Long-Term Agreements Reshape Margins; JPMorgan Initiates SanDisk at Overweight

Overview

Following the lifting of coverage restrictions on SanDisk, JPMorgan initiates coverage with an Overweight rating. The report argues that long-term agreements will shift most of SanDisk's business from spot-priced commodity flash toward a long-term, high-margin revenue model featuring minimum purchase commitments. At the same time, AI inference-driven growth in data-center storage demand, BiCS technology evolution, and HBF positioning should support stronger long-term earnings power.

Core views

The core bullish thesis includes: first, long-term agreements raise the margin floor and materially improve revenue and price visibility; second, AI inference and persistent KV cache drive expansion in data-center NAND demand, while industry supply constraints support pricing; third, BiCS10 and subsequent technology iterations increase bit output per wafer, while HBF offers potential incremental growth; fourth, high free cash flow and a commitment to return all residual cash to shareholders should drive sustained compounding EPS growth.

Analysis framework

The report assesses earnings stability using long-term agreement contract value, term, floor-price gross margin, and bit-coverage ratio; it evaluates growth and pricing conditions through NAND market size, data-center demand, industry capacity utilization, and technology roadmaps; valuation uses a forward P/E approach, applying the midpoint of the historical storage-industry forward P/E range to the 2027 EPS forecast.

Methodology notes

• Business Model AnalysisLong-Term Agreements (NBM/LTA)

  Long-term purchase and pricing agreements

  By agreeing monthly or quarterly volumes, product mix, and pricing, while setting price floors and ceilings and financial guarantees, these agreements improve the predictability of revenue and gross margin.

• Relative ValuationForward P/E Method

  Target price = forward EPS × target P/E

  A $2,250 target price is derived from 2027 EPS of $250 and a 9x forward P/E; 9x falls within the historical 7x to 10x range.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• SNDK.US

  Directly Covered Security

  Strengths

  Long-term agreements provide multiyear visibility into volumes, product mix, and pricing; AI inference drives data-center NAND demand; BiCS10 progress and higher bit output per wafer support capital efficiency; strong free cash flow and buyback commitments benefit EPS growth.

  Weaknesses

  The company remains in a highly competitive memory market, with part of bit demand dependent on traditional end markets such as smartphones, PCs, and consumer electronics.

  Comparison

  The target valuation applies a 9x 2027 forward P/E, representing the midpoint of the historical 7x to 10x storage-industry range.

  Risks

  A slowdown in AI infrastructure spending, industry capacity expansion and price competition, and macro-driven declines in traditional end-market demand could all pressure pricing and margins.

Key data

• Investment RatingOverweightChanged from Unrated to Overweight.
• Target Price$2,250Target date is December 2027.
• Reference Share Price$1,528.11Closing price on August 13, 2026.
• Potential UpsideApproximately 47%Relative to the reference share price cited in the report.
• Number and Value of Long-Term Agreements8; approximately $94BAggregate contract value calculated at floor prices.
• Average Term of Long-Term AgreementsMore than 4 yearsWeighted-average contract term.
• Floor-Price Gross MarginApproximately 80%Gross margin the report expects long-term agreements to sustain at floor prices.
• Financial Guarantees$16.5BUsed to ensure customers fulfill their purchase obligations.
• NAND Market Size ForecastApproximately $70B in 2025; more than $300B in 2026; approximately $500B in 2027Includes company and JPMorgan expectations.
• 2027 EPS Forecast$250Used in the target-price calculation.

Impact & implications

If long-term agreements execute as expected, SanDisk's margins and cash flow should prove more resilient than under the traditional NAND spot cycle, and the market may gradually assign greater confidence to its earnings. AI inference demand and data-center storage penetration are key drivers of earnings upgrades, while share-count reduction from buybacks could further amplify EPS performance.

Risks

• A material slowdown in AI infrastructure capital spending, project cancellations, or stricter return-on-investment scrutiny could result in reduced enterprise SSD orders and earlier pricing pressure.
• NAND suppliers may expand capacity aggressively due to high margins or cut prices to gain share, potentially causing ASP declines and margin erosion.
• Weak smartphone, PC, and consumer-electronics demand due to a global recession or soft consumption could push excess supply into the market and weigh on blended ASPs.
• Customer execution, floating-price mechanisms, or coverage scale under long-term agreements could fall below expectations, weakening margin stability.
• Customer adoption and commercialization timing for new technologies such as HBF and 3D Matrix Memory remain uncertain.

What to watch

• Signing and expansion of new long-term agreements, and their share of shipped bits.
• Long-term agreement floor prices, realization of floating pricing, and whether gross margins approach the approximately 80% expectation.
• Data-center and AI inference-related enterprise SSD orders and persistent KV cache demand.
• NAND industry wafer capacity, utilization rates, supplier expansion plans, and ASP trends.
• BiCS10 mass-production progress, bit-output improvements in subsequent BiCS generations, and HBF sampling and customer-validation progress in 2027.
• Free cash flow, share-repurchase scale, and changes in shares outstanding.