SanDisk Stock Crashed 13% and What the Memory Selloff Signals

SanDisk lost 12.63% on Monday, July 13, 2026, and finished as the single worst performer in the entire market. It fell harder than SK Hynix's US-listed line, which dropped 9.32%, harder than Seagate, and nearly three times harder than Micron, which lost 4.32%. None of the damage started with SanDisk.

The trigger came out of Seoul, where SK Hynix crashed 15.4% in its largest single-day fall on record. Everything that hit SNDK was downstream of that, amplified because storage sits at the highest-beta end of the AI supply chain.

Monday's snapshot, July 13, 2026

- SanDisk (SNDK) down 12.63%, the worst single-name performance in the market

- SK Hynix down 15.4% in Seoul, its worst day ever, three days after a record Nasdaq debut

- Micron (MU) at $937.00, down 5.51%, with Marvell down 6.79% and Intel down 4.47%

- Philadelphia Semiconductor Index down 10.8%, roughly $1.3 trillion of market value erased

- Brent crude at $83.31, up 10.76%, its biggest one-day gain in over six years

- Bitcoin at $62,470, no shelter from the risk-off move

Here is what actually broke, why storage broke hardest, and what the oil market has to do with any of it.

What Actually Triggered the Memory Selloff

Korea Investment & Securities cut its estimates on SK Hynix, and the numbers were ugly enough to reset the entire memory complex. The firm now projects Q2 2026 operating profit of 60.4 trillion won against a consensus near 65 trillion, roughly 8% below the Street, paired with 9% and 11% cuts to its 2026 and 2027 estimates.

Read that carefully. It is a projection from one brokerage, not a reported result, and a single analyst revision took 15.4% out of one of the largest memory manufacturers on earth. That says more about how tightly wound positioning had become than about the business.

The timing made it worse. The cut landed three days after SK Hynix's July 10 Nasdaq debut, a listing that raised a record $26.5 billion. Every account that bought that deal was underwater within 72 hours, and forced selling from a fresh, crowded shareholder base is how one estimate revision becomes a record down day.

Why SanDisk Took the Most Damage

SanDisk does not have an SK Hynix problem. It has a beta problem, and this is the part worth understanding because it will repeat.

Memory and storage are the most cyclical, most commoditized layer of the AI stack. A NAND die is a NAND die. Unlike NVIDIA, which sells a proprietary compute platform with a software moat and can hold its price, storage vendors sell an interchangeable product into a market that clears on price alone. They have the least pricing power in the chain and the most operating leverage, a combination that cuts both ways.

When memory prices rise, almost every incremental dollar of revenue falls to the bottom line because fab costs are largely fixed. Earnings estimates go vertical and the stock outruns the sector on the way up, which is exactly what storage did through the first half of 2026. But operating leverage in reverse is the same math with the sign flipped. A modest cut to pricing assumptions takes a disproportionate bite out of forward earnings, and a stock priced for the up-cycle has a long way to fall before it finds a floor.

So when the market decided on Monday that AI memory earnings might have already peaked, it did not re-rate every semiconductor name equally. It punished them in order of beta. Marvell took 6.79%, Micron took 5.51%, and the storage names took double digits. On July 13 the highest-beta name in the group was SanDisk.

The Oil Shock Underneath the Chip Selloff

None of this happened in a vacuum. Over the weekend, Trump reinstated a blockade on Iranian shipping through the Strait of Hormuz and the mid-June ceasefire collapsed with it. Brent closed Monday at $83.31, up 10.76%, its biggest one-day gain in more than six years, while WTI rose 9.08% to $77.99.

The transmission chain from a Middle East headline to a NAND stock is shorter than it looks. An oil shock is inflationary, inflation reprices the rate path toward higher-for-longer, and higher-for-longer compresses the multiple on every long-duration asset in the market. That is precisely the bucket AI semiconductors have been sitting in for two years.

Nothing about SanDisk's business changed on Sunday night. What changed is the rate at which the market discounts its future earnings back to today, and for a name whose valuation rests on profits arriving in 2027 and 2028, that adjustment is violent. Crypto felt the same tremor, with Bitcoin at $62,470 and no diversification benefit when the shock is a rates shock.

The Rotation Tell Hiding in Monday's Tape

Here is the detail most of the coverage missed. On the same day the chip complex bled out, three of the biggest companies in the market closed green.

Money did not leave the AI theme on Monday. It rotated out of the companies that receive AI capex and into the megacaps that spend it, which is a positioning shift rather than a panic. The market has started questioning the return on the spending, not the spending itself.

Hyperscaler AI capex jumped 67% to roughly $650 billion annually, with guidance stretching toward $725 billion for 2026. When your customers spend three quarters of a trillion dollars a year and the Street starts asking what the payback looks like, the suppliers get repriced first. The Philadelphia Semiconductor Index is down 10.8% in this drawdown, roughly $1.3 trillion of value gone, and Morgan Stanley calls it a "mid-cycle reset" rather than a peak.

The Memory Bull Case Is Being Stress-Tested, Not Buried

Be honest about the other side, because it is louder than the tape suggests. Baird upgraded Micron to Outperform, Citi still names Micron its top US semiconductor pick, and UBS says memory prices are set to rise from here, the exact opposite of what Korea Investment & Securities just modeled.

Two credible camps are reading the same NAND and DRAM supply curves and reaching opposite conclusions, and price action is doing what it always does when analysts disagree. It picks the more fearful answer first and asks questions later. Samsung, SK Hynix, Micron and SanDisk all still sell into demand nobody claims is falling.

The resolution will not come from another analyst note. ASML reports Wednesday, July 15, and its bookings tell you what the fabs are committing to build. TSMC reports Thursday, July 16, and its revenue and capex commentary is the best proxy for the question everyone is now asking, which is if AI demand is decelerating or simply being repriced.

Frequently Asked Questions

Why did SanDisk stock crash 13% today?

SanDisk fell 12.63% on July 13, 2026, on a sector-wide memory de-rating that started with SK Hynix crashing 15.4% in Seoul after a brokerage estimate cut. Nothing company-specific happened at SanDisk. It is the highest-beta name in storage, so it absorbed the largest share of the sector's losses.

Is the memory chip cycle peaking in 2026?

Nobody knows yet, and the analysts are openly split. Korea Investment & Securities cut its SK Hynix estimates 9% for 2026 and 11% for 2027, while UBS says memory prices are set to rise and Citi still calls Micron its top semiconductor pick.

Why does an oil price spike hurt semiconductor stocks?

Higher oil is inflationary, which pushes the expected rate path higher for longer and raises the discount rate applied to future earnings. High-multiple growth names carry most of their value in distant earnings, so they lose the most when that discount rate rises.

Are memory stocks a buy after this selloff?

The honest answer is that it depends on what ASML and TSMC report this week, and buying before those prints is a bet on data you do not have. What the tape says right now is that money is rotating toward AI spenders and away from AI suppliers, and that rotation usually runs longer than one session.

The Bottom Line

One brokerage note took 15.4% out of SK Hynix and 12.63% out of SanDisk without a single company reporting a miss, which tells you how much of the memory trade was built on estimate momentum rather than delivered earnings. The high-beta names always find out first.

The decision rules for the rest of the week are simple. Clean bookings from ASML and clean numbers from TSMC, and the "mid-cycle reset" framing holds, which makes Monday a de-rating rather than a top. A soft print from either, and the market gets the confirmation it went hunting for. Until then, storage stays the fastest way to be right and the fastest way to be wrong.

This article is for informational purposes only and does not constitute financial or investment advice. Cryptocurrency trading involves substantial risk. Always conduct your own research before making trading decisions.