SanDisk 30% Crash|Samsungs Record Profit Triggered the Sell-Off

The Day Good News Broke the Trade

SanDisk shares have dropped 30% from their June 22 peak, with another 11% decline on July 9 alone — even as the AI memory thesis has never looked stronger on paper.

The week's defining event was not a demand warning. Samsung Electronics reported operating profit of 89.4 trillion won, a 19-fold jump from the prior year, with revenue up 129% year over year. Those are record numbers by any measure.

Yet Samsung shares fell 6.9% in Seoul trading on the announcement. SK Hynix dropped 6.1%. The Korean Kospi briefly entered bear market territory, falling as much as 8.2%, dragged by chips. And the damage spread instantly to US names — Micron fell 7.3%, Western Digital dropped 8.14%, and SanDisk led all declines.

The surface read says something broke. The deeper read says the best possible news was already priced into every share, and the moment it was confirmed rather than merely anticipated, the trade had nowhere higher to go.

That bottleneck — earnings-as-exit-signal rather than earnings-as-catalyst — is the mechanism this video traces, because whether it signals a temporary reset or a cycle turn decides whether the 30% pullback is an entry or a trap.

Why Record Earnings Became the Sell Signal

SanDisk's 2026 run tells the problem plainly. The stock rose roughly 858% in the first half of the year, from a 52-week low of $40.10 to its June peak. Micron nearly tripled. The Roundhill Memory ETF, whose top three holdings are Samsung at 25%, SK Hynix at 24%, and Micron at 24%, rose in lockstep.

That compression of future value into current price is what makes Samsung's record earnings a clearing event rather than a catalyst.

The rotation mechanics confirm this. On the same day SanDisk fell 14% in early July, the Dow Jones Industrial Average hit a record closing high. Capital was not leaving the market — it was leaving memory and rotating into blue chips and software. The June jobs report came in at 57,000 payrolls versus expectations, dimming Fed rate-hike odds and lifting rate-sensitive names. But the Nasdaq fell as chip stocks dragged it lower. These moves happened simultaneously and in opposite directions.

What makes this a deeper problem than simple profit-taking is the $1.5 trillion in semiconductor market cap erased since June 25. Micron alone shed roughly $350 billion over that stretch. Twenty-five semiconductor names in the group are down at least 20% from their June peaks. The Roundhill Memory ETF is down 19% from its peak.

That scale of move is not portfolio rebalancing. It reflects a repricing of the AI memory trade's return horizon — specifically, the question of whether the pricing power that drove margin expansion survives the supply additions already announced by Samsung and SK Hynix coming online through 2027.

The reversal card the market is playing is not "AI demand is wrong." It is "the cycle's best earnings are now visible, which means the easy money has been made, and the next question is how much supply enters before demand grows into it."

Two Analysts, One Stock, Opposite Verdicts

The conflict is not theoretical. Two named analysts looked at the same SanDisk at the same moment and drew opposite conclusions — and both can cite evidence from the same earnings cycle.

Bank of America's Wamsi Mohan raised his SanDisk price target to $2,500 from $2,100 and reiterated a Buy rating, projecting June-quarter revenue of $9.1 billion against company guidance of $7.75 to $8.25 billion. The core thesis is that NAND supply-demand imbalance persists through 2027, that SanDisk's new long-term supply agreements feature fixed or range-bound pricing with upfront financial commitments, and that these contracts structurally reduce earnings volatility. Bernstein separately raised its target to $3,000 from $1,700, also Outperform.

Morningstar's Lorraine Tan told Bloomberg TV that a large slice of AI names could give back 20 to 30 percent before becoming buyable again. The concern is the inverse of BofA's: Samsung and SK Hynix are expanding capacity, and announced supply additions are expected to soften memory pricing precisely as AI capital expenditure is projected to peak in 2026 and taper after. That combination targets the same pricing power BofA is betting on.

The buried assumption in BofA's read is that fixed-pricing contracts insulate SanDisk from a spot-price reset. But those contracts were signed during the current shortage. Whether they hold their terms if buyers face oversupply — and whether counterparties honor them or renegotiate — is the untested premise.

The buried assumption in Morningstar's read is that Samsung and SK Hynix can execute their capacity additions on schedule. The same chip infrastructure buildout that took longer than expected on the demand side has repeatedly surprised on the supply side. HBM production in particular uses three times the wafer capacity of conventional DRAM, meaning capacity additions take longer to translate into NAND oversupply than a simple tonnage count implies.

Neither analyst is fabricating the data. They are drawing different conclusions from the same supply-demand curve, at different points on the horizon.

The Variable That Decides the Trade

SanDisk's upcoming earnings report is the sharpest discriminator available. The company guided June-quarter revenue of $7.75 to $8.25 billion. Bank of America projects $9.1 billion and earnings per share of $37.01. If the actual print arrives above guidance and the fixed-pricing contracts hold into the second half, the BofA thesis survives the Morningstar challenge.

If NAND spot pricing shows softening or if contract renewal terms reveal buyer pushback, Morningstar's reset thesis gains the evidence it currently lacks.

The cleaner leading signal is SanDisk's enterprise SSD order book and the pricing terms in its long-term supply agreements — these will be disclosed on the earnings call before quarterly aggregates confirm the trend. That is the variable holders and watch-list candidates should track before the broader revenue print.

For holders, the existing position's case rests on BofA's contract thesis. The invalidation trigger is any language on the earnings call signaling that fixed-pricing terms are being renegotiated or that customers are delaying draw-downs under the take-or-pay structure. That confirmation makes the 30% pullback a trap rather than a flush.

For those watching from the sideline, the entry case requires the opposite: confirmation that pricing holds through the current quarter, volume commitments are being met, and Samsung's and SK Hynix's capacity additions remain on a 2027 timeline rather than pulling forward. A revenue print above BofA's $9.1 billion forecast with maintained contract terms would be the signal the rotation was technical rather than fundamental — and the 30% pullback would resolve as an entry window rather than a warning.

The last note: SanDisk stock has gained more than 3,750% over the past 12 months. A 30% pullback from a 52-week high, on record-beating AI memory demand, and with two major analysts maintaining Buy, does not require a buyer to believe the cycle is broken. It requires only that the next quarterly print confirms the contracts the cycle's pricing power rests on.