On July 9 I put Micron on WATCH for a pullback entry of $900–940, rated SPECULATIVE, with an explicit note: "extended after the $250B U.S. investment headline — do not chase." MU closed $991.64 that day. Today it is $969.82 and has traded as low as $954.13 intraday as the SK Hynix listing pulls capital. That call is working — the "don't chase, wait for the dip" discipline is being rewarded in real time. This deep dive is the follow-through: the dip is arriving, and I now have the conviction to act on it rather than just watch.
South Korean memory giant SK Hynix began trading on the Nasdaq this morning under ticker SKHYV (switching to SKHY Tuesday). ADRs were priced at $149 and opened at $170, up ~14%, raising $26.5 billion — the largest foreign IPO in U.S. market history. SK Hynix controls roughly 60% of the high-bandwidth memory (HBM) market and counts Nvidia and Apple as customers. Chairman Chey Tae-won told CNBC that even after committing to double capacity in five years, "all my customers said… 'that's not enough, man, we need more.'"
The sell-side reaction has been almost entirely about positioning: Morgan Stanley's desk called the ADR supply a "case for more downside"; UBS warned "positioning remains susceptible to further unwind"; Mizuho's Jordan Klein asked whether the listing is "additive to Micron, SanDisk, Lam Research… or does it pull money out of those names?" Micron opened down ~3% and saw ~700,000 options contracts trade off the open (≈87% of a full day's volume) with implied-volatility rank at 92%.
Micron Technology (MU, Nasdaq, market cap ~$1.15 trillion) is one of only three companies — with Samsung (~38% DRAM share) and SK Hynix (~29%) — that together control >95% of global DRAM. Micron makes DRAM (76% of revenue), NAND flash storage (24%), and increasingly HBM, the stacked DRAM bolted next to AI accelerators. Memory was a sleepy, brutally cyclical commodity for decades. AI changed that: HBM requires >3x the wafers to deliver the same bit count as standard DRAM, so every generation of HBM structurally cannibalizes conventional DRAM supply — turning a commodity into a scarce strategic asset.
Micron reported fiscal Q3 (ended May 28) after the close on June 24. I read the earnings press release / 8-K Exhibit 99.1 and the prepared remarks. The headline numbers were spectacular, but the structural disclosure matters more.
| Metric | Q3 FY26 | YoY |
|---|---|---|
| Revenue | $41.46B | +346% |
| Non-GAAP EPS | $25.11 | +1,215% |
| Non-GAAP gross margin | 84.9% (record) | +45.9pp |
| Operating cash flow | $25.39B | — |
| DRAM ASP (sequential) | up ~low-60s% | — |
| NAND ASP (sequential) | up ~mid-80s% | — |
Revenue beat consensus by $5.64B; EPS beat by $4.40. But guidance is what earnings should be judged on, and the Q4 guide was $50.0B revenue (vs. $43.45B consensus — a $6.55B beat), non-GAAP EPS $31.00 (vs $25.43), gross margin ~86%, and adjusted free cash flow projected >$30B in a single quarter. Shares spiked 14.6% after-hours to $1,199.52 on the print.
In the Q&A and CEO Sanjay Mehrotra's prepared remarks, Micron disclosed 16 multi-year Strategic Customer Agreements (SCAs) — take-or-pay contracts with data-center, consumer and auto customers, covering minimum committed volumes through calendar 2030, with cumulative minimum revenue of ~$100 billion and ~$22B of customer cash deposits/commitments (~$18B cash, ~$4B letters of credit). Management expects, when all SCAs are signed, ~half or more of total revenue under contract, with ~40% at fixed prices or ceilings near current levels and floor prices that "sustain gross margins well above any prior peak in Micron's history."
Supply context confirms it: Mehrotra said Micron has no "line of sight as to when memory supply will catch up with demand," and now expects tightness in both DRAM and NAND to persist beyond calendar 2027. HBM3E and HBM4 are fully booked through 2027 with demand into 2028; HBM4 is ramping at 2x the pace of HBM3E. Independent corroboration: TrendForce sees DRAM contract prices +13–18% in Q3 2026 (still rising, though decelerating); DigiTimes reports HBM4 prices set to roughly double by 2027; multiple houses see no meaningful price relief until 2028.
On July 9, Micron also raised its planned U.S. investment to >$250 billion through 2035 and poured first concrete at its Clay, NY fab ahead of schedule (backed by a $6.1B CHIPS award) — reinforcing long-term supply security and political goodwill.
At $970, MU trades at a trailing P/E of ~22.8x (TTM EPS $44.17). But trailing is meaningless for a company in a vertical earnings ramp. Consider the forward math:
So the tension is stark: a company guiding 86% gross margins, >$30B quarterly FCF and ~$100B of contracted revenue trades at ~8–10x forward earnings. That is either (a) an extraordinary bargain if the cycle has structurally lengthened, or (b) a classic peak-earnings trap where the low multiple is the warning. The SCAs are the evidence that tilts me toward (a) — but I respect (b) enough to size and stage the entry.
Sell-side has largely moved to (a): post-Q3 targets are Barclays $2,000, TD Cowen $1,500, Cantor $1,500, RBC $1,200; consensus ~$1,247 (29 analysts, Buy). Goldman raised its target but stayed Neutral — the honest skeptic's position (great fundamentals, but a lot already priced after +250% YTD).
Micron is the #3 DRAM player but the U.S. pure-play, and it has closed much of the HBM gap: HBM4 ramping fast, >$1B HBM4 already shipped, and it is the only leading-edge memory maker building material U.S. capacity (CHIPS-advantaged, politically insulated from tariff/China risk). SK Hynix is the higher-quality HBM franchise (~60% share) and now directly investable via SKHY; Samsung is the scale leader but has lagged on HBM qualification. The three-player oligopoly with >95% share and visible capital discipline is the moat — nobody is racing to flood the market, because greenfield fabs take years and each HBM generation eats more wafers.
MU ran from a $103 low (Aug 2025) to a $1,255 ATH on June 25, spiked to $1,199 after-hours on the Q3 print, and has since digested down to ~$970 (−23% from ATH). The $862 area (the prior Fibonacci-extension breakout / pre-earnings consolidation shelf) is the key structural support; the post-earnings gap and 50-DMA sit in the ~$900–955 band — which is exactly where the SK Hynix flow is pushing it. A weekly close back below ~$840 would tell me the SCA/structural re-rate is being rejected as "just another cycle top."
HIGH CONVICTION BUY MU — scale in on SK-Hynix-driven weakness.
My base case: ~$1,300 (≈11–12x a conservative ~$110 FY27 EPS — a modest re-rate for contracted durability). Bull ~$1,700 (cycle extends to 2028, SCAs cover half of revenue, 13–14x); bear ~$800 (DRAM rolls over in 1H27, multiple stays ~7x on peak earnings).
| Scenario | Prob. | Trigger | MU path |
|---|---|---|---|
| Bull | 35% | DRAM/HBM prices keep climbing into 2027; more SCAs signed; Fed holds. FY27 EPS $120+ | Retest $1,255, new highs toward $1,500–1,700 |
| Base | 45% | Prices decelerate but stay elevated (TrendForce +13–18% Q3 then flatten); SCAs hold margins. Choppy digestion | Range $900–1,300; grinds to $1,200 in 1–3 mo |
| Bear | 20% | SOX -17% drawdown + DRAM contract rollover in 1H27; peak-earnings de-rate | $840 breaks → $700–800 |
SKHYV/SKHY) — WATCH, don't chase day-one. The best pure HBM franchise (~60% share), but fresh IPOs run hot then cool: SpaceX (SPCX) fell ~26% from its June peak and below its debut price within weeks. Let it establish a base after the +14% open; revisit once the ADR/Korean-share premium normalizes.SNDK, $1,874, +0.9%) — the NAND-leverage play; NAND ASPs +mid-80s% sequentially at Micron. Extended after a massive run; I prefer MU's DRAM/HBM mix and contracted revenue.Everyone is trading the SK Hynix listing as a reason to sell memory. I think it's the opposite. The listing adds shares, not wafers, and it drops MU into a buyable zone just weeks after a quarter that structurally de-risked the entire memory cycle via $100B of take-or-pay contracts. At ~8–10x forward earnings with 86% gross margins and >$30B quarterly FCF, the risk/reward favors buyers — provided you stage the entry to survive a stretched-SOX air-pocket. Buy MU $900–955, targets $1,200 then $1,500, invalidation weekly close <$840 or a DRAM-price rollover.
Sources (primary, linked): Micron Q3 FY2026 earnings press release & prepared remarks (investors.micron.com); CNBC — SK Hynix opens at $170; CNBC — What the SK Hynix listing means for U.S. memory stocks; CNBC — Kilburg MU options trade; CNBC — BTIG SOX warning; CNBC — Micron $250B U.S. investment. This is analysis and opinion, not personal investment advice. I/we may hold positions in names mentioned. Prices intraday July 10, 2026.