Sandisk stock has been upgraded to Strong Buy by Seeking Alpha analyst Esxeleryn Analytics, with the bullish call centered on High Bandwidth Flash, surging AI-related NAND demand and a business model that could make the historically cyclical memory industry far more predictable. The upgrade comes just weeks after Sandisk reported fiscal Q4 revenue of $8.97 billion, up 51% sequentially, and outlined a long-term financial model targeting mid-to-high-teens revenue growth and roughly 80% non-GAAP gross margins.
The central question for Sandisk stock is no longer whether the current NAND upcycle is strong. The numbers already show extraordinary demand and pricing. The more important issue is whether technologies such as High Bandwidth Flash, long-term customer agreements and AI inference workloads can transform Sandisk from a volatile memory-cycle trade into a structurally higher-growth, higher-margin semiconductor company.
Sandisk Stock Upgrade Is Really an AI Memory Call
Esxeleryn Analytics upgraded SNDK to Strong Buy because it sees Sandisk’s High Bandwidth Flash, or HBF, as a potentially disruptive solution to one of the AI industry’s biggest emerging bottlenecks: memory capacity close to the compute layer. Seeking Alpha’s summary of the call specifically highlighted HBF’s potential cost and efficiency advantages as the main reason for the upgrade.
That matters because the AI infrastructure debate is broadening beyond GPUs. Large models require enormous amounts of memory and storage, particularly as the industry shifts from training toward inference, where models continuously process user prompts, generate tokens and maintain increasingly large context windows.
Traditional high-bandwidth memory offers tremendous speed but is expensive and capacity-constrained. Sandisk’s HBF concept uses NAND flash to provide much larger amounts of near-compute memory at lower cost, potentially allowing AI systems to retain more information close to accelerators without relying exclusively on expensive DRAM-based HBM.
Related articles
Sandisk is effectively betting that AI inference will create a new layer in the memory hierarchy — one where conventional storage is too slow, HBM is too expensive, and HBF can fill the gap.
High Bandwidth Flash Is Moving Beyond a Concept
The bullish thesis would be far weaker if HBF existed only on Sandisk’s product roadmap. Instead, an ecosystem is beginning to form around the technology.
Sandisk and SK hynix released the first Open Compute Project technical specification for HBF in August, establishing guidelines for interfaces, reliability, packaging and software. Google and Tenstorrent also joined the standardization process, giving the effort considerably more credibility than a proprietary Sandisk-only architecture would have had.
The standard is designed specifically around AI inference systems that need more capacity positioned close to compute while maintaining high bandwidth and improved power efficiency. Sandisk argues that this could help data-center operators improve token economics — essentially reducing the cost of serving AI responses at scale.
That is a strategically important distinction. Training created enormous demand for GPUs and HBM, but inference may ultimately represent a much larger and more persistent workload because every production AI application requires continuous processing. If HBF becomes an accepted part of future AI architectures, Sandisk’s addressable market could expand well beyond traditional enterprise SSDs.
Sandisk’s Latest Results Already Show an AI Boom
Investors do not need to wait for HBF commercialization to see AI’s impact on Sandisk’s financials. Fiscal fourth-quarter revenue reached $8.97 billion, up 51% sequentially, with roughly one-third of the increase coming from higher volumes and two-thirds from higher pricing. Non-GAAP diluted EPS reached $39.25.
For the full fiscal year, revenue surged 175% to $20.25 billion, while Data Center revenue increased 437%. The company said its performance reflected a shift toward higher-value customers alongside sharply stronger memory pricing.
Those figures demonstrate how dramatically Sandisk’s earnings power has changed during the current memory cycle. The challenge is that NAND has historically been notorious for boom-and-bust conditions: tight supply produces extraordinary pricing and margins, which encourages capacity expansion, eventually creating oversupply and collapsing profitability.
That historical pattern is precisely what Sandisk is trying to break.
Long-Term Customer Deals Could Reduce the Memory Cycle
One of the most important developments from Sandisk’s August Investor Day was its new business model, or NBM, agreements. These contracts include committed volumes, minimum financial guarantees and structured pricing mechanisms designed to make future revenue and cash flow more predictable.
Sandisk has signed such agreements with eight customers, covering approximately 50% of fiscal 2027 bit shipments and roughly two-thirds of fiscal 2028 bits. That gives management significantly more visibility into demand and pricing than traditional spot-market NAND economics would normally provide.
For shareholders, that could be transformational. If large cloud and AI customers increasingly commit to multiyear supply agreements, Sandisk may be able to plan production more rationally while reducing exposure to sudden price collapses.
The bullish investment case therefore extends beyond AI demand itself. It assumes Sandisk can use the current shortage environment to negotiate a structurally better commercial model before memory supply eventually catches up.
Sandisk’s Margin Targets Are Extraordinary
The company’s long-term financial model explains why analysts have become so enthusiastic. Sandisk expects revenue to grow at a mid-to-high-teens rate from fiscal 2028 through fiscal 2030, while targeting approximately 80% non-GAAP gross margins and 75% non-GAAP operating margins. It also expects adjusted free-cash-flow margins around 50%.
Those figures are extraordinary for a company historically associated with commodity-like NAND economics. Operating margins around 75% would place Sandisk among the most profitable large semiconductor companies in the world.
Management argues those margins can be supported by disciplined supply, improved customer agreements, higher-value data-center products and the increasing strategic importance of flash memory to AI systems.
Investors should nevertheless treat these targets as management’s long-range framework rather than guaranteed results. Memory markets remain highly sensitive to supply additions, customer inventories and pricing, while competitors including Kioxia, Samsung, Micron and SK hynix are all investing in next-generation technology.
The upside is enormous if Sandisk delivers. The disappointment risk is equally significant if today’s margins prove cyclical rather than structural.
Q1 Guidance Suggests the Momentum Is Still Building
Near-term guidance supports the bullish argument. Sandisk expects fiscal first-quarter 2027 revenue of $10.3 billion to $10.8 billion, significantly above the $8.97 billion generated in Q4. Management also forecasts non-GAAP diluted EPS of $44 to $46.
That means Sandisk expects another substantial sequential increase even after Q4 revenue jumped 51%. The outlook reinforces evidence that data-center demand and NAND pricing remain exceptionally strong heading into the new fiscal year.
The company is also returning capital aggressively. Sandisk’s board approved an additional $14 billion share repurchase program, bringing its remaining authorization to $15.5 billion.
At its Investor Day, management went further, saying it expects to return 100% of excess cash to shareholders after investing in the business. If free-cash-flow margins eventually approach the 50% long-term target, the resulting buyback capacity could become another major driver of per-share earnings.
Wall Street Is Becoming Aggressively Bullish
The Seeking Alpha Strong Buy call is not occurring in isolation. JPMorgan recently resumed coverage of Sandisk with an Overweight rating and a $2,250 December 2027 price target, arguing that the company is uniquely positioned to benefit from the structural expansion of NAND demand driven by AI inference.
Other bullish targets include approximately $2,000 from Wedbush, $1,900 from Mizuho, $2,500 from Bank of America and $2,800 from Evercore ISI, while Cantor Fitzgerald has published a $2,900 target.
The range illustrates just how dramatically analyst expectations have moved. Sandisk is no longer being evaluated primarily as a consumer-storage or conventional NAND company. The market increasingly views it as a potential infrastructure supplier to the AI inference economy.
That reclassification matters because investors typically assign much higher valuation multiples to companies with structurally growing data-center exposure than to cyclical commodity-memory businesses.
But Sandisk Stock Is No Longer Cheap
The biggest argument against chasing SNDK stock is valuation after its spectacular rally. Sandisk recently traded near $1,600 per share, with one valuation service putting the stock at approximately 22 times trailing earnings and 25.5 times forward earnings.
Those multiples do not look extreme relative to premier AI semiconductor companies, but they are far above what investors historically paid for deeply cyclical NAND producers. The valuation therefore already assumes that Sandisk’s improved margins and AI demand will prove more sustainable than previous memory booms.
That is the critical risk. If NAND pricing weakens materially, earnings estimates can fall extremely quickly. A stock that looks inexpensive based on peak-cycle earnings can suddenly become expensive once normalized profits are used instead.
Investors should therefore watch contract pricing, competitor capacity additions and customer inventory levels as closely as HBF announcements.
Competition and Execution Remain Major Risks
HBF itself still carries execution risk. Standardization is progressing and the industry partners are credible, but broad commercial deployment will require semiconductor designers and AI-system builders to redesign memory architectures around the technology.
Sandisk must also prove that HBF can deliver meaningful advantages in real-world workloads and that customers are willing to adopt it alongside existing HBM and SSD solutions.
Its broader NAND roadmap is advancing quickly. Sandisk and Kioxia recently unveiled new QLC flash technology with up to a 60% increase in bit density versus BiCS8, while another 9th-generation design targets AI and data-intensive applications with capital-efficient manufacturing.
Those developments strengthen the product story, but they also underline how competitive the sector remains. Sustaining today’s economics requires Sandisk to keep innovating while avoiding the excessive capacity expansion that historically destroyed NAND profitability.
Outlook: Sandisk Is Trying to Rewrite the NAND Playbook
The Strong Buy upgrade captures why Sandisk stock has become one of the most intriguing AI infrastructure plays outside GPUs and traditional HBM. Revenue is exploding, Data Center demand is surging, HBF could create an entirely new memory category, and multiyear customer agreements may make future earnings considerably less cyclical.
The financial targets are equally ambitious. Sandisk is aiming for mid-to-high-teens long-term revenue growth, roughly 80% gross margins, 75% operating margins and a 50% adjusted free-cash-flow margin while promising to return excess cash to shareholders.
Investors should now watch HBF customer adoption, Q1 execution against the $10.3 billion-$10.8 billion revenue guide, NAND pricing, the durability of NBM contracts and whether Sandisk’s exceptional margins survive the eventual normalization of memory supply.
Sandisk’s bull case is no longer simply that AI needs more storage. It is that AI could permanently change what flash memory is worth — and if HBF becomes part of the standard AI architecture, Wall Street may have to rethink what SNDK itself is worth as well.
