SanDisk’s $93.9B Floor Is Now the Only Question

SanDisk (SNDK) posted one of the most impressive quarters in the history of the semiconductor industry on August 5, then watched its stock fall anyway. Fiscal fourth quarter revenue surged sequentially, and profitability metrics reached extraordinary levels for a company of this size. These are numbers that almost no technology company in history has ever produced at this scale. And the stock still fell.

The Big Question

Here is what institutional investors are actually debating this morning: not whether SanDisk had a great quarter, but whether the company’s disclosed New Business Model contract base, described by management as including a substantial minimum-revenue “floor”, is a structural moat that redefines how NAND companies should be valued, or simply a record of peak-cycle pricing locked in at exactly the wrong moment.

The answer to that question determines whether the guidance miss is a buying opportunity or a warning signal.

Why Wall Street Cares

SanDisk guided fiscal Q1 revenue below where many analysts had been modeling, and the stock fell in early trading on August 6. On its face, this looks like a straightforward miss-on-guidance reaction. But the magnitude of the selloff requires a more precise explanation, because the headline numbers from Q4 were genuinely strong.

SanDisk’s adjusted earnings per share topped expectations, and the company’s fiscal fourth quarter revenue edged above consensus. Reported results beat across the board. What the market is selling is the forward picture, and specifically one version of it.

The professional investor framing is simpler: if the guidance midpoint for Q1 is roughly in the middle of the range provided, the sequential step-up from Q4 is still enormous. But memory stocks do not trade on sequential moves. They trade on how far the current environment is from the inevitable mean reversion.

The Bull Case

The strongest bull argument is not the margin or the revenue. It is the floor.

The most important long-term disclosure was SanDisk’s expanded New Business Model contracts, which are multi-year customer agreements covering supply, volume and pricing. Pricing includes fixed and variable components, with floors and ceilings designed to protect acceptable margins even if market prices decline.

SanDisk has described its minimum contracted NBM revenue at floor pricing as a very large number relative to its current revenue base, and has also described financial guarantees supported by customer collateral. This is the crux of the bull case. If NAND pricing weakens materially, SanDisk does not fall off a cliff the way it would in a traditional commodity downcycle. The floor pricing holds. The customers who signed these agreements also deposited real money as collateral. CEO David Goeckeler characterized the change in position as dramatic in an interview reported by Reuters.

Add to that a large share repurchase authorization that management has highlighted alongside the quarter’s cash generation. A company generating that kind of cash with minimal leverage and a sizable base of contracted minimums is not obviously cheap at roughly 7.5 times an annualized earnings run rate.

The AI demand story also has room to run. The workloads driving datacenter demand, including inference, retrieval-augmented generation, KV cache and autonomous agentic systems, require substantial high-performance low-latency NAND flash at scale. SanDisk has argued that data centers become the largest NAND end market by 2026, with demand forecasts moving higher.

The Bear Case

The bear argument is more subtle and more dangerous than simple valuation.

The contract floor was built during a period of near-unprecedented NAND pricing. The concern is that the NBM agreements are capturing a significant portion of the cycle’s peak. If the contracts locked in elevated floor pricing, and that pricing is structurally above long-run equilibrium, then the floor itself becomes a liability in the eyes of customers who need to renegotiate.

The bear case is more time-tested: NAND is a commodity, and commodity margins eventually return to the mean. Even if the current quarter’s results are stellar, the stock could still face downward pressure if management provides conservative guidance for fiscal 2027, or fails to clearly articulate the actual effectiveness of long-term supply agreements in smoothing out industry cyclicality.

Goldman Sachs has outlined key downside hazards, including the risk that a structural change in NAND pricing fails to materialize, Chinese competitor YMTC iterating on its roadmap, and SanDisk failing to gain enterprise SSD traction. YMTC is a risk that has been easy to dismiss, but it gets harder to dismiss as the company’s technology roadmap advances and geopolitical dynamics shift.

There is also a customer concentration concern embedded in the long-term agreement model. Deeper reliance on a small group of large hyperscalers supports the less-cyclical story, but could limit SanDisk’s pricing flexibility if those customers push back once more NAND supply comes online.

The Evidence

The trajectory of this business in fiscal 2026 was extraordinary. SanDisk reported that fiscal year 2026 revenue climbed sharply year over year, driven by mix shift toward higher-value customers, a surge in datacenter revenue, and higher pricing.

SanDisk has said that a handful of signed long-term deals will cover a meaningful portion of NAND shipments for fiscal 2027, and that the proportion could rise further. It has also described minimum revenue from contracts signed in fiscal Q3 as extremely large and backed by customer collateral and penalties. These are not soft take-or-pay arrangements. They have teeth.

The guidance miss, viewed in that context, is a function of a market that had extrapolated Q3’s blowout results indefinitely. In fiscal Q3, the company reported revenue of $5.95 billion, up 97% sequentially. The Street made the mistake of expecting every successive quarter to produce the same upside coefficient.

The company’s forecast for adjusted earnings per share for the current quarter was broadly in line with many estimates. The revenue shortfall is the problem, and it reflects both a more conservative pricing ramp assumption and deliberate mix management.

The Mavens’ View

The honest debate in institutional portfolios this week is not whether SanDisk is a good business. Almost nobody argues otherwise. The debate is whether the current valuation, after an extraordinary run, has fully priced the NBM model’s protection against cyclicality.

At the August 5 close, the stock traded at a high multiple of fiscal 2026 non-GAAP earnings, while annualizing the next-quarter guidance implies a much lower headline multiple. The real decision for investors is not whether the implied multiple is cheap. It is whether that run-rate earnings level is remotely representative of sustainable annual earnings.

Portfolio managers who have been long this stock for most of the year are now sitting on massive gains and facing a genuinely uncertain forward-earnings outlook. The NBM contracts provide a floor, but they do not eliminate the cyclicality question. They restructure it. The question becomes: where does floor pricing settle when the next contract round is negotiated, and does the AI demand wave continue to grow fast enough to absorb incremental NAND supply from Samsung, SK Hynix, and eventually YMTC?

With long-term demand locked in, SanDisk has argued it can invest with greater confidence in capacity expansion and next-generation technology, including a High-Bandwidth Flash prototype line expected to launch this year and a more complete solution targeted for the first half of 2027. HBF, if it reaches commercial deployment on schedule, would be a genuine product differentiation event rather than another commodity pricing story. That timeline is the one most bears are underweighting.

What Investors Are Missing

The selloff conversation is almost entirely about the revenue gap between the Q1 guidance midpoint and the consensus estimate. That gap is real but may be the least interesting data point in the entire earnings package.

What the market is not pricing is the asymmetric structure of the NBM agreements. Pricing includes fixed and variable components, with floors and ceilings designed to protect acceptable margins even if market prices decline. In a downcycle, the floors hold. In an upcycle, the variable component allows participation above the floor. The structure is not a cap on earnings. It is a ratchet.

The second overlooked item is the Investor Day on August 13, 2026. Every analyst on the Street will be listening for specifics on how many additional NBM agreements are in advanced discussion, what the fiscal 2027 contracted revenue mix looks like at the end of this year, and whether the HBF prototype timeline is still on track. That event may be a far more important datapoint for the stock than the guidance miss is today.

Also absent from most post-earnings commentary: the buyback. Management has emphasized a sizable repurchase authorization alongside the quarter’s cash generation. A company with minimal long-term debt and multibillion-dollar quarterly free cash flow buying back a meaningful percentage of its equity at a low multiple of sustainable forward earnings is a very different investment proposition than the headline reaction suggests.

Stocks to Watch

SanDisk (SNDK). The central name. The guidance miss is real and the stock deserved to correct after an extraordinary run over the past year. But the NBM contract floor, the buyback, the HBF timeline, and the August 13 Investor Day all argue against treating the current selloff as a structural inflection point. The honest position is uncertainty about cycle duration, not certainty of decline.

Micron Technology (MU). Micron is the cleanest comparable, with overlapping NAND exposure and similar AI-driven margin dynamics. Micron’s own long-term agreement disclosures in recent quarters were part of what sparked more bullish views on the storage complex. If SanDisk’s forward guidance disappointment signals a broader moderation in hyperscaler NAND procurement pace, Micron will face similar pressure.

Western Digital (WDC). Former parent Western Digital also weakened after beating estimates, reinforcing the view that storage expectations had outrun near-term forecasts across the sector. Western Digital has its own enterprise SSD exposure and its own long-term agreement ambitions. The selloff in both names simultaneously points to a sector-wide reset, not a company-specific story.

Kioxia. SanDisk’s joint venture partner in the Yokkaichi and Kitakami flash fabrication plants is the least visible but most directly tied name in this debate. SanDisk has disclosed an extension of its Yokkaichi joint venture through December 2034, including payments to Kioxia tied to manufacturing services and supply arrangements. If the NAND supercycle is entering a more mature phase, the fab utilization economics at these shared facilities matter enormously.

NVIDIA (NVDA). The indirect name. The AI workloads driving NAND datacenter demand sit alongside NVIDIA graphics processing units in the AI server stack and must match their throughput demands. A moderation in hyperscaler NAND procurement pace would reflect the same capex rationalization debate that NVIDIA faces every quarter. The two demand signals are not independent.