SanDisk Has Fallen 45% From Its Peak. The Stock Split Story Misses the Point.

The market is debating whether SanDisk deserves a stock split. That is the wrong question.

SanDisk (SNDK) spun off from Western Digital in February 2025 at around $22 per share. It hit an all-time high of $2,354 on June 22, 2026 — a gain of roughly 3,000% in 14 months. As of August 7, it sits around $1,212, down 45% from that peak. The split conversation only makes sense when the price action is going one direction. Right now it isn't.

What matters is whether this pullback is the beginning of a cyclical collapse or an overdone correction in a company that has fundamentally transformed. The numbers say the latter. The risks say investors still need to be careful about what the next 12 months can actually deliver.

The transformation is real

SanDisk's fiscal 2026 — its first full year as an independent company — produced $20.25 billion in revenue, up 175% year-over-year from the $7.36 billion it generated under Western Digital's ownership. Gross margin expanded from 30.1% to 71.5%. Free cash flow hit $11.49 billion. Return on invested capital was 80.6%. The company carries essentially no net debt and holds $4.76 billion in cash.

These aren't recovery numbers from a cyclical trough. They're the profile of a company that has shifted its revenue mix into the highest-margin end of the flash storage market.

The segment breakdown tells the story. Datacenter revenue — the AI-infrastructure bucket — grew 437% year-over-year to $5.15 billion for the full fiscal year, representing roughly 25% of total sales. The Edge segment (industrial, automotive, enterprise) grew 195% to $12.16 billion. Consumer — the traditional flash drive and memory card business that drove the old cycle — shrank to $2.94 billion, or just 15% of revenue. In the fourth quarter alone, datacenter was $2.98 billion (33% of quarterly revenue) and consumer was $556 million (6%).

The business SanDisk was nine months ago no longer exists.

The margin question

The fourth quarter closed with a gross margin of 84.6%, up from 26.2% a year earlier. About two-thirds of sequential revenue growth came from pricing, one-third from volume. That margin level is extraordinary for a hardware manufacturer and would be remarkable for any company.

But it's also where the bear case lives. NAND flash is a commodity. Historically, NAND upcycles have lasted 4 to 7 quarters before prices reverse and margins collapse. During previous downturns, industry players saw net margins swing negative — SK Hynix fell to roughly -28% at its cyclical low. An 84.6% gross margin in a commodity business is unsustainable indefinitely. No matter how compelling the AI demand curve, someone will eventually build capacity or find a substitute.

Management's answer is the New Business Model (NBM). These are multi-year agreements with firm financial commitments and upfront cash deposits from hyperscalers and enterprise customers. SanDisk has signed 10 NBM deals since April. The company disclosed $41.6 billion in remaining performance obligations from those contracts alone, which covers roughly one-third of its fiscal 2027 bit output.

That's a genuine structural shift — moving a portion of the business from a spot-market commodity to a contracted service model. But two-thirds of production remains exposed to open-market NAND pricing. If the cycle turns, the unprotected portion of the business gets hit first.

The selloff mechanics

The 45% decline from the $2,354 peak came in two waves. The first was a 24% drop over three sessions in late July, triggered by the CXMT IPO. CXMT is a Chinese DRAM (not NAND) manufacturer, but the memory sector got swept up in a basket unwind that also dragged down Micron and SK Hynix. That was a sector trade, not a SanDisk-specific repricing.

The second wave hit on August 5, when SanDisk reported fiscal Q4 results. Revenue of $8.97 billion beat the $8.44 billion consensus. Non-GAAP EPS of $39.25 beat the $34.80 estimate. But Q1 FY2027 guidance of $10.3 to $10.8 billion came in slightly below the ~$10.82 billion consensus. The stock fell 13% the next day.

The market wanted perpetual acceleration. It got a company that grew 175% in its first fiscal year, then guided for 50% sequential growth. That's still massive. But after a 3,000% rally, even good news that doesn't match perfection gets punished.

Valuation in the pullback

At $1,212, SanDisk's trailing P/E sits around 15.8x based on the last twelve months of earnings. The forward multiple gets messy depending on what earnings base you use, but the underlying arithmetic is striking: a company generating $11.5 billion in free cash flow on $20.2 billion of revenue, with a market cap of roughly $181 billion.

That's a price-to-free-cash-flow multiple of about 15.8x. For a business with 71.5% gross margins and 80%+ return on invested capital, the market is pricing in a scenario where those margins compress dramatically and the AI-driven datacenter demand stalls. The forward P/E is arguably in the low-to-mid single digits if you extrapolate from the recent quarterly run-rate, assuming earnings power holds even modestly.

The question isn't whether SanDisk is cheap in a vacuum. The question is whether the underlying earnings power is durable enough to justify the market cap, or whether this is a cyclical peak being valued like a structural shift.

Price action says: selling hasn't exhausted

Technically, the stock is below its 50-day moving average of $1,688, and the MACD is negative at -126.83. The RSI at 41.8 is not in oversold territory. Over the past 20 days the stock has fallen 36.7%. One-day volatility is running at 10.57%.

There's no bear trap signal yet. The volume remains heavy — daily turnover of $17.2 billion — and the price has not found a floor. The 200-day moving average at $872 is the only technical support level that would represent a meaningful discount. The stock hasn't reached it.

This is the difference between a stock that's been beaten down into a buying zone and one that's still falling. SanDisk is arguably the latter. The fundamentals don't support a collapse from here, but the price action doesn't yet show the kind of selling exhaustion that marks a contrarian entry.

The NBM floor vs. the open-market ceiling

The structural change matters. Goldman Sachs forecasts NAND shortages of 4.4% in 2026 and 4.6% in 2027. TrendForce projects NAND contract price growth will slow to 10-15% quarter-over-quarter in Q3 calendar 2026, down sharply from 70-75% in the prior quarter. SanDisk has reportedly sold out its production capacity for the remainder of 2026, with 2027 capacity already being purchased.

The NBMs create a floor. But they don't create a ceiling. Two-thirds of production is still at the mercy of whatever the open market pays. If Chinese competitors like YMTC scale domestic NAND capacity with homegrown equipment — the long-term threat that's always one cycle out — the open-market portion of SanDisk's business could face margin pressure faster than investors expect.

What to do

SanDisk is not a falling knife. The fundamentals are too strong for that. Revenue is growing, margins are expanding, cash flow is exceptional, and the NBM model is real structural progress. But it's not a confirmed bottom either.

The 3,000% rally has baked in a lot of perfection. The 45% pullback is a correction, not a panic sell-off. The price action hasn't yet told us where the floor is. Until the RSI dips into oversold territory and daily volume starts drying up — signs that selling pressure is exhausting — the lower-risk entry is below the current level.

A position in the $800-1,000 range, near the 200-day moving average, would offer asymmetric risk/reward. Below that, the forward valuation on even a modest earnings decline becomes hard to argue against. Above $1,400, you're chasing momentum in a stock that just lost 45%.

I don't think investors need to rush in here. The setup is constructive for a patient buyer, but the timing isn't confirmed. Watch for selling exhaustion — not the stock split headline. That one doesn't change the underlying business.

I would reassess the bullish case if Q1 FY2027 revenue guidance falls below $9 billion (suggesting the pricing deceleration is steeper than expected), if NBM contract signings slow materially, or if the stock breaks below the 200-day moving average without a reversal. Any of those would suggest the cycle is turning faster than the current pricing implies.