit reached US$12,389 million, while company-adjusted operating income was higher still because adjustments remove selected charges. Net income also benefited from an unrealized US$807 million gain on Nanya shares. Taxes then reduced the result through a 12.2% effective tax rate. The specialist therefore separates product economics, operating costs, investment gains and taxes.

These rows show adjusted operating income, EBITDA, net income, EPS and the effective tax rate.

The company's own adjusted measures, and tax

US$ millions

5.3 — Cash

Sandisk now generates substantial cash, but customer advances make conventional free cash flow look stronger than cash earned from completed deliveries.

Cash from operations was only US$84 million in FY2025, and free cash flow was negative US$120 million after capital spending. In FY2026, operating cash reached US$11,671 million and conventional free cash flow reached US$11,494 million. The improvement was too large to explain simply through working-capital collection.

Sandisk’s New Business Model, or NBM, includes deposits from cloud customers securing future NAND supply. Those receipts finance production, but they also create a delivery obligation. NBM prepayments and deposits reached US$2,476 million in FY2026. After removing them and including Flash Ventures activity, adjusted free cash flow was US$8,743 million.

Capital spending also has more than one useful definition because joint-venture funding, leases and wafer economics run through different lines. Inventory days rose from 135 to 178, and the cash conversion cycle lengthened from 136 to 162 days. Inventory deserves attention when prices can fall. Deducting US$232 million of stock compensation leaves “real free cash flow” of US$11,262 million, an allowance for dilution rather than another cash payment.

These rows trace operating cash, working capital, capital spending, free cash flow and stock compensation.

Follow the money: real free cash flow

US$ millions

5.4 — Balance sheet and debt

Sandisk has repaid its reported borrowings, but manufacturing partnerships continue to impose fixed obligations.

At July 2026, Sandisk held US$4,762 million of cash and reported zero debt, compared with US$1,900 million a year earlier. Its current ratio fell from 3.6 times to 2.3 times, but liquidity still strengthened because both current assets and liabilities grew. The quality of those assets matters: receivables and inventory are not the same as cash.

The debt came from a US$2.0 billion borrowing in February 2025, of which US$1.5 billion funded a distribution to WDC. Sandisk repaid the remaining principal in March 2026 and recorded a US$46 million extinguishment loss. Debt and interest exposure therefore fell substantially.

Flash Ventures are Sandisk’s manufacturing partnerships with Kioxia. Utilization means the share of available factory capacity in use. Sandisk reduced its wafer purchases during weaker conditions but still pays half of certain fixed costs and places rolling three-month orders. Underutilization charges fell from US$249 million to US$11 million, but zero debt does not mean zero downturn exposure.

These rows show cash, investments, assets, liabilities, leases, equity and liquidity measures.

Balance sheet

US$ millions

5.5 — The latest quarter

The July 2026 quarter confirms an operating acceleration, but pricing contributed more than volume.

Revenue and gross margin reached new highs, and operating income rose sharply from the year-earlier quarter. Because the recovery had already appeared in earlier quarters, the result is more than the disappearance of one accounting charge.

The eight-quarter sequence shows revenue weakening in early 2025 before climbing through the rest of the year and accelerating in 2026. Management attributed roughly one-third of sequential revenue growth to volume and two-thirds to pricing. That mix makes the quarter powerful but also exposes it to future price reversals.

Growth was uneven across the business. Datacenter and Edge drove the expansion, while Consumer declined. Net income also exceeded adjusted earnings, showing that investment and other items still mattered.

Operating cash and conventional free cash flow were both strong. Adjusted free cash flow was lower after customer prepayments and Flash Ventures activity, but it still showed substantial operating improvement. The quarter establishes momentum, not a normal long-term margin.

These rows show revenue, gross margin, operating income, net income, EPS, operating cash and free cash flow across eight quarters.

The last eight quarters

US$ millions

5.6 — Cost per bit and process transitions

Manufacturing progress matters only when denser designs produce more reliable, saleable storage at lower cost.

A wafer is a disc on which memory chips are made. Cost per bit measures the manufacturing cost of stored information, while yield is the share of chips that pass testing. A wafer can hold more potential storage and still be uneconomic if too many chips fail.

BiCS8, Sandisk and Kioxia’s 218-layer generation, supplied the majority of Sandisk’s bit production in FY2026. That proves meaningful manufacturing adoption rather than mere sampling. BiCS10 reached sampling with 332 layers and a claimed 59% density improvement over BiCS8.

TLC stores three bits per cell; QLC stores four. QLC offers more capacity, but the denser design can make reading and rewriting harder. Sandisk must therefore prove bonding yield, reliability, power, endurance, customer approval and cost. The durable advantage is usable bits produced faster than manufacturing and transition costs rise.

5.7 — Enterprise SSD growth and product mix

Enterprise SSD growth has reached paid shipments, but the accounts do not yet prove the margins will last.

An enterprise SSD is a solid-state drive for servers and storage systems. It combines NAND with a controller, which manages data placement, and firmware, the software that directs the drive. A hyperscale customer operates enormous cloud infrastructure and tests the complete product under demanding workloads.

That testing is called qualification. It can take months because failures after deployment are expensive. Once a product passes, switching suppliers may require another round of testing, giving the approved supplier some protection.

Sandisk’s datacenter revenue rose from US$325 million in FY2024 to US$5,153 million in FY2026. Its QLC Stargate platform began shipping for revenue, which marks commercialization rather than merely an engineering announcement.

Yet Sandisk ranked fifth by enterprise SSD revenue. Growth therefore proves progress, not leadership. The accounts also do not disclose Datacenter operating margins, so they cannot show whether controllers, firmware, qualification and support costs leave a durable return.

5.8 — Customer qualification, mix and pricing power

Sandisk’s realized pricing depends on customer approval, product mix and incentives, while channel inventory remains a hidden risk.

Storage products are not interchangeable. A cloud customer that has qualified one drive cannot necessarily replace it immediately, because performance and software behavior must be tested again. That creates commercial value, but not permanent pricing power.

Mix also matters. Enterprise products can command more per gigabyte than retail storage. In FY2025, Edge revenue rose despite lower capacity shipments because ASP improved, while Consumer revenue barely changed as higher volume met lower ASP. In FY2026, Consumer revenue rose even as its capacity volume fell, again showing that dollars and gigabytes can tell different stories.

Sales incentives and marketing programs fell from 19% of gross revenue to 11%. That supports better realized pricing, although it may also reflect a different customer mix. Top-ten customer concentration rose from 40.0% to 44.0%, making large programs more important.

Channel inventory is stock held by distributors and retailers before users buy it. Strong shipments can therefore fill shelves without proving sell-through. Sandisk needs replenishment, stable allowances and repeat orders to show that qualification has become durable demand.

SanDisk Corporation makes NAND flash storage, a memory that keeps data when power is off, and its shares now price in a dramatic recovery. The question is whether that recovery has become durable cash-earning power or whether unusually strong NAND profits are making the stock appear cheaper than it is.

6.1 — What an investor pays today

SanDisk looks inexpensive on forecast earnings only because forecast earnings assume a much stronger business than trailing earnings show.

A share price is not the value of the company. Market value adds up all common shares, while enterprise value adjusts that figure for cash and debt to show what investors are paying for the operating business. The forward P/E compares the price with predicted earnings, so its apparent cheapness depends on those earnings arriving. Enterprise value divided by revenue adds another warning: sales matter only if they become profit and cash.

Free cash flow means cash left after operating needs and spending on property and equipment. SanDisk’s calculation is more complicated because Flash Ventures affects cash movements, while customer advances and deposits are removed. Customers may pay before delivery, but that cash is still a liability until SanDisk performs.

SanDisk and Kioxia jointly use factories through Flash Ventures. Kioxia owns the facilities and provides manufacturing services; SanDisk owns 49.9% of each venture, receives about half the output, and pays its share of variable and fixed costs. It may also have to fund investment when the venture lacks cash. That obligation makes accounting profit less informative without the cash-flow notes.

The float is the stock readily available for public trading. Western Digital’s sales or exchanges can increase that supply without creating new shares, while buybacks retire shares and reduce it. SanDisk had $15.5 billion of repurchase but had repurchased $4.5 billion during fiscal 2026. The table gathers the price, valuation multiples, cash yield, book value, and dividend income that investors face today.

What an investor pays today

US$ millions

6.2 — Against its own past

SanDisk’s independent trading history is too short to establish a normal valuation range.

SanDisk began independent trading in February 2025, leaving only one post-spin-off year-end in its history. Earlier figures cannot show what investors would have paid for this separately listed company, and the old SNDK ticker belongs to a different ownership and balance-sheet history.

The business did give investors a reason to reassess it. Gross margin is the share of sales left after direct production costs. It rose from 30.1% in fiscal 2025 to 71.5% in fiscal 2026, reaching 84.6% in the fourth quarter. More revenue therefore remained available for operating expenses and profit.

Bits measure storage capacity, while ASP means average selling price. If SanDisk sells the same amount of storage at a higher ASP while costs change slowly, revenue rises faster than costs and profit expands. Management said one-third of fourth-quarter sequential revenue growth came from additional bits and two-thirds from pricing. That mix makes the recovery powerful, but also exposes it to a reversal in NAND prices.

NAND prices fell by more than 50% during the 2022–2023 downturn. The recent margin recovery may therefore reflect both better customer demand and a cyclical shortage. SanDisk’s own history cannot settle that question; comparisons with rivals offer the next test.

Five years of history

US$ millions

6.3 — Against rivals

SanDisk’s rival multiples provide context, but they do not create a simple cheapness ranking.

Kioxia and Micron are closer comparisons because they make memory, yet Micron and SK hynix also sell DRAM, the memory used for active computation. Samsung combines memory with a much broader electronics business. Comparing SanDisk’s forecast earnings with rivals’ trailing earnings would also mix two different periods.

P/E is especially unstable in NAND. When ASP rises, profit can grow much faster than storage volume, so the P/E falls even if the share price does not move. When prices fall, the multiple can rise sharply as earnings collapse. A specialist therefore examines physical demand, selling prices, and margins across a cycle rather than treating one year’s P/E as a lasting measure.

EV/EBITDA compares enterprise value with operating earnings before interest, tax, depreciation, and amortization. Depreciation spreads equipment costs over time, but adding it back does not make replacement equipment free. SanDisk must pay half of Flash Ventures’ fixed costs and may fund capital investment. Its share of lease guarantees was $923 million at July 3, 2026.

Price-to-book has a similar problem. SanDisk records Flash Ventures through the equity method instead of consolidating every factory asset and liability. Silicon Motion designs controllers, while Phison, ADATA, and Transcend operate different kinds of component and module businesses. Their book values therefore support different profit engines. The table shows the companies, ownership status, revenue, margins, market values, and trailing P/E figures available for comparison.

The rivals

6.4 — How the price moved

SanDisk’s enormous rise reflects a reassessment of earnings power, while its sharp reversals show how demanding expectations have become.

The performance table shows a stock that rose dramatically over one year but also suffered substantial recent declines. The 52-week range shows the same tension: a powerful advance followed by a large retreat from the high. Such movement suggests that investors are revising both expected earnings and the multiple they will pay for them.

The main business catalyst was a recovery in NAND pricing alongside stronger datacenter demand. Datacenter customers need storage for large datasets and systems that repeatedly retrieve information. When demand meets constrained supply, manufacturers can charge more, and higher prices can move quickly into profit.

A second catalyst was SanDisk’s addition to the Nasdaq-100, an index tracked by funds that must buy newly included companies. SanDisk joined before trading opened on April 20, 2026. That demand can lift the share price without changing the amount of NAND SanDisk sells.

The risk is that strong results may still disappoint if investors expected even more. Consumer affordability may limit demand, while supply tightness could ease in the second half of 2027. The tables show the market’s reaction; the final question is what that reaction assumes about future cash flow.

The share price today and over 52 weeks

US$

Share price performance

6.5 — What the price implies

Today’s valuation depends on SanDisk turning exceptionally high NAND margins into repeatable cash earnings through a downturn.

The forward multiple implies about $30.7 billion of next-twelve-month earnings. But forecast earnings can make a cyclical stock look cheap. A shop bought for $100 and expected to earn $14 next year sells for about seven times earnings; if a normal year produces only $5, the same purchase price represents twenty times normal earnings.

Management’s August Investor Day model targets mid-to-high-teens revenue growth for fiscal 2028–2030. It also targets about 80% non-GAAP gross margin and 75% non-GAAP operating margin. Gross margin subtracts production costs; operating margin also subtracts research, sales, administration, and other operating expenses. Non-GAAP means management excludes selected costs, which does not make them disappear.

New Business Model agreements provide some protection. These multiyear contracts combine customer volume commitments with fixed and market-linked pricing. A floor limits how far prices can fall, while a ceiling limits how far they can rise. Eight customers had signed agreements covering about half of fiscal 2027 bits and two-thirds of fiscal 2028 bits. Their $93.9 billion contract value is spread across the agreements’ lives, not received as one year’s revenue.

The cash test is demanding. A 10% annual cash return on today’s equity value would require about $21.5 billion. At a 50% adjusted cash-flow margin, SanDisk would need roughly $43 billion of annual revenue, compared with $20.248 billion in fiscal 2026. The business was already approaching that annualized scale in its fiscal first-quarter guidance, but one quarter is not a full-year forecast.

The price therefore assumes that contract floors, growing storage demand, and high margins survive easier supply. If customers regain cheaper alternatives, the stock’s low forward multiple may simply be measuring peak earnings.

Sandisk Corporation makes flash storage, the memory used in solid-state drives and other data devices. Its plan depends on selling much more of it at high margins while paying for new factories, technology and customer commitments.

7.1 — What management has guided

Management expects a strong fiscal Q1 2027 and several years of unusually profitable growth.

For the next quarter, Sandisk forecasts revenue of $10,300–$10,800 million and non-GAAP earnings of $44.00–$46.00 per diluted share. Non-GAAP is management’s version of profit after removing selected items from GAAP, the standard accounting result. The exclusions include stock-based compensation: issuing shares may not require an immediate cash payment, but it dilutes existing owners. The forecast is therefore useful for judging operations, but it should be checked against GAAP profit, share count and cash flow.

Guidance in force

US$ millions

Management’s longer-range model targets average non-GAAP gross margins of 80.0%, operating margins of 75.0% and adjusted free-cash-flow margins of 50.0% across FY2028–FY2030. These are averages, not promises for every quarter. Free cash flow is the money left after operating needs and investment; the adjusted version also removes company-specific items.

The main protection is Sandisk’s New Business Model. Longer contracts commit customers to specified purchases, while fixed and variable pricing includes minimum and maximum prices. Deposits and other guarantees improve visibility, but they are not automatically profit: Sandisk must still manufacture and deliver the storage. The contracts can soften a price collapse, not eliminate factory or demand risk. That risk determines how much of the promised excess cash can safely be returned to shareholders.

7.2 — What analysts expect

Analysts expect another strong quarter, so delivering the current forecast matters more than producing another isolated earnings beat.

Consensus means the collected forecast of analysts who follow Sandisk. It changes as analysts revise their models, and different providers can calculate it differently. For fiscal Q4 2026, one estimate put expected non-GAAP earnings at $34.59 a share. Sandisk delivered $39.25, while revenue reached about $8,970 million against an expected $8,420 million.

This was part of a longer run. Sandisk beat both revenue and earnings estimates in all six reported quarters from fiscal Q3 2025 through fiscal Q4 2026. That record shows strong recent execution, but it does not yet show how the newly independent company performs through a complete memory-industry cycle.

For fiscal Q1 2027, the cited consensus was revenue of $10,620 million and earnings of $44.21 a share. Revenue expectations sat above the middle of management’s range, while earnings expectations sat below it. So the guide cannot be called simply optimistic or disappointing; the answer depends on the measure and the estimate used.

The useful test is what supports the result. More shipments at durable prices would strengthen the investment plan. Fewer shares or generous exclusions would offer less reassurance. Those questions lead directly to the production spending required to make the forecast possible.

7.3 — The spending ahead

Sandisk must spend on manufacturing capacity and technology before the growth in its forecast can produce cash.

Kioxia, Sandisk’s manufacturing partner, operates the Flash Ventures joint ventures with it. The ventures produce wafers, the thin silicon discs from which memory chips are made, and generally divide output between the partners. Sandisk pays its share of fixed costs and must fund 49.9%–50.0% of capital investment when a venture cannot pay from its own cash. In effect, Sandisk partly co-owns and partly rents a factory whose costs continue even when demand weakens.

An agreement signed January 29, 2026 extended this arrangement through December 31, 2034. Sandisk also committed about $1.2 billion during 2026–2029 for manufacturing services and continued supply access. K2 at Kitakami in Japan began production during FY2026, and tenth-generation production was announced there in July.

Management expects FY2027 capital spending to rise as it expands BiCS8 and BiCS10. BiCS is Sandisk’s three-dimensional NAND architecture; TLC stores three bits in each memory cell. BiCS10 TLC promises 59% greater bit density than BiCS8, meaning more storage from the same wafer area. That can lower the cost of each gigabyte, but only if early production achieves good yield and customers approve the product.

The timing matters. Equipment may be installed before production becomes efficient, and chips may be made before customers accept them. Sandisk therefore has to finance the manufacturing system before the hoped-for revenue arrives, which makes its cash position the next question.

7.4 — Can they pay for it

Sandisk can support its near-term commitments, but no debt does not mean no unavoidable payments.

As of July 2026, Sandisk held cash and no debt, with no debt due in each of the next five years. Cash had risen substantially from January through July, so the immediate financing problem was production investment rather than refinancing a loan.

Cash and debt, last eight quarters

US$ millions

Debt coming due after the latest fiscal year end

US$ millions

Cash exceeded the disclosed FY2027 requirements by $1,371 million. That comparison is reassuring but not a cash-runway forecast: the commitments fall at different times, and the company continues to receive and spend cash while it operates.

Operating cash generation was especially strong. Fiscal Q4 produced $7,126 million of operating cash flow and $5,035 million of adjusted free cash flow. The adjusted figure excluded $1,938 million of New Business Model prepayments and deposits. Those payments help fund production, but they carry obligations to supply products later, so Sandisk cannot treat every dollar as unrestricted surplus.

A $1,500 million revolving credit facility, available at year-end and expiring February 21, 2030, provides a backstop. The larger exposure remains Flash Ventures: fixed costs continue even if Sandisk reduces purchases. Buybacks are authorized rather than required, so preserving cash for factories is the sensible defence if prices turn down.

7.5 — What to watch

The October results will test execution, but taxes, tariffs and claims will decide how much of the profit becomes spendable cash.

Sandisk’s fiscal Q1 earnings call is scheduled for October 29, 2026, at 4:30 p.m. Eastern. Investors should watch revenue, earnings, shipment growth, realized pricing and the next outlook. Those figures will show whether customers are taking the capacity Sandisk is funding and whether the following quarter still supports the plan.

Taxes create a separate claim. Changes to foreign-income taxation and global minimum taxes affect FY2027, while Sandisk expects corporate alternative-minimum-tax exposure. Malaysian tax holidays expire during FY2028–FY2031, which can raise taxes without weakening factory performance. Cash taxes matter as well as the accounting tax charge because expense recognition and payment can occur at different times.

Sandisk also retains $128 million of tax-indemnity liabilities tied to Western Digital and pre-separation tax positions. U.S. trade investigations could change tariff exemptions; a 10% additional tariff on Malaysian products would either reduce margins or force a price increase customers may resist.

Fiscal FY2026 included $93 million of non-operating legal settlements, with other proceedings unresolved. The essential questions are separate: are products earning the promised margins, are taxes and claims consuming cash, and are manufacturing commitments growing faster than customer receipts? The final question concerns whether new products can make that investment worthwhile.

7.6 — Enterprise SSD growth ahead

Enterprise SSD growth becomes dependable only when customer approvals turn into repeat shipments.

An enterprise solid-state drive stores data in servers. Qualification means a customer has tested that the drive works with its software, servers and workloads. A design win means the customer selected it for a planned system. Neither proves that large orders will follow. The commercial sequence is qualification, selection, volume shipment, deployment and repeat purchase.

Sandisk’s QLC Stargate platform began shipping for revenue in fiscal Q4. QLC stores four bits in each memory cell, allowing high capacity, while Sandisk is also expanding TLC enterprise drives for cloud and artificial-intelligence infrastructure. These products serve different storage needs, so product announcements alone do not establish the size of the opportunity.

The evidence has moved beyond announcements. Fiscal Q4 datacenter revenue reached $2,977 million, up 103% sequentially. High-capacity QLC drives were entering a volume ramp, meaning production was moving from initial shipments toward sustained larger output. The enterprise SSD market is forecast to grow 35 percent annually from 2024 through 2030, and the leading supplier held 35.1% of the market in 2Q26.

Sandisk still needs repeat orders, additional qualifications and a growing share of customer purchases. Those signals show that deployed drives worked, manufacturing scaled and customers chose Sandisk again. Only then will enterprise growth become durable cash capable of funding the next generation of production.

8.1 — The verdict

The forecast's central outcome is modest near-term weakness followed by stronger three-year upside.

The weighted price is $1,331.50 in October 2027, a 10% loss including dividends. It becomes $1,581.80 in October 2028, a 6% gain. By October 2029, it reaches $2,063.61, a 39% gain.

The last-date range runs from $573.23 to $4,471.21. Today’s price is $1,487.33.

The bear case has a 30% probability. The base case has a 52% probability. The bull case has an 18% probability. These odds reflect the tension between strong current demand and an unusually cyclical earnings record.

The table shows the three paths and their probability-weighted result.

What the shares could be worth

USD per share at each date; in brackets the return since today's 1487.33, dividends included

8.2 — The model

The forecast values each date from normal operating profit, a selected enterprise-value multiple, net debt and the projected share count.

That approach fits Sandisk because NAND pricing, utilization and supply additions can move reported earnings sharply. Operating income ranged from a $2,035M loss to $12,389M over the supplied five-year record. The forecast therefore avoids treating FY2026’s 61.2% operating margin as permanent; the five-year revenue-weighted margin was about 19.4%.

Net cash is important because Sandisk had cash of $4,762M and no debt at the latest reported year-end. Cash generation funds investment and buybacks, while weaker pricing can reverse that flow.

The share count changes with repurchases, equity awards and employee share purchases. Sandisk repurchased about 3M shares in FY2026, while 6.6M awards remained outstanding. Dividends stay at zero because every reported year in the supplied history had no dividend, and management has not given dividend-specific guidance.

8.3 — The drivers

Normal operating profit is the main separation between the cases. The base case uses $5,000M in October 2027 and $5,500M in October 2028. It reaches $6,500M in October 2029. That path assumes revenue growth from data-center and enterprise demand, but a 25% normalized margin, above the historical 19.4% and well below FY2026’s peak.

The bear case falls to $3,000M, then $2,500M, before recovering to $3,500M. It assumes supply outruns demand, prices reverse and fixed Flash Ventures costs amplify the decline. NAND prices fell more than 50% in the last completed downturn. The bull case rises from $8,000M to $10,000M and then $12,000M, requiring enterprise demand to absorb new bits and BiCS10 to lower cost per bit. BiCS10 is claimed to improve density by 59% over BiCS8.

The table holds the base share count at 144M. The bear count rises from 147M to 149M and then 151M as restrained buybacks fail to offset new equity. The bull count falls from 143M to 142M and then 141M because stronger cash generation supports larger repurchases. The assumptions reflect FY2026’s $4.5B buyback and management’s policy of returning excess cash after investment.

Dividends remain $0 in all cases. Strong FY2026 free cash flow of $11,494M could support a new payout, but the company has no established dividend habit and returned cash through buybacks instead.

The base multiple rises from 40x to 45x as customer visibility and repeat enterprise orders improve. The bear case compresses from 25x to 20x before a limited recovery, because pricing and utilization weaken. The bull case expands from 42x to 52x, requiring durable demand, successful BiCS10 execution and a more predictable cash stream. Peer reference multiples near 28x to 31x support neither the base premium nor the bull premium without stronger evidence of durability.

Net debt remains negative in every case, meaning net cash. The base case moves from negative $5,930M to negative $7,309M and then negative $9,510M as cash generation exceeds investment and $3,000M annual buybacks. The bear case assumes only $1,000M annual buybacks and working-capital outflows. The bull case assumes $7,500M annual buybacks, leaving negative $5,994M net debt by the final date. The cash build uses a 13% tax rate and depreciation and amortization equal to 3.8% of revenue.

The table gathers the selected assumptions.

The forecast's table

base case, then bear / bull in brackets; USD millions, per share in USD, unless shown

8.4 — Bear and bull

The bear case begins when NAND tightness unwinds before data-center demand can absorb the added supply. Pricing and mix fall, fixed Flash Ventures costs remain, and weaker enterprise execution compresses both profit and the multiple. The price reaches $542.70 in October 2027 and $573.23 in October 2029. Its 30% probability reflects three prior operating-loss years, the documented price collapse and the fact that about two-thirds of recent sequential revenue growth came from pricing.

The bull case requires several linked successes. AI and cloud demand must absorb new bits, BiCS10 must ramp with competitive yields, and enterprise SSD qualifications must become repeat orders. Strong cash generation then funds investment and buybacks while the market sustains a premium multiple. The price reaches $2,375.73 in October 2027 and $4,471.21 in October 2029. Its 18% probability is lower because the assumed 32%–33.3% margins and rising multiple are not yet established through a full cycle.

8.5 — What would change the view

The first test is the Q1 FY2027 report, due on October 29, 2026. Revenue near the guided $10,300M–$10,800M range would support the base path; a miss would increase bear risk.

Through FY2027 and FY2028, the key evidence is whether supply growth outruns demand, whether NAND pricing weakens, and whether utilization revives fixed-cost pressure. NBM agreements covering 50% of FY2027 bits and 66.7% of FY2028 bits support the base case, but do not eliminate cycle risk.

The view would move bullish if BiCS10 improves cost per bit and QLC enterprise drives convert qualifications into repeat volume. It would move bearish if buybacks shrink because investment and working capital consume cash, or if earnings prove too cyclical to sustain the selected multiple.