SanDisk Earnings Today: The NAND Upcycle Is Supply Discipline, Not AI Demand

The consensus narrative

SanDisk is one of the greatest AI infrastructure winners. Data center demand for NAND flash memory has created an unprecedented shortage, and the company-spun off from Western Digital in February 2025-has become the pure-play beneficiary. Its stock has surged approximately 4,000% over the past year, and analysts continue to raise targets. Lynx Equity pushed its price target to $750 from $300 in January and has maintained its bullish stance through June. That is the story investors are carrying into fourth-quarter fiscal 2026 earnings today.

The structural reality

The NAND flash recovery is not primarily driven by demand growth. It is being driven by manufacturers divesting NAND capacity to pursue higher-margin HBM and advanced DRAM production, creating a supply constraint that is structurally different from previous memory cycles. The demand story is real but secondary. The supply story is what determines whether pricing power persists.

The supply constraint

Major NAND manufacturers-Samsung, SK Hynix, and Micron-are structurally reducing NAND production capacity as they reallocate fab space and capital toward HBM, the high-bandwidth memory that powers AI training and inference. Industry research from Isaiah Research projects that global NAND capacity in 2027 will be approximately 40% lower than the 2022 peak. That is not a short-term adjustment. It is a structural reallocation of manufacturing capacity toward where the highest margins are.

The result is that NAND pricing has entered an upcycle that is fundamentally different from the demand-led recoveries of 2016 or 2020. This is a supply-constrained environment where limited incremental capacity meets growing demand from both AI storage and enterprise solid-state drives.

SanDisk's position within the constraint

SanDisk is one of five major NAND suppliers globally. After its separation from Western Digital, it became a focused NAND flash operator serving consumer drives, enterprise storage, and increasingly AI workloads. The company is also advancing its process technology-Kioxia and SanDisk announced the start of 10th-generation 3D NAND flash production in July, maintaining competitiveness in node advancement.

The financial results show the effect of the supply-constrained environment. For the quarter ending in April 2026, SanDisk reported diluted EPS of $23.41 against consensus estimates of approximately $14.17-a beat of roughly 65%. Trailing twelve-month margins are striking: gross margin at 56%, operating margin at 41%, and free cash flow margin at 34%. These are not the margins of a commodity semiconductor operation in a balanced supply-demand environment. They are the margins of a supplier with pricing power derived from limited capacity.

Table 1: SanDisk Financial Metrics (TTM as of latest quarter)

Metric	Value
Revenue Growth YoY	82.8%
Revenue Growth QoQ	96.7%
Gross Margin	56.0%
Operating Margin	40.7%
Free Cash Flow Margin	33.8%
ROIC	38.0%
Capital Expenditure TTM	$179M
Free Cash Flow TTM	$4.46B

Source: Public market data service, 2026-08-05.

The capex discipline question

The number that matters most in this table is capital expenditure: $179 million over the trailing twelve months. Against $4.46 billion in free cash flow, that is an extraordinarily low reinvestment rate. SanDisk is not building new capacity to meet the current demand surge. It is extracting cash from a supply-constrained market without the capital outlay that would eventually restore the competitive balance.

That is the mechanism that separates this cycle from the traditional demand-led recoveries. In previous NAND upcycles, strong pricing would trigger heavy capex, new capacity would come online, and margins would compress within two to three quarters. This time, the structural diversion of fab capacity toward HBM means that even aggressive NAND-capex spending would be difficult to execute. The constraint is not just capex discipline-it is physical fab space and equipment that has been committed to HBM.

The implication for SanDisk's cash generation is that it can sustain elevated margins as long as the broader supply reallocation toward HBM continues. If HBM demand decelerates and manufacturers begin converting capacity back to NAND, the pricing environment would deteriorate.

The valuation problem

SanDisk currently trades at approximately $1,399 per share with a market capitalization of $207 billion. The trailing PE ratio sits at 46 times, and EV/EBITDA is approximately 37 times. By comparison, Micron-operating in the same memory sector with comparable cyclical exposure-trades at a PE of 20 and EV/EBITDA of 14.5. SanDisk's valuation commands a premium of more than 2x on earnings multiples for what is fundamentally the same type of supply-constrained cyclicality.

Table 2: Valuation Comparison - SanDisk vs. Micron

Metric	SanDisk (SNDK)	Micron (MU)
Market Cap	$207B	$1.01T
PE TTM	45.9x	20.0x
EV/EBITDA TTM	36.7x	14.5x
PB	15.0x	10.0x
PS TTM	15.7x	11.2x

Source: Public market data service, 2026-08-05.

A 2x PE premium for NAND-only versus diversified memory is defensible only if SanDisk's margin trajectory sustains materially above Micron's for an extended period. The current 41% operating margin would need to remain elevated while Micron's cyclical average pulls down the comparison. If the NAND upcycle normalizes within 12 to 18 months-and historical memory cycles have compressed within that timeframe-the 46x PE would require earnings growth that the historical NAND cycle cannot support.

The bifurcation

The recent weekly performance of memory stocks illustrates the emerging split. Over the past week ending July 31, Samsung rose 8.3% on record earnings and long-term HBM supply deals. SK Hynix was up modestly. SanDisk fell 10.9%, Micron, which also has HBM exposure, dropped 5%, and Kioxia declined 15%. The divergence is clear: companies with HBM exposure are being rewarded, while pure-play NAND operators face pressure despite the same favorable pricing environment.

This bifurcation reflects where investors believe the constraint will tighten further. HBM has its own supply bottleneck-in advanced packaging and high-bandwidth stack technology-and the market is pricing that structural scarcity at higher multiples. NAND, even in a supply-constrained upcycle, is viewed as having a shorter path to normalization once the HBM buildout reaches equilibrium.

What to watch in today's earnings

SanDisk expects Q4 revenue between $7.75 billion and $8.25 billion, driven by both higher bit shipments and improved pricing. The management commentary on three items will determine whether the current thesis holds:

1. Capex guidance for FY2027. If capital expenditure remains near current levels ($179M TTM), the supply-constraint thesis is reinforced. If guidance jumps toward $1 billion or more, the market should interpret that as the beginning of capacity normalization.

2. Bit shipment versus ASP split. If the revenue increase between Q3 and Q4 is driven predominantly by ASP (average selling price) rather than unit volume growth, the supply-constraint narrative remains intact. If bit shipments are surging independently, demand is playing a larger role than the supply story suggests.

3. Data center storage commentary. The AI narrative requires sustained demand growth from hyperscalers for enterprise SSDs and eSSDs. Any softening in data center guidance would undermine the demand-side anchor of the thesis, even if supply discipline continues.

   

Investor Takeaway

The SanDisk upcycle is being driven by structural supply constraints, not by a demand surge that the market has misattributed as the primary catalyst. NAND capacity is being permanently reallocated toward HBM production, and SanDisk is extracting cash at extraordinary margins while spending almost nothing on new capacity. That is a durable story only as long as the broader memory manufacturers maintain their HBM-first capex allocation.

The current valuation-46 times trailing earnings, versus 20 times for Micron-prices in the assumption that this margin environment is structural rather than cyclical. Historical NAND cycles have compressed within 12 to 18 months once pricing peaked. The constraint that makes this cycle different is real, but it is contingent on HBM demand sustaining the capacity diversion.

The key issue is not whether SanDisk will deliver strong earnings today. The more important question is whether the company's capex guidance confirms continued supply restraint or signals the beginning of capacity reinvestment. If the former, the thesis holds. If the latter, the 46x PE multiple is pricing in a margin trajectory that the NAND cycle has never sustained.