The week of July 21 to 25 is the single most instructive week in the memory supercycle so far. Not because of the Tuesday gains. Because of what happened on Friday. Understanding both moves together tells you more about how to trade this sector than any analyst note written this year.
Here is the full sequence. Memory stocks fell 25% across July to their lows. Monday July 20 saw a 4 to 6% snap-back as no fresh Korea headlines appeared over the weekend. Then Tuesday exploded: SNDK surged 14.27%, SKHY ADR jumped 13.75%, MU climbed 12.17%, WDC was up 12%, the DRAM ETF gained 11%. The catalyst was a Morgan Stanley report forecasting a 25% memory price increase from Q2 to Q3, with the firm explicitly stating it was buying the dip. BofA called the slump a summer reset. Semiconductor ETFs absorbed over $2.1 billion in a single session.
Then Friday happened. SNDK dropped 9% to $1,467. MU fell 6% to $929. SKHY fell 6% to $158. WDC dropped 6%. The DRAM ETF sank 7%. A Korea-led selloff with no fresh fundamental trigger. No earnings miss. No demand signal.
That is the environment you are operating in. And next week, SK Hynix reports July 29. That is the binary that resolves the question this entire week raised.
1. Why Friday Had No Fundamental Trigger
Micron's fiscal Q3 revenue of $41.46 billion, up 346% year on year, with non-GAAP EPS of $25.11 against a $20.28 consensus and Q4 guidance of $50 billion, were all already public before Friday opened. SanDisk's most recent quarter posted revenue of $5.95 billion with EPS of $23.41. None of those numbers explain Friday's drop.
What explains it is structure, not fundamentals. Three forces converged.
SanDisk carries a trailing P/E of 55x, which makes it the first target for institutional profit-taking when the tape softens. After a 14% single-session gain on Tuesday, mechanical risk management forces trimming. That is not bearish conviction. It is portfolio housekeeping.
The Korea correlation has become structural. Every time the KOSPI weakens, US-listed memory names trade the relationship. BOK rate hike risk, the KIS note overhang, and the thin float on SKHY ADRs create persistent cross-market volatility with no fundamental read-through on the US names.
MULL and MUU are amplifying every move in both directions. A 6% drop in MU produces approximately a 12% drop in these daily-reset products. The forced rebalancing at close creates mechanical selling pressure that amplifies the underlying move. On days when MU is down 6%, MULL holders are down 12% and margin calls force further MU selling. The tail is wagging the dog.
2. The MULL and MUU Question
MULL is the GraniteShares 2x Long MU Daily ETF. MUU is Direxion's equivalent. Both are daily-reset instruments. MUU's trailing 12-month return is approximately 3,936%, a number that was earned during a near-linear bull trend. That run is real. The current tape is not a linear trend. It is 14% up on Tuesday and 9% down on Friday, which is exactly the choppy environment where NAV decay destroys leveraged ETF holders even when the directional thesis is correct.
Over a week where MU drops 6% and then recovers 7%, you do not break even in MULL. The daily reset and NAV decay mean your net position is worse than holding MU directly. This effect compounds over weeks, especially in volatile sideways conditions.
The appropriate use of MULL and MUU right now is as a defined-catalyst trade. If SK Hynix beats on July 29 and HBM4 scaling is confirmed, MU trades 8 to 12% higher in that session. MULL gains 16 to 24% in the same day. That is a precise entry and exit around a single known event, which is what these products are designed for. It is not a thesis hold.
If you hold MULL currently and your thesis is the 12-month memory supercycle, switching to MU directly preserves the thesis while eliminating compounding decay. The thesis survives. The instrument may not.
3. Morgan Stanley's 25% Call: What It Actually Prices In
Morgan Stanley's Joseph Moore said the selloff created a strong entry point and expects memory and NAND prices to rise 25% or more from Q2 to Q3. That is not a minor upgrade. It is a structural statement that the pricing power seen in Micron's $41.5 billion quarter is accelerating into Q3, not peaking.
The framing one investor used this week is worth noting: SNDK's Q3 capex was just 1.4% of revenue due to its Kioxia partnership. That makes it an asset-light AI storage infrastructure business, not a commodity cyclical. At 6.4x 2027 EPS and 16.2x forward EV/EBITDA, the re-rating argument is credible if the memory price trajectory confirms. Whether the market accepts that reframing will be determined largely by what SK Hynix says on July 29.
4. Next Week's Projection
Monday July 28 sets the tone. If the weekend produces no fresh negative Korea headlines and no new Iran escalation, the memory group holds in the $1,400 to $1,500 range for SNDK and $900 to $950 for MU. The fundamental floor is solid. The sentiment floor is fragile.
Tuesday through Wednesday will be positioning days. Expect lower volume as traders reduce or build exposure ahead of the July 29 binary. SKHY will be the most actively traded name in the complex as earnings approach.
July 29 is the event that matters. SK Hynix's Q2 earnings answer the question the KIS note raised three weeks ago: did HBM fixed contracts cap blended ASP enough to produce another consensus miss despite a record profit quarter? Consensus Q2 operating profit sits around 13 to 14 trillion won. If SK Hynix confirms HBM4 scaled in Q2 and blended ASP improved, the KIS concern is resolved and the entire complex re-rates. If HBM4 scaling was still delayed and ASP undershot again, the 14% Tuesday gain gives way to another 10 to 12% sector-wide selloff.
Four specific things to watch on July 29: Q2 operating profit versus 13 to 14 trillion won consensus, HBM4 shipment volume in Q2, Q3 ASP trajectory commentary given Morgan Stanley's 25% forecast, and whether CEO Kwak Noh-jung reiterates that 2027 will be the worst supply year in the industry's history. Each of those either confirms or breaks the thesis that the market is currently holding at elevated multiples.
5. The Three-Name Question: MU, SKHY, or SNDK
MU at $929 is the cleanest trade in the complex right now. Forward PE of just 6x against analyst targets near $1,507, implying 62% upside. Zero direct BOK rate exposure. Eighth consecutive EPS beat. $50 billion Q4 guidance. The 6x forward PE on a company with 85% gross margins and a sold-out order book is the most straightforward valuation argument in the entire memory sector. Korea selloffs punish MU for structural reasons that have nothing to do with its fundamentals. That is the entry, not the exit.
SKHY at $154 to $158 is the highest-leverage play on the HBM supercycle and the most direct binary on July 29. 58% global HBM market share, 60 to 70% of Vera Rubin HBM4 volume estimated, HBM4E samples delivered ahead of schedule. The structural bull case is the strongest of the three names. The near-term risk is another KIS-style earnings miss where record profits still fall short of elevated expectations. Positioning into July 29 on SKHY is positioning for a binary, not a thesis hold.
SNDK at $1,433 to $1,467 has the most complex setup. The 55x trailing PE is the highest in the group and makes it the most volatile in both directions, as Tuesday's +14% and Friday's -9% confirm. The Kioxia partnership and 1.4% capex-to-revenue ratio are genuine structural advantages that a higher multiple can eventually justify. For long-term conviction, SNDK's margin structure supports a position. For a tactical trade around July 29, MU offers better risk-adjusted exposure.
The memory supercycle is intact. The DRAM supply-demand gap is intensifying into 2027. The volatility is not the thesis breaking. It is the price you pay to own the best-positioned sector in AI infrastructure at multiples that are dramatically lower than the broader market.
July 29 is the next answer. Between now and then, size for the binary.
I am not a financial advisor. Trade wisely, Comrades.
Comments