Sandisk expects mid-to-high-teens revenue growth from FY2028 through FY2030, broadly aligned with bit growth.
Management targets approximately 80% non-GAAP gross margin, 75% operating margin and 50% adjusted free cash flow margin.
Eight multiyear New Business Model agreements already cover around 50% of expected FY2027 bits and roughly two-thirds of FY2028 bits.
Sandisk expects the enterprise data center flash market to reach 1.2 zettabytes by 2030 as AI inference increases storage intensity.
HBF, QLC NAND and a more capital-efficient manufacturing roadmap are central to Sandisk’s longer-term AI strategy.
Sandisk’s 2026 Investor Day was less about the next quarter and more about what the company believes its NAND business can become by the end of the decade. At its August 13 event, Sandisk introduced a financial framework for FY2028 through FY2030 built around sustained AI infrastructure demand, higher storage intensity, multiyear customer commitments and tighter capacity planning. Management expects mid-to-high-teens revenue growth while sustaining roughly 80% non-GAAP gross margins and generating adjusted free cash flow margins of approximately 50%.
For a NAND producer, those numbers are unusually ambitious. The more important question, however, is not whether Sandisk can produce another strong year while memory supply remains tight. Investor Day was an attempt to explain why management believes the business can remain highly profitable even after the current cycle matures.
Sandisk’s 2030 Financial Model Is About Durability, Not Just Growth
In its
2026 Investor Day update, Sandisk said revenue should grow at a mid-to-high-teens rate from FY2028 through FY2030, roughly matching expected growth in the amount of storage capacity it ships. Over the same period, the company expects non-GAAP gross margins to remain around
80%, non-GAAP operating margins around
75%, and adjusted free cash flow margin around
50% after taxes, capital expenditure and working-capital requirements.
The margin assumptions are arguably more important than the revenue target. NAND has historically produced very strong profits when supply is tight, followed by sharp margin compression when capacity catches up with demand. Sandisk is effectively arguing that its future economics will depend less on extracting maximum pricing from a short-term shortage and more on matching capacity with committed customer demand. That is a fundamentally different claim from simply forecasting another memory upcycle.
This is also why the FY2028–FY2030 time frame matters. Sandisk is not presenting an 80% gross-margin target only for the current period of exceptional NAND pricing; it is explicitly incorporating that level into a multiyear model. Whether the company can deliver it will depend heavily on the next part of the strategy: its New Business Model agreements.
NBM Agreements Are the Foundation of the New Sandisk Model
Sandisk’s New Business Model, or NBM, agreements are multiyear arrangements built around committed volumes, structured pricing and minimum financial guarantees. According to Sandisk, the company has now signed NBMs with eight customers, covering approximately 50% of expected FY2027 bits and roughly two-thirds of FY2028 bits. Reuters reported that the customer group includes three U.S. hyperscalers.
The importance of these agreements is operational as much as financial. NAND manufacturers must make capacity decisions well before demand is fully visible. In a traditional cycle, strong pricing can encourage suppliers to invest aggressively, only for demand to soften after new capacity arrives. Multiyear commitments give Sandisk a clearer view of future volumes before it decides how much capacity to add.
That does not make NAND non-cyclical. Pricing, customer demand and technology transitions can still change. But it potentially changes the amplitude of the cycle by making supply decisions less dependent on short-term spot-market signals. This is why Sandisk describes NBM as increasingly becoming its predominant way of doing business and why the company connects these contracts directly to its long-term margin and free-cash-flow targets.
AI Inference Could Make Data Centers Far More Storage-Intensive
The demand side of the Investor Day thesis centers on AI inference. Sandisk expects the total addressable market for enterprise data center flash to reach 1.2 zettabytes by 2030, arguing that the transition from AI training toward much broader inference workloads is changing how data centers use memory and storage.
Training initially concentrated much of the AI investment cycle around GPUs, HBM and networking. Inference creates a different challenge because AI systems must repeatedly access model weights, KV cache, embeddings, user context and increasingly large volumes of multimodal data. As inference scales across cloud services, enterprise applications and AI agents, keeping every piece of active information in scarce and expensive HBM is not economically realistic.
That creates room for a deeper memory hierarchy in which NAND flash provides far more capacity at lower cost. The important distinction is that Sandisk is not arguing NAND will replace HBM. Instead, it sees flash becoming a larger complementary layer as AI systems require both extreme bandwidth and vastly greater memory capacity.
This is one reason Sandisk’s long-term AI story is broader than the current enterprise SSD boom. The company is developing products across conventional high-capacity flash, QLC NAND and new memory architectures designed specifically around inference.
HBF Is Sandisk’s Bet on a New AI Memory Layer
High Bandwidth Flash, or HBF, is the most strategically interesting part of that roadmap. HBF is designed to combine the large capacity and economics of NAND with significantly higher bandwidth, creating a potential memory layer between conventional SSD storage and premium HBM. According to
Reuters’ coverage of the Investor Day, Sandisk has already taped out its first HBF memory die and is working toward initial customer samples for AI inference devices in 2027.
The technology is still early, so HBF should not be treated as a major contributor to current Sandisk earnings. Its importance is strategic. If inference workloads continue to increase memory requirements faster than premium HBM capacity can economically scale, there may be room for an additional high-capacity, high-bandwidth memory tier.
Sandisk is also trying to turn HBF into an industry architecture rather than a proprietary niche. Its work with SK hynix and participation from major AI infrastructure customers indicate that the company is building an ecosystem around the technology, although commercial adoption will ultimately depend on performance, system integration and customer economics.
QLC and Manufacturing Efficiency Matter Just as Much as HBF
Investor Day also highlighted a less visible but potentially more immediate part of Sandisk’s roadmap: higher-density QLC NAND and more capital-efficient technology transitions. Sandisk introduced a two-dimensional scaling strategy based on CMOS directly Bonded to Array, or CBA, which allows the company to combine different generations of array and CMOS technology rather than replacing the entire architecture at once. Its new BiCS9 QLC product is an early example, while the upcoming BiCS10 QLC node is designed to deliver approximately 60% higher bit density than BiCS8.
Higher bit density matters because the AI storage opportunity is ultimately a capacity problem. Hyperscalers need more terabytes and petabytes per rack while controlling power consumption and total cost. QLC stores more bits per cell and therefore offers attractive economics for high-capacity workloads where cost per bit matters more than achieving the absolute lowest latency.
Combined with CBA, Sandisk is effectively trying to improve two variables at once: deliver more storage capacity per wafer while avoiding unnecessarily expensive manufacturing transitions. That capital efficiency is important if the company wants to grow bits at a mid-to-high-teens rate without recreating the overinvestment that has historically destabilized memory markets.
Sandisk Is Pairing AI Growth With a Stronger Capital-Return Story
The financial model does not end with margins. Sandisk said it intends to return 100% of excess cash to shareholders after investing in the business, connecting its AI growth strategy directly with capital returns. Management expects roughly 50% adjusted free cash flow margins during FY2028–FY2030, although those targets remain forward-looking and depend on demand, pricing, execution and the broader memory cycle.
This makes the Investor Day thesis more comprehensive than a conventional semiconductor growth story. Sandisk is effectively proposing a model in which AI drives sustained bit demand, NBM contracts improve visibility, capital-efficient manufacturing limits unnecessary spending, high margins generate cash, and excess capital flows back to shareholders.
The key word is sustainability. Investors already know that NAND companies can generate enormous profits during favorable pricing environments. Sandisk is now trying to show that a combination of AI storage growth and a different commercial model can make those profits structurally more durable.
Access Sandisk and SNDK-Related Markets on MEXC
Eligible users following Sandisk can access several different forms of SNDK-related market exposure on MEXC. For leveraged long or short positioning, the
SNDKSTOCK_USDT perpetual futures contract tracks Sandisk-related price movements through a USDT-margined perpetual product. Traders should account for leverage, funding and liquidation risk.
Users looking for traditional equity exposure can also explore
Sandisk through MEXC RealStocks, while
SNDKON/USDT provides tokenized economic exposure to Sandisk through Ondo. SNDKON is a tokenized product and does not make the holder a direct shareholder of Sandisk. MEXC also provides a broader
Sandisk stock and AI storage guide for readers researching the company’s NAND, SSD and data-center exposure.
The Bigger Message From Sandisk Investor Day
Sandisk’s 2026 Investor Day was ultimately a bet that the economics of NAND are changing at the same time as its end market is changing. AI inference could create a much larger need for storage capacity, while NBM agreements give Sandisk a mechanism to align that demand with production more carefully.
The company still operates in a competitive and historically cyclical semiconductor market, and ambitious FY2028–FY2030 targets should be treated as management expectations rather than guaranteed outcomes. But the strategy is now much clearer: Sandisk wants to combine AI-driven bit growth, a more predictable commercial model, capital-efficient technology scaling and high cash returns.
That is a substantially different SNDK thesis from simply betting on the next move in NAND prices.