On June 24, Micron reported the best quarter in the history of the memory industry. Revenue of $41.5 billion, up 346% year over year. Earnings of $25.11 per share against expectations near $20.86. A record 85% gross margin. Guidance for the October quarter of $50 billion, a full $6.5 billion above consensus. The stock gapped up 16% the next day and touched an all-time high of $1,255.
Then it fell for seven of the next eight sessions. By Tuesday's close MU sat at $938, down 25% from the high, with roughly $350 billion of market cap erased in twelve calendar days.
I own the stock, and I spent the last day pulling this apart properly: every catalyst, the analyst tape, the technicals, the flows, and the SK Hynix angle everyone keeps whispering about. Here is what I found.
Nothing in the numbers broke
The selloff was not a reaction to any Micron datapoint. It was a stack of sentiment events landing on the most crowded trade of 2026, almost one per day.
June 23: a South Korean regulator's warning about leveraged Samsung and SK Hynix ETFs triggered a global chip rout and a 13% down day, before earnings even hit. June 25: the day of the all-time high, a DRAM price-fixing class action was filed against Micron, Samsung, and SK Hynix in the Northern District of California. June 26: profit-taking on the heaviest volume of the stretch, plus disclosure of the CEO's $46 million share sale under a pre-set trading plan. July 1: the worst single day, down 10.6%. July 2: Michael Burry disclosed a short from $1,051.87, calling Micron "a destroyer of capital" in a "psychological bubble." July 7: Samsung posted a record $58 billion preliminary operating profit that beat estimates by only 6%, and the whole memory complex sold the news.
Notice what is not on that list: an order cut, a pricing miss, a guidance walk-back, a demand warning. There isn't one.
The SK Hynix listing is real, and it's Friday
If you've heard that the market is waiting for SK Hynix to list in America, that's not a rumor. The board approved it June 24. The F-1 is filed with the SEC. The ADRs price Thursday July 9 and begin trading on Nasdaq Friday July 10 under the ticker SKHY, raising up to $29.65 billion. That is the largest first-time US share sale by a foreign company in history, bigger than Alibaba in 2014. Baillie Gifford and Coatue have reportedly indicated up to $7 billion of combined interest.
Two theses compete. The drag thesis says $29 billion of fresh AI-memory paper absorbs institutional demand, hands US investors the HBM market leader at a cheaper multiple, and kills Micron's scarcity premium as the only US-listed pure memory play. The lift thesis says the listing removes the Korea discount, raises the whole sector's multiple ceiling, and pulls global capital into memory as an asset class.
Here is the tell that it's not a simple rotation: SK Hynix fell harder than Micron during this selloff, down about 26% from its peak versus Micron's 25%. If money were leaving MU to wait for SKHY, SK Hynix would have outperformed. It didn't. Both got de-rated together. Friday's debut resolves the question either way: a strong open that holds says capital is flowing into the complex; a fade says the marginal buyer of AI-memory exposure is saturated.
Cheap and expensive at the same time
At $938, Micron trades at 21x trailing earnings, about 13x fiscal 2026 estimates, and about 6x fiscal 2027 consensus of $150 per share. On forward earnings it is one of the cheapest trillion-dollar companies ever printed. On price-to-book it trades at 10.5x, versus low single digits at every prior memory cycle peak. Both facts are true, and the entire debate collapses to one question: are the 2027 estimates real?
Consensus fiscal 2027 EPS was about $20 last December. It is $150 today. Estimates went up seven-fold in seven months. If they hold, the stock is cheap at almost any reasonable multiple. If 2027 HBM contract negotiations disappoint or hyperscaler capex pauses, the estimates deflate underneath the price. That is the actual risk, not this week's tape.
The counterweight is contractual, and it is new to this cycle: 16 non-cancelable take-or-pay agreements running through 2030, roughly $100 billion of minimum revenue, backed by $22 billion of customer deposits and letters of credit, with price floors set above the best gross margin of any prior cycle. No memory downturn in history started with the customers legally locked in.
What Wall Street says
Forty-five analysts cover the stock: 31 Strong Buy, 9 Buy, 5 Hold, zero Sells. Average target $1,486, median $1,550. The high is Melius at $2,200, with Cantor, Susquehanna, Barclays, and DA Davidson all at $2,000. The lone major-bank holdout is Goldman at $1,100, and even that is a Hold on valuation discipline, not a demand call. The loud bears are not sell-side: Burry, positioning desks warning about crowding, and the uncomfortable fact that insiders are selling at the fastest rate since 2010.
What the tape says
Price broke the 8-day and 20-day moving averages but still sits about 9% above the 50-day near $862 and more than 100% above the 200-day near $448. Daily RSI reads 49: neutral, not oversold, which after a 25% drawdown tells you how stretched the run-up was. The volume pattern is the uncomfortable part: every major down day printed above-average volume while the only bounce came 35% below average. That is a distribution-shaped tape short term.
But short interest is just 3.3% of float with about a day to cover. Nobody is pressing a big short except Burry. This is longs lightening up, not shorts attacking, and that kind of selling can reverse as fast as it started once the sellers are done.
The levels that matter: $972 and $1,014 above (a close over $1,014 projects $1,090), $892 to $896 below (tested and held on Tuesday), then the zone I care about most, $860 to $900, where the 50-day meets the biggest volume shelf. A high-volume weekly close below $860 breaks the 2026 uptrend structure and puts the 100-day near $706 in play.
Shakeout or top?
My base case, roughly 55%: this is a violent positioning flush inside an intact supercycle. The decline is sector-wide and event-driven, the Street's numbers are still rising, the demand side is contractually documented like no memory cycle before it, and the biggest overhang resolves this Friday. I expect chop between $860 and $1,050 that resolves higher as the October quarter confirms the $50 billion guide and 2027 HBM pricing lands. At a historically normal 10x fiscal 2027 earnings, that is a $1,500 stock.
The bull case, call it 25%: SKHY prices strong, index money and the Korea-discount re-rate lift the complex, and the December 9 CHIPS unlock (after which management returns 100% of excess free cash flow, with BofA modeling $31.7 billion of buybacks next fiscal year) puts a permanent bid under the stock. The $2,000+ targets come into view through 2027.
The bear case, 20%, and I take it seriously: this was the top. The SKHY debut fades, Samsung's full report on July 23 shows HBM pricing pressure, the $575 billion Korean capex wave plus China's CXMT normalize DRAM through 2027, estimates stop rising and then get cut, and the multiple compresses the way it always has late-cycle. The insider selling and the neutral RSI were the tells.
What would flip me: fiscal 2027 revisions turning negative, reported HBM contract pricing coming in flat instead of higher, a hyperscaler capex pause, or that weekly close below $860 on expanding volume.
Watch Friday's SKHY debut, Samsung on July 23, TrendForce's monthly pricing data, and Micron's own report in late September. The market just handed everyone exact dates.
Disclosure: I am long MU. This is my own research and opinion, not financial advice. Do your own work, especially with a 2-beta trillion-dollar memory stock.