JPMorgan upgrades Sandisk to Overweight on AI storage demand
JPMorgan upgraded Sandisk to Overweight with a $2,250 target, driven by AI NAND demand and $94 billion in long-term contracts. The firm sees margins stabilizing around 80% as data center share rises to 50% by 2026. Sandisk targets mid-to-high teens revenue growth through FY30.
*this image is generated using AI for illustrative purposes only.
JPMorgan upgraded Sandisk Corp. (NASDAQ: SNDK) to an Overweight rating from Not Rated on Friday, citing strong AI-driven storage demand and a new business model anchored by long-term customer agreements. Analyst Harlan Sur set a December 2027 price forecast of $2,250, implying approximately 47% upside from the company's Aug. 13 closing price of $1,528.11.
The bullish thesis centers on the expansion of the NAND flash market, which JPMorgan estimates will grow from about $70 billion in 2025 to more than $300 billion in 2026. Sandisk projects the market could approach $500 billion by 2027. Data centers are expected to account for roughly half of the market in 2026, up from approximately 30% in 2025, driven by storage-intensive AI inference workloads such as persistent key-value cache.
Long-Term Agreements Stabilize Revenue
Sandisk has signed eight agreements representing about $94 billion in total contract value at floor pricing. These contracts have a weighted-average duration of more than four years and include $16.5 billion in financial guarantees. JPMorgan estimates gross margins could remain around 80% even at floor pricing.
The agreements are expected to cover more than half of Sandisk's bits in fiscal 2027 and roughly two-thirds in fiscal 2028. This structure aims to shift much of the business away from volatile spot-priced flash toward long-dated, high-margin revenue, potentially breaking the historic boom-bust cycle of the memory sector.
Financial Outlook and Technology Roadmap
Sandisk's long-term financial model targets mid- to high-teens percentage revenue growth in fiscal 2028 through fiscal 2030. The company also aims for an adjusted gross margin of about 80%, an operating margin of roughly 75%, and an adjusted free cash flow margin near 50%. JPMorgan expects this revenue growth to translate into an EPS compound annual growth rate above 25%.
Sur forecasts calendar 2027 EPS of $250. His $2,250 price forecast applies a nine-times multiple to that estimate, within the historical seven- to 10-times forward earnings range for memory stocks. Sandisk plans to return 100% of excess cash flow to shareholders, which JPMorgan expects will reduce share count and further support per-share earnings growth.
What the Numbers Show
The shift toward long-term agreements represents a significant structural change in Sandisk's revenue composition. With contracts covering more than half of its bit output in fiscal 2027 and two-thirds in fiscal 2028, the company is effectively locking in high-margin economics for a majority of its production. This contrasts sharply with the traditional spot-market exposure that has historically driven volatility in memory chip pricing. The $16.5 billion in financial guarantees embedded in the $94 billion total contract value further underscores the depth of customer commitment, providing a stable base against which AI-driven volume growth can be leveraged.
On the technology front, Sandisk's BiCS10 chip is sampling ahead of schedule and packs 65% more bits per wafer than BiCS8. The company expects a 27% compound annual growth rate in per-wafer bit productivity across several BiCS generations. Additionally, Sandisk is developing High Bandwidth Flash (HBF) for AI inference workloads, with the first HBF memory die taped out and initial product samples targeted for 2027.
Sandisk shares were up 7.41% at $1,641.31 at the time of publication on Friday.
How might the shift from spot-market pricing to long-term contracts affect Sandisk's agility in responding to sudden downturns in AI infrastructure spending?
What are the potential risks if the projected $300 billion NAND flash market expansion slows due to saturation in AI inference workloads by 2026?
Could competitors like Samsung or Micron replicate Sandisk's long-term agreement model, and how would that impact industry-wide pricing power and margins?
20
Create Custom Screeners 
Access 200+ Pre-built Filters 
Use 50+ Readymade Screener 
