SanDisk Beat Earnings and Fell. The Question Was Never the Quarter.

The consensus narrative is that SanDisk's Q4 fiscal 2026 results failed to live up to expectations. That narrative is wrong on the facts and misleading on the mechanism. SanDisk reported non-GAAP EPS of $39.25 for Q4, beating the consensus estimate of $33.38 by 17.6%. Revenue guidance was $7.75 billion to $8.25 billion, representing a roughly 30% to 39% sequential increase from Q3's $5.95 billion. The stock fell 5.4% on the day to $1,351, on volume of nearly 15 million shares and $20.7 billion in turnover. The market did not sell because the quarter was weak. The market sold because the quarter was strong enough to force a reckoning with what the multiple implies.

The Valuation Does Not Match the Business Model

SanDisk is trading at 44.4 times trailing earnings. Micron - a direct memory competitor generating 65.6% operating margins, 58.6% return on invested capital, and 167% year-over-year revenue growth - trades at 20 times earnings. Western Digital, whose business is concentrated in hard disk drives rather than NAND flash, trades at 28 times earnings. The market is assigning SanDisk a 122% multiple premium over Micron for a business whose pricing power derives from the exact same supply-demand dynamics. The premium makes sense only if the upcycle is permanent. For a cyclical commodity, that assumption is the entire risk.

The earnings trajectory tells the acceleration story. After spinning off from Western Digital on February 24, 2025, at approximately $38 per share with gross margins collapsed to 7%, SanDisk has posted the following quarterly progression:

• Q1 FY2026: $1.22 EPS, $2.31 billion revenue
• Q2 FY2026: $6.20 EPS, $3.03 billion revenue
• Q3 FY2026: $23.41 EPS, $5.95 billion revenue (reported), gross margin expanded to 78.4%
• Q4 FY2026: $39.25 EPS, approximately $8 billion revenue

That is not a growth trajectory. It is a recovery-from-the-floor trajectory. The comparison base for Q1 and Q2 was the deepest NAND downcycle in a decade, when gross margins collapsed and the company posted operating losses. The Q3 and Q4 numbers represent the peak of the current upcycle. The market is valuing the peak as if it were the new baseline.

Supply Discipline Is the Driver, Not Demand

The semiconductor recovery is not being driven by a surge in NAND unit demand. It is being driven by constrained supply. Post-2023, the major NAND manufacturers - Samsung, Kioxia, Micron, and SanDisk itself - collectively restrained output to avoid repricing their balance sheets. Average selling prices recovered because bit supply growth was kept in check, not because the market needed more storage than it had been making. AI-driven data center demand for enterprise solid-state drives has accelerated this dynamic, with SanDisk's data center segment now representing nearly 25% of revenue, up roughly 7x year-over-year. But AI demand is the amplifier, not the origin. The origin is supply discipline.

And supply discipline is inherently self-defeating. The higher NAND prices climb, the more incentive every manufacturer has to add capacity. SanDisk's fabless model - with trailing twelve-month capex of just $179 million, compared to Micron's $25.26 billion - means SanDisk does not directly control wafer-level NAND supply. SanDisk purchases NAND capacity from contract foundries and joint-venture partners. The constraint that has supported SanDisk's pricing belongs to other companies' capital allocation decisions. That makes SanDisk's pricing power derivative, not autonomous.

The Long-Term Agreements Are a Floor, Not a Ceiling

The defense most frequently offered for SanDisk's premium valuation is its portfolio of five long-term customer agreements, covering approximately 60% of volumes at floor prices around $0.29 per gigabyte. Bernstein analysts have modeled that even in a worst-case scenario of NAND prices collapsing to $0.11 per gigabyte, SanDisk's fiscal 2030 EPS would land at $214. That analysis is internally consistent. It is also not the question.

Those agreements prevent the earnings from going to zero. They do not prevent the multiple from compressing from 44x to 20x while the earnings are still positive and still large. A cyclical stock priced at peak earnings with a peak multiple faces downside from two directions: earnings normalization and multiple compression. The long-term agreements address the first. They do nothing for the second. The market has already applied the first-quarter-of-next-cycle earnings to the current price. When the cycle turns, both variables move against the holder simultaneously.

Where the Market Is Misattributing the Risk

The competitor headline frames this as a story about SanDisk's forecast not living up to high expectations. That framing puts the causal arrow on the wrong end of the mechanism. The forecast beat expectations. The stock fell because the expectations embedded in the price were structurally unsustainable, and today's results made that fact harder to ignore. When a stock has risen 3,900% from spinoff (as of late June) and 468.9% year-to-date, no quarterly result - even one that beats - can justify the implied permanence of peak-cycle conditions. The sell-off is not about the quarter. It is about the math.

SanDisk's financial position is not weak. The company generated $4.46 billion in free cash flow over the trailing twelve months, carries $3.7 billion in cash against $3.3 billion in total debt for a net-debt profile that is effectively neutral, and has authorized a $6 billion share repurchase program. These are real strengths. But they describe a company executing well in the current cycle, not a company whose business model has escaped its cyclicality.

Investor Takeaway

The key issue is not whether SanDisk's data center demand remains healthy or whether its BiCS8 and BiCS10 chip generations maintain their performance advantage. The more important question is whether NAND suppliers maintain the supply restraint that is currently supporting pricing power. Historically, they do not. When ASPs stay elevated for two or three quarters, capacity additions follow. That is not a prediction about demand weakening. It is a prediction about the incentive structure of a commodity market that has been suppressed and is now re-pricing. The stock at 44 times earnings assumes supply discipline lasts indefinitely. The history of NAND cycles says it does not. The investor who bought the 3,900% run-up is now holding a peak-cycle multiple on a business whose peak-cycle margins are the very thing that invites the next supply expansion.