The call: BUY Micron (MU) on the Taiwan-strike dip, $980–$1,080 zone — SPECULATIVE, toward $1,536 consensus then $2,100–$3,000 on the buyback/HBM path, 1–3 months / 6–12 months.
Why now: Record FY2026 ($133B revenue), HBM sold out through 2027, CHIPS buyback locks come off Dec 9 — and the stock still trades ~6x FY27 earnings.
The disagreement: Consensus sees a 266%-up cyclical top; Luria sees contracted, non-cancelable revenue that deserves a 19x multiple. I side with Luria on the cash flows, not the multiple.
The level that changes everything: A daily close below $980 kills the trade — it means the cycle, not the contract story, is winning.
Two headlines collided this morning, and that collision is the trade. D.A. Davidson's Gil Luria raised Micron's 12-month target to $3,000 from $2,100 — implying ~187% upside from Tuesday's $1,045.56 close — while a Taiwan union representing ~2,000 Taoyuan workers won authorization to strike over bonuses. The stock shrugged: MU traded to $1,074.52 (+2.8%) intraday Wednesday after dipping premarket.
The strike's absurdity is the signal. Micron already offered Taiwan workers 35–68 months of pay in fiscal 2026 rewards; the union demands ~83 months. When labor fights over how many years of salary the bonus should be, you are not early-cycle — you are at the top of the most profitable memory year in history. FY2026 revenue hit a record $133.19 billion, Q4 alone did $54.2 billion (+379% y/y) with EPS of $33.42 vs. $32.56 expected. That is the context for everything below.
Why Micron over the macro noise (record S&P at 7,787 with a $44M SPY put spread betting on a 35% crash, 10Y at 5.36%, Brent at $102 on Hormuz threats, BTC sliding under $83k)? Because Micron compresses all of it — AI capex demand, supply-chain fragility, and the multiple-re-rating question — into one actionable name. Yesterday's Constellation/Google nuclear piece covered the power side; today is the memory side of the same AI bottleneck.
Luria's argument is simple enough to check: MU earns ~$157/share in FY27 (implied by 6x at ~$1,016) and deserves 19x, hence $3,000. The bridge from 6x to 19x rests on three claims, each worth interrogating:
First, demand outstrips supply through 2028. Management guided on the Sept 30 call that calendar 2027–2028 will be tighter than 2026, with 2026–2027 HBM supply fully sold out. Third-party data supports it: DRAM contract prices jumped 50%+ quarter-on-quarter entering 2026, Bernstein sees ~20% further upside this quarter, and Citi models HBM bit demand +62% in 2027 and +69% in 2028. With only three global DRAM suppliers (Micron, Samsung, SK Hynix) and capex diverted from NAND to HBM, the oligopoly can hold price — my opinion: this is the strongest leg, and the "50% of revenue under long-term strategic agreements that aren't easily cancelable" detail is what converts a spot-price story into a contracted-earnings story.
Second, the multiple deserves to triple because earnings quality changed. Non-cancelable contracts plus $68B+ net cash turn a boom-bust commodity into something closer to an infrastructure annuity. Luria's "if you don't buy it, they will" is a hint: strategic buyers (hyperscalers vertically integrating memory, à la SpaceX reportedly raising $40B in debt for Nvidia chips) could pay 19x without blinking. My opinion: 19x is a takeover multiple, not a public-market through-cycle multiple — fair as a stretch target, aggressive as a 12-month base. The honest base is 10–12x on $160–$190 FY27 EPS = $1,600–$2,280, which still doubles the stock.
Third, the December 9 buyback unlock is a forced buyer. CHIPS Act direct-funding terms capped repurchases; those fall away Dec 9, 2026. Bulls sketch $50–60B of capacity against $68B net cash. Retiring even 5–8% of shares outstanding at 6x earnings is violently accretive. My opinion: this is the most underpriced near-term catalyst — a mechanical bid arriving in eight weeks that no model of "cyclical top" includes.
On forward math MU is the cheapest AI hardware name in the S&P; on cycle history it is the classic top. Both are true:
| Metric | Level (Oct 7, ~12pm ET) | What it means |
|---|---|---|
| Price / move | $1,074.52, +2.8%; +266% YTD, +463% 1-yr; 52-wk $179.61–$1,255 | 14% below June $1,255 high — a breakout retest, not a breakdown |
| Forward P/E | ~6x FY27 EPS (~$157–188); TTM P/E ~14x on $74 EPS | Cheapest vs. NVDA ~30x, AVGO ~28x, AMD ~35x — the re-rating gap is the trade |
| Growth | Q4 rev +379% y/y; FY rev $133B record; 46 of 50 analysts Buy | Peak-growth prints mark tops as often as they confirm trends |
| Balance sheet | $68B net cash; $100B+ FCF chatter; div yield ~0.6% | Funds the Dec 9 buyback without touching capex |
| Comps today | SNDK $1,711 +3.1%; STX $804 flat; WDC $405 -1.4%; AMAT $519 -2.1%; LRCX $326 -2.2% | Memory (MU/SNDK) outperforming equipment — pricing power over units |
The Seeking Alpha "peak cycle" Hold note out this morning is the honest bear in one line: stellar results + sold-out HBM is exactly what prior tops looked like. Memory veterans remember 2018 and 2021 — sold-out quarters preceded 40–60% drawdowns when hyperscaler digestion hit. The difference this time, and it matters: contracted revenue share and HBM's 2–3 year qualification moat. Standard DRAM could roll while HBM holds — and HBM is now the margin engine.
Four risks, each with a mechanism and a number:
Flagship: BUY the strike-fear dip; do not chase the $3,000 headline. The edge is the Dec 9 buyback bid plus sold-out HBM against a 6x multiple — a 1–3 month mechanical re-rating toward consensus $1,536, with a 6–12 month path to $2,100+ if contracts hold.
| Field | Level |
|---|---|
| Entry zone | $980–$1,080 (strike-fear dip; add washout $940–$960 only if no walkout date) |
| Target | $1,536 (Street consensus) base; stretch $2,100 (prior Luria high) / $3,000 (new bull, needs 19x + buyback proof) |
| Invalidation | Daily close below $980 — one level, measurable, outside the base case |
| Timeframe | 1–3 months (buyback unlock Dec 9) / 6–12 months (HBM contract repricing) |
| Conviction | SPECULATIVE (one confirmed signal — contracted scarcity; unconfirmed — multiple re-rating) |
| Audience / size | Half-size vs. HIGH-conviction book; pair against existing EQ-MU-1 washout add ($980–$1,000) — no double-count |
Bull (25%): HBM 2027 pricing set up 20%+, Dec buyback $30B+ announced, 12x $190 = $2,280 by mid-2027. Base (50%): contracts hold, spot flattens, 8–10x $160–170 = $1,350–$1,650 into year-end — the $1,536 consensus is the magnet. Bear (25%): hyperscaler digestion + Taiwan stoppage + 10Y >5.5% → $90 EPS × 8x = $720.
What would prove me wrong: (1) daily close below $980; (2) a Taiwan walkout lasting >7 days; (3) December buyback authorization under $10B; (4) any cut to "tighter in 2027–2028" language on the next call; (5) DRAM spot down >15% in a quarter while HBM pricing flat — the mix shield failing. Any two and the thesis, not just the trade, is wrong.
Second-order notes (one line each): SNDK is the cleaner NAND pure-play if MU feels too consensus; AMAT/LRCX weakness today is the tell that equipment is not participating — buy memory pricing, not tool orders; the 10Y >5.36% + Brent $102 combo is why this stays SPECULATIVE not HIGH; BTC under $83k on the same macro confirms risk-off breadth, not a crypto-specific break.
Open-call check: EQ-MU-1 (add washout $980–$1,000, tgt $1,200, kill <$980) is live and consistent — today's $980–$1,080 entry overlaps it; I am not issuing a second independent MU call, this deep dive upgrades its target path ($1,200 → $1,536 base) while keeping the same $980 kill. No contradiction with EQ-SNDK-1 or TACT-SPX-FADE-1 (a market flush takes MU with it — that is what the $980 kill is for).
Contents: A. Data snapshot · B. Model & assumptions · C. Sources
MU $1,074.52 (+2.77%) intraday Oct 7, 2026 ~16:00 UTC (prev close $1,045.56 Oct 6); 52-wk $179.61–$1,255. SNDK $1,711.43 +3.07%; STX $803.81 -0.23%; WDC $405.35 -1.38%; AMAT $519.38 -2.05%; LRCX $326.39 -2.25%; NVDA $237.24 -0.84%; AMD $644.74 -0.72%; AVGO $374.73 -0.29%; MRVL $283.14 -1.35%. SPY $776.39 -0.35%; QQQ $756.41 -0.43%; TLT $76.93 -0.45%. BTC ~$83,365 -2.8%; ETH ~$2,569 -5.0%; fear/greed 71 (Greed). 10Y 5.36%, 30Y 5.73%, Brent ~$102, WTI ~$91 (intraday Oct 7). MU Q4 FY26 (Sept 30): rev $54.229B vs. ~$52.6B est., EPS $33.42 vs. $32.56 est.; FY26 rev $133.19B record. DA Davidson: $3,000 from $2,100, Buy, ~19x FY27 EPS. 46/50 Buy per LSEG. HBM sold out 2026–2027. CHIPS buyback limits end Dec 9, 2026. Taiwan: 1,994/2,012 strike authorization; union seeks ~83 months bonus vs. offered 35–68. CEO sales Aug 21 $970–$990. Dates verified: Oct 7, 2026 is a Wednesday; Dec 9, 2026 is a Wednesday.
FY27 EPS bridge: $133B FY26 rev × ~15% growth (price > volume) ≈ $153B × ~32% net margin ≈ $49B net ÷ ~290M fully-diluted shares ≈ $165–170/share (range $157–188 across Street). Multiples: bear 8x trough ($720–$1,350), base 10x ($1,600–$1,700), bull 12–14x ($2,000–$2,400), Luria 19x ($3,000). Load-bearing assumption: non-cancelable contracts actually hold through a digestion quarter. If they reprice or push out, EPS falls to ~$90 and no multiple above 10x applies — answer becomes $720–$900, i.e., the bear case is the model, not a overlay.
CNBC — Micron to nearly triple, D.A. Davidson $3,000
Morningstar/MarketWatch — 19x FY27 math, Dec buyback
Cointelegraph — BTC $82.7k, 10Y 5.36%, Brent $102
CNBC — record S&P, $44M SPY put spread, Meta large call trade
Micron Q4 FY26 prepared remarks (PDF)
Seeking Alpha — Taiwan strike authorization detail
CNBC Investing Club — post-Q4 target raise