This piece follows two threads. One is the existing SanDisk report, SanDisk +3,685% in 1 Year, Has the NAND Cycle Ended?. The other is the memory/FX series: Every Time the Won Weakens, SK Hynix Cracks. How Low?, the structure edition Memory's Great Power Reversal, and the event edition KOSPI 8000 Breaks: Real Crisis or a Fit?.
The question is simple: "SanDisk's target keeps climbing to $3,000, so why does the stock keep falling?" Korea (Hynix, Samsung) you can chalk up to FX, but for SanDisk the news is all good. This piece unpacks that paradox. Jargon is explained inline, with a glossary at the bottom. Figures are cross-checked against reporting and data as of July 2, 2026, and estimates are marked as such.
Today's Scene: Memory Is Blue in the US Too
Memory is blue in the US market too (in the US, down is red, but our series uses the Korean convention where blue means down).
SNDK, July 2, 2026
- Price
- $1,957.83 (about 3.01M won), 7/2 close. Second straight down day from 7/1 $2,032 (-9.91%)
- Recent high
- $2,335 (6/25 close), $2,354 intraday (6/22). About -16% from the high
- YTD return
- About +725% (around +880% at the June high)
- Valuation
- Forward P/E about 77-79x, a 'priced for perfection' level
- Latest targets
- Bernstein $3,000, BofA $2,500. Targets were actually raised
| Indicator | Value | Note |
|---|---|---|
| SNDK price | $1,957.83 (about 3.01M won) | 7/2 close |
| Recent path | 7/1 $2,032 (-9.91%) → 7/2 $1,957 | Second straight down day |
| Recent high | $2,335 (6/25 close), $2,354 intraday (6/22) | about -16% from high |
| YTD return | about +725% (around +880% at the June high) | vs. year start |
| Valuation | Forward P/E about 77-79x | "Priced for perfection" level |
| Latest targets | Bernstein $3,000, BofA $2,500 | Targets actually raised |
Here is the strange part. On Tuesday, when Bernstein set a $3,000 target, it jumped +11%, then reversed -11% the next day and slid to around $2,016. It did so even though BofA raised to $2,500 the same day. In other words, it sells off on the very day good news (a target hike) lands. The observation is accurate.
Why? The answer is not one thing but four. And one of those four happens to be SK Hynix.
The Root of the Paradox: The Target Rises but the Price Falls
The Drop Was Already Foretold
Two months ago, the SanDisk analyst report (2026-05-09) warned of exactly this situation when the stock was $1,562 (YTD +558%). Rereading it is eerie.
- "No chasing in the short term (HOLD). Enter on scaling into $1,300-1,400." The market ignored it and raced to $2,354.
- "Beta 5.04, one macro shock can trigger a -30%+ crash." This is that phase.
- "Base scenario (45% probability), 12-month price $900-1,300," it saw a decline as the base case from the start.
- "Supply-normalization risk if Samsung and SK Hynix restart capex in 2027." The two companies made a 4,700-trillion-won expansion official, pulling that risk forward.
In other words, today's drop is not a sudden shock but the mean reversion that valuation and beta foretold. The problem is that in the meantime the price blew past even the report's bull target ($1,886) to the $2,300s. It went too far, so the give-back is large too.
For a Stock Up 850%, Good News Is Selling Liquidity
One core concept. A stock moves on "are expectations already in the price," not "is it a good company."
SanDisk is up 700-880% YTD at a 77-79x P/E. At that level, several years of good news are already embedded (priced for perfection). When news like "$3,000 target" lands in this state, it becomes an exit through which early buyers sell out, in a zone with no one left to buy. This is sell-on-news, the same mechanism I described for the Hynix ADR in the Korea piece, except SanDisk's valuation is far more extreme, so the reaction is more violent.
Why It Falls: Four Forces
① Sell-on-News plus Valuation (the foundation)
Just as above. Up 700%+ YTD at a 77-79x P/E is "a price premised on perfection." Any good news is already reflected, so good news becomes a profit-taking trigger rather than upside fuel. Add a beta of 5.04 (ultra-high volatility that moves 5% when the market moves 1%), and once a correction starts, the drawdown is amplified. For reference, Korea even has leveraged products like a TRADR SanDisk 2x ETF (the same "single-stock leverage" logic as the Korea piece), so on down days mechanical selling piles on and deepens the whipsaw.
② The Same Macro Deleveraging as Korea, Minus the FX
The 7/2 US session saw semiconductors collapse. Philadelphia Semiconductor Index -6.27%, Micron -10.57%. The market is now in a phase of pulling money out of AI chips and hardware and rotating into AI software. SanDisk, a high-beta pure-memory name, sits dead center of that wave.
This shares its root with the deleveraging Korean chips are taking (see the event edition). The only difference is one thing: Korea has an extra layer of won weakness (1,540) that amplifies foreign selling, whereas SanDisk is a dollar asset and lacks that layer. So it is not "because of FX," but it takes the same bigger wave of AI-memory deleveraging all the same.
③ The SK Hynix ADR (7/10) Steals SanDisk's Scarcity Premium
This is SanDisk's own, and the most interesting, factor.
Until now SanDisk was "almost the only pure AI-memory name US institutions could easily buy." Micron has a large DRAM mix, and Samsung and Hynix are Korea-listed, so US passive and institutional money found them cumbersome to hold. As a result, SanDisk carried a "scarcity premium," because US dollars wanting to bet on pure memory had nowhere else to go.
But on July 10, the larger, more diversified SK Hynix lists on Nasdaq via ADR, and at a record scale ($29 billion, #1 in HBM globally) at that. For US investors, there is now a more attractive option for "dollars to put into AI memory." So SanDisk's monopoly position wobbles, and that premium is re-rated down.
In the Korea piece I addressed "will the Hynix ADR steal the Korean common's liquidity?" But what the ADR actually steals is not the Korean common (arbitrage re-ties the common's price) but rather SanDisk's scarcity premium. It is a mirror image where the same ADR works in the opposite direction.
④ Western Digital Overhang plus NAND-Cycle Skepticism
Overhang (pending supply) is the first. SanDisk was spun off from Western Digital (WD) on February 24, 2025, and WD keeps selling its stake via secondary offerings (5.82 million shares at $545 on February 18, 2026, plus shelf registrations and further sale programs). Supply that comes out constantly, regardless of fundamentals, presses the price.
NAND being the weak link of the supercycle is the second. The real engine of this cycle is HBM and DRAM. There is suspicion that SanDisk's core, NAND, is relatively more commoditized and thus more vulnerable to oversupply. Management says "this NAND cycle is different" (that NBM contracts weakened the cyclicality), but the market discounts that good news somewhat.
Korea (Hynix, Samsung) and SanDisk: Same Disease, Different Clothes
The three names collapsed in the same week, but the "outerwear" of the cause differs. Strip it off and the core is the same.
| Outerwear (that name's own layer) | Core (shared) | |
|---|---|---|
| Hynix, Samsung | Won weakness (1,540) → foreign-selling loop + single-stock leverage | AI-memory deleveraging (supercycle winners sold first in a liquidity-withdrawal phase) |
| SanDisk | 850% surge and 77x P/E valuation + Hynix-ADR premium erosion + WD overhang | Same as above |
The common root is exactly what the structure and event editions described. The winners of the "power-reversal" supercycle where memory became the price-setter (Hynix, Samsung, Micron, SanDisk) get sold first and hardest in a deleveraging phase stacked with rates, the AI narrative, and leverage. Because the most-appreciated winner is the cash machine.
That said, SanDisk also has a shield others lack: NBM (multi-year supply contracts). As of the 5/9 report it had already locked in $42 billion of revenue, $11 billion of financial guarantees, and one-third of FY27 bits. In theory it structurally dampens memory cyclicality. Yet valuation beats this shield right now. No matter how good the contracts laid down, at a 77x P/E the market treats even those contracts as already in the price.
How the Hynix ADR Spreads Across the Whole Memory Group: SanDisk vs. Micron
Then a question arises: "Micron also fell -10.57% on 7/2, so is that because of the Hynix ADR too?" The answer is "partly yes, but not the main culprit." The Hynix ADR is not a Hynix-only event; it is a "gravity event" that soaks up capital and attention within the memory sector, so it spreads to neighbors too. But its weight is smaller than you might think.
First, the Weighting (to avoid misreading)
The overwhelming culprit behind the 7/2 crash is macro deleveraging (Philadelphia Semiconductor -6.27%, rotation from AI chips into software) plus an extreme valuation give-back. Weighting it roughly:
| Factor | Estimated weight | Nature |
|---|---|---|
| Macro deleveraging / rotation | ~60% | Whole market (event edition) |
| Valuation / sell-on-news | ~25% | Stock overvaluation |
| Hynix ADR related | ~10-15% | Intra-sector capital shift (SanDisk larger than Micron) |
| Idiosyncratic overhang, etc. | Remainder | WD sales, etc. |
So the ADR is one layer on top, not the floor. Not "it crashed because of the ADR" but "the ADR helped it along a bit" is accurate.
And the Channels Differ for the Two Companies
| Channel | Directness | Size | |
|---|---|---|---|
| SanDisk (SNDK) | Scarcity-premium erosion | Direct | Meaningful (one of four factors) |
| Micron (MU) | ① Reinforced oversupply narrative ② Diluted relative appeal ③ Passive rebalancing | Indirect | Small |
SanDisk is direct, so it matters. The source of SanDisk's premium was the scarcity of being "almost the only pure-memory name easily bought in the US." When the larger, more diversified pure-memory leader (Hynix) lists in the US, the options for "dollars to put into pure memory" grow and that scarcity premium is directly shaved. So for SanDisk, the ADR actually operates as one of the four downside factors.
Micron is indirect, weak but present. Micron is already US-listed, large, and ETF-included, so it has no "access premium" to lose. So the direct hit is small. Instead it seeps in three ways. First, reinforced oversupply narrative, and this is the biggest. The Hynix ADR raises $29 billion as HBM and expansion money. That is money that increases future supply, so it stokes "2027-28 oversupply" fear and re-rates the whole memory group down. Micron, in the same group, gets hit too (the same logic as the Korea piece's "4,700-trillion-won capex is the seed of future oversupply"). Second, diluted relative appeal. The seat of "the flagship HBM/memory large-cap you can buy in the US" is now shared with Hynix. The "US memory flagship" premium that concentrated on Micron thins. Third, diluted passive rebalancing. When the Hynix ADR is included in US semiconductor ETFs like SOXX, existing holdings (Micron included) are trimmed slightly to make room, or new passive money is spread across more names.
The Timing Is a Hint
The listing is 7/10, but the crash is before it (7/2). So the ADR effect showing up now is not "the listing itself" but positioning ahead of the listing: trimming a crowded memory trade before a big event, or selling other memory names to raise ammunition for the Hynix ADR.
A Double-Edged Sword
If the 7/10 debut draws strong demand, it becomes validation that "the supercycle is real" and can lift SanDisk and Micron too; if it flops, it drags the whole group down. In other words, 7/10 is a fork-in-the-road event for the entire memory sector. That is why scenario ② below (ADR event volatility) is wide open in both directions.
So How Low Can It Go: Scenarios
Coordinates and Support
Reading the daily data:
| Phase | Price zone | Note |
|---|---|---|
| June high | $2,335 (6/25 close), $2,354 intraday (6/22) | Near all-time high |
| Whipsaw zone | 6/24 $1,914 → 6/25 $2,335 (+22%) → 7/2 $1,957 | The fingerprint of beta 5.04 |
| 7/2 now | $1,957.83 | -16% from high |
First support is around $1,900 (where the 6/17 and 6/24 lows and the 6/29 intraday low overlap). If that breaks, second is $1,560-1,600 (the 5/8 prior high, now attempting to flip to support). Below that is $1,300-1,400 (the early-May supply zone and the "scale-in zone" the 5/9 report proposed).
Reviewing Against the Existing Report's Targets
The 5/9 report's probability-weighted 12-month target was $1,800 (recomputed bull $1,886 / base $1,038 / bear $513). The current $1,957 already sits above even that bull target. By the report's yardstick alone it is still in overvalued territory, and the drop reads as the process of closing that gap.
Four 12-Month Scenarios (Subjective Estimate)
| Scenario | Price zone | Premise | Probability (est.) |
|---|---|---|---|
| ① Valuation normalization (underway) | $1,500-1,900 | Macro deleveraging + sell-on-news + ADR premium erosion. Fundamentals firm, pure valuation give-back | ~40% |
| ② ADR event volatility (around 7/10) | $1,400-2,000, high vol | Capital shift/sentiment weakness on the Hynix ADR, then a bounce if memory strength is reconfirmed | ~20% |
| ③ Cycle-peak recognition | $900-1,300 | NAND prices roll over + GM normalizes + Samsung/Hynix capex adds supply. The 5/9 report's base to bear | ~20% |
| ④ Re-rally | $2,300-3,000 | 8/13 earnings with more NBM signings and Stargate revenue + AI inference demand explosion → re-rating resumes | ~20% |
The direct answer to "how low":
The most probable near-term floor is $1,500-1,900. Not because fundamentals broke, but the spot where deleveraging, the ADR, and valuation force the spike to be given back. $900-1,300 only opens if a "real cycle peak" is confirmed, that is, when NAND ASP, bits, and GM actually roll over or the 8/13 earnings disappoint. We are not at that signal yet. Conversely, if the 8/13 earnings stack more NBM, today's drop becomes "a correction that only shaved off the expensive froth" and a re-rally is possible too.
Is This a Fit, or a Real Decline? How to Tell
The same principle as the Korea piece: watch leading indicators, not the price. A checklist tuned for SanDisk:
- NAND spot ASP. It leads contract prices and revenue. If it rolls over first, that is a real cycle signal.
- Bit-shipment trend. Already slowing to -high teens. If price rolls over too, revenue deleverages simultaneously.
- New NBM signings (8/13 earnings). If contracts keep stacking, the "cycle buffer" narrative holds; if they stall, it reverts to a cyclical multiple.
- Gross margin. 78.4% is an industry all-time high. Missing the Q4 guide of 80% is peak recognition.
- Samsung and SK Hynix NAND capex. The root of supply normalization. How much of the 4,700-trillion-won expansion goes to NAND.
The current read (July 2, 2026): what is falling now is only the price (valuation, flows, sentiment), and there is no sign yet that fundamental leading indicators like NAND ASP and NBM have rolled over. If so, today's -16% is closer to scenario ① (valuation normalization). The real fork is the August 13 Q4 earnings, where NBM, GM, and Stargate revenue are decided.
So How Does It Trend for Now: The Triggers That Decide Direction
The honest answer to "so does it just keep falling for now?" is "the weight leans toward weak and high-volatility, but it is not a one-way decline and hinges on a few triggers."
The reason the weight leans downside for now is that the four pressing forces from earlier are all still on: macro deleveraging and the AI-chips-to-software rotation (underway), the valuation give-back behind a 77x P/E and 700%+ (still expensive), positioning ahead of the 7/10 ADR (lasts until the event), and the Western Digital overhang (constant). So the base picture is scenario ① (valuation normalization, a $1,500-1,900 box) plus high volatility.
But do not conclude "one-way decline." Today's drop is not broken fundamentals but a valuation/flow give-back. Such declines can V-bounce on a single catalyst. In fact, being beta 5.04, a +22% single-day bounce (6/24 $1,914 to 6/25 $2,335) already happened. So the right phrase is not "slowly bleeds down" but "tilted down but swings hard both ways." Even with the direction down, it is violent in both directions.
The four triggers that decide direction:
| Trigger | Timing | Up (bounce) | Down (further drop) |
|---|---|---|---|
| Hynix ADR debut | 7/10 | Strong demand → "supercycle validation" → group-wide bounce | Weak → group-wide weakness |
| Q4 FY26 earnings | 8/13 | More NBM, GM 80%+, Stargate revenue | GM misses 80%, NBM stalls → opens $900-1,300 |
| Macro | Constant | Rates and AI-software rotation calm | Deleveraging persists, rates rise |
| NAND leading indicators | Constant | Spot ASP, bits, GM stay firm | ASP, bits, GM actually roll over |
7/10 is a double-edged sword, the biggest near-term variable that can move big either way. 8/13 is the real fundamental fork, where the leading indicators above get confirmed in numbers.
To sum up the near-term outlook, a weak, high-volatility bias is right. The floor is roughly $1,500-1,900 (first support around $1,900, next around $1,560), and below that $900-1,300 only opens if NAND leading indicators actually roll over or the 8/13 earnings disappoint. But since this drop is driven by valuation, flows, and ADR positioning rather than fundamentals, it is a give-back that can flip up at any time around 7/10 and 8/13. This is scenario analysis, not investment advice. Being a beta 5.04 name, it can move big against the direction too.
In One Line
SanDisk falling on good news is not because the news is bad. Up 850% at a 77x P/E, good news became selling liquidity (sell-on-news), it takes the same AI-memory deleveraging as Korea minus only the FX, the 7/10 Hynix ADR steals its "only pure-memory name in the US" scarcity premium, and the Western Digital overhang plus NAND-cycle skepticism piled on. Even with the NBM shield, valuation wins right now. The most probable near-term floor is $1,500-1,900, and the real cycle drop to $900-1,300 only opens if NAND leading indicators roll over or the 8/13 earnings disappoint. Korea (an FX loop) and SanDisk (valuation + ADR + overhang) just wear different clothes; the core is the same deleveraging.
Glossary
| Term | Plain explanation |
|---|---|
| Sell-on-news | A stock that rose on anticipation of good news falls on profit-taking once that news actually lands. The more it had risen, the worse. |
| Priced for perfection | A state where all the future good is already reflected in the price. Then ordinary good news can't lift it, and small disappointments knock it down hard. |
| P/E (price-to-earnings) | Price divided by earnings per share. 77-79x means 77-79 years of earnings are prepaid into the price, so very high expectations are embedded. |
| Beta | Volatility relative to the market. Beta 5.04 means the stock moves about 5% when the market moves 1%, ultra-high volatility. |
| Scarcity premium | A markup that attaches "because there's nothing else to buy." SanDisk got it as "the only pure-memory name easily bought in the US," and it vanishes once the Hynix ADR appears as a substitute. |
| De-rating | The price falling as the market assigns a lower P/E to the same earnings (the opposite of re-rating). |
| Deleveraging / rotation | Deleveraging is cutting debt and risk assets to raise cash. Rotation is money shifting by selling sector A to buy sector B (here, AI chips to AI software). |
| Overhang / secondary offering | Overhang is the burden of pending supply about to hit the market. A secondary is an existing major holder (WD) selling its stake into the market. |
| NAND / ASP / bit shipments | NAND is storage memory that keeps data when powered off (SSD, USB). ASP is average selling price. Bit shipments are volume. Revenue is roughly ASP times volume. |
| NBM (New Business Models) | SanDisk's multi-year supply contracts. A device to reduce memory cyclicality via minimum-revenue lock-ins and financial guarantees ($42 billion locked in). |
| HBF (High-Bandwidth Flash) | A new category SanDisk leads. Ultra-fast flash for AI inference, effectively the NAND version of HBM. |
| ADR (American Depositary Receipt) | A receipt that lets a foreign (Korean) stock trade in dollars in the US. Hynix lists on Nasdaq at a record scale on 7/10. |
| HBM/DRAM vs. NAND | HBM/DRAM are the main engine of this supercycle (prices soaring). NAND is the "weak link," relatively more commoditized and more vulnerable to oversupply. |
| Passive rebalancing / ETF inclusion | When a new name joins an index-tracking ETF, existing holdings are mechanically trimmed to make room, or money spreads across more names. If the Hynix ADR joins US semiconductor ETFs, Micron and others may be slightly diluted. |
| Gravity event | A large listing/event that soaks up a sector's capital and attention, affecting even neighbors' supply-demand. How the Hynix ADR ripples across the whole memory group. |
Disclaimer
This piece is research collected and cross-checked from public reporting, statistics, and local data for informational purposes, and is not investment advice recommending the purchase, sale, or holding of any specific security. The companies mentioned (SanDisk, SK Hynix, Samsung Electronics, Micron, Western Digital, etc.) are analytical examples, and the author may hold some of them. The 7/2 close, YTD, and P/E may differ slightly on a data or reporting basis. Target prices, NBM lock-in amounts, NAND share, and scenario price zones and probabilities are subjective estimates based on research and reports as of the writing date (July 2, 2026) and change frequently. In particular, being a beta 5.04 name, it can move big against the direction too. All investing carries the risk of principal loss, and investment decisions and their consequences rest with the investor.
References
Today's market and cause of the drop
- Why Sandisk Stock Is Plummeting Today · Motley Fool (2026-07-01)
- Why SanDisk Is Plummeting Despite a $3,000 Price Target · Stocks Down Under
- SNDK Moved Down 9.91% on Jul 1 · TradingKey
- SanDisk Fell 10%, Management Says the NAND Cycle Is Different · TIKR
Supply overhang (Western Digital)
- Sandisk Prices Secondary Offering (WD stake, 5.82M shares @ $545) · SanDisk IR
- Sandisk Form S-3ASR / 424B7 (shelf registration, sales) · SEC EDGAR
Hynix ADR (competitive threat)
For existing material, this drew on the SanDisk analyst report (2026-05-09) for Q3 FY26 results, NBM, the $1,800 target, and beta 5.04, and the data was cross-checked against local SNDK daily data (250 bars, 7/2 close of $1,957.83).