MU Crashed 22% After Record Earnings. Here's Who's Actually Selling

TL;DR

• Micron beat fiscal Q3 EPS by 18.6%, posted record revenue, and raised guidance on AI memory demand. Then the stock gave back 22% from its post-earnings highs.
• Nothing broke. HBM is still sold out. The guide is still up.
• What changed: SK Hynix listed on the Nasdaq and gave AI-memory money a second home, Korean memory profit-taking spilled into Boise, and the whole chip complex threw a $1 trillion valuation tantrum in the same week.
• $900 is the line. Above it, this is a shakeout. Below it, the memory supercycle gets its first real stress test.

Best Earnings In Company History, Straight Into A Woodchipper

Micron printed record fiscal Q3 numbers on June 24. EPS beat by nearly 19%. Revenue record. Guidance raised specifically because HBM demand is running ahead of what they can physically make. The stock ripped to fresh highs and everyone on the timeline was a genius for about four trading days.

Then it dropped 22%.

Micron's post-earnings pop and the round trip that followed, week of July 6 to 10.

If you panic-sold that move thinking something was wrong with the business, you got played by the oldest trick in the book. The print was clean. The selling came from somewhere else entirely.

Who's Actually Selling

Everyone who wanted in was already in. MU ran hard into earnings and harder after. When a beat that big buys you only four green days, the marginal buyer is gone, and price falls under its own weight from there. Classic sell-the-news, textbook edition.

SK Hynix stole the trade. On July 10, SK Hynix raised $26.5 billion on the Nasdaq, the biggest foreign IPO in US history, and popped 13% on day one. For two years, MU was the only liquid US ticket to the HBM boom. Monopolies on a hot trade earn a premium. That monopoly ended this week, and a chunk of the premium walked across the street to ticker SKHY.

Korea sneezed, Boise caught the cold. Profit-taking ripped through South Korean memory names first and spilled straight into MU, because the market prices the entire memory complex as one trade. They rise together, they fall together, and this week they fell.

The AI valuation tantrum. Intel lost 21% in a week. The chip complex shed over a trillion dollars of market cap. Strategists dusted off their June 2000 charts. When the whole sector is getting repriced, the best house on a burning block still smells like smoke.

The $900 Line

• MU closed Friday at $979 after tagging the mid-$950s.
• $900 is the level the whole street is watching. It held on the first test this week.
• Hold it, and this was a positioning flush that gets bought aggressively into the next guide.
• Lose it, and the tourists who bought the top at $1,255 start capitulating. That gets ugly before it gets better.

The Options Angle

That 22% drawdown pumped implied volatility, which means calls are now priced for the panic you just watched. Don't pay panic prices for upside. Sell them instead.

• Sell cash-secured puts at the $850 to $880 strikes, 30 to 45 days out. Get assigned, and you own the best US-listed memory franchise 30% off its highs with a freshly raised guide. Don't get assigned, and you pocketed fear premium from people hedging a bottom that already happened.
• Want direct upside? Wait for a weekly close back above $900, then buy September at-the-money calls. Two months is enough time for the SKHY novelty to wear off and for the next earnings print to remind everyone the HBM book is sold out through next year.
• Skip the weeklies. Positioning resets take weeks to resolve. Paying theta on a 5-day expiry here is donating to your market maker's boat fund.

Sooo... Buy Or Sell?

Buy, with a plan. The company just told you demand exceeds supply and raised its own targets. The stock fell anyway, on flows, a shiny new listing, and sector-wide fear. That gap between the business and the tape is the entire opportunity. Above $900 I'm selling puts and buying dips. Below $900 I sit on my hands and let the last tourists finish leaving before stepping in bigger.