Calls › MU › Oct 15, 2026

MU Lean bullish Micron Technology, Inc. · Technology

Peak-multiple compression, not peak earnings — yet.

Published Oct 15, 2026 · Horizon 3-6 months Audited

Edge6/10

Evidence7/10

Entry (close)

$914.13 2026-10-14

Since publish

+1.6% to 2026-10-27

vs SPY

+1.6%

3-month verdict

Closes 2027-01-13

Thesis

MU trades at ~5x FY27 consensus EPS ($176) after a 25% drawdown driven mostly by sector-wide AI-trade de-risking (SOXX -11% same week) rather than company-specific deterioration. Demand visibility is unusually concrete — >75% of FY27 shipments committed, HBM largely sold out at prices above 2026 levels, management guiding supply-demand tighter through FY28. If DRAM/HBM pricing merely decelerates rather than rolls over, earnings revisions keep rising into a stock priced for decline, producing re-rating toward the $1,100-1,250 range. The bear case (Morningstar $700) requires a 2028-29 downturn that the stock is already partially pre-paying.

Bull case

Five times forward earnings for a company whose next three quarters are contractually underwritten, with estimates still rising 28-to-2 and supply guided tighter through FY28. The selloff is ~two-thirds sector beta (SOXX -11%) plus a weakly-sourced China scare; CXMT capacity cannot affect HBM/advanced DRAM before 2028 at earliest. If Q4 pricing confirms +~20% and Q1 FY27 (reporting ~December) beats the $61.5B guide as Q4 did, the multiple re-rates toward even 7-8x, implying $1,230-1,410 (+35-55%).

Bear case

This is the classic cyclical trap: earnings peaks are marked by maximum estimate optimism, minimum multiples, and maximum capex — all present. Micron is spending ~$55B to build capacity that lands precisely when Morningstar models the downturn (2028-29), with depreciation that will compress margins through the trough; CXMT introduces a structurally new low-cost supplier into the next downcycle; and if AI-capex ROI skepticism (the actual driver of the October selloff) causes any hyperscaler order pause, the 25% of FY27 volume NOT contractually committed reprices instantly. Morningstar's $700 fair value is 23% below current price — and Morningstar-type normalized valuations have historically been right on memory.

What would change our mind

Any sequential decline in DRAM contract prices in Q4 2026 or Q1 2027 (not deceleration — decline)
Q1 FY27 revenue at or below the low end of guide ($60B) or gross margin guidance cut from ~86%
Evidence that Strategic Customer Agreement commitments are being renegotiated or pushed out
Verified CXMT HBM/advanced-DRAM qualification at a major customer before mid-2027
FY28 EPS consensus beginning sustained downward revisions while FY27 still rises — the market pricing the cliff

**MU's selloff exceeds the sector de-risking move.** Rebased over the recent drawdown, this shows whether Micron's 25% drop was mostly amplified sector-wide AI unwind versus a company-specific deterioration.

Since publication vs SPY · closed horizons are frozen

1 weekfinal

-0.5%

-0.1% raw · miss

1 monthrunning

+1.6%

+1.6% raw

3 monthsrunning

+1.6%

+1.6% raw

6 monthsrunning

+1.6%

+1.6% raw

The full research everything the view rests on

Direct answer to the question asked

MU's 25% drawdown is the market refusing to capitalize near-peak memory earnings, not evidence the cycle has turned: contract prices are still rising ~20% QoQ, >75% of FY27 shipments are contractually committed, and FY27 EPS estimates ($176) are still being revised up 28-to-2. At ~5x forward earnings, the price already discounts substantial normalization, making the risk/reward mildly asymmetric to the upside over 3-6 months — but the ~$55B FY27 capex plan and the CXMT China shock are genuine 2028-29 oversupply seeds, not noise. Lean bullish, with the next two DRAM pricing prints as the arbiter.

Where the edge would come from

What the price assumes

The price embeds peak-cycle skepticism: a ~5-7x forward P/E vs. a 10-13x historical normalized range means the market assumes FY27 earnings of ~$176 are materially above mid-cycle and applies a trough multiple to peak earnings. Implicitly, consensus pricing assumes meaningful EPS normalization beginning around 2028 — i.e., the market is paying roughly Morningstar's framework (peak early 2028, downturn 2029) while sell-side targets cluster $1,250-1,900.

Why the market may be wrong

The market may be conflating rate-of-change deceleration with absolute decline. Server DRAM price increases slowed to 13-18% QoQ but remain strongly positive; >75% of FY27 volume is contractually locked, muting spot-market sensitivity; and HBM's 3-4x wafer intensity tightens conventional DRAM even as competitors expand. If the cycle peak is 2028 rather than 2027, four more quarters of upward revisions into a 5x multiple is a poor short. The risk is the market is right early — this exact pattern (falling stock, rising estimates) is textbook late-cycle behavior.

The one thing to watch

The variable that actually resolves the debate is not earnings — those are locked upward for 2-3 quarters — but DRAM contract price *direction* versus Micron's own capex trajectory: the ~$55B FY27 capex (construction landing late-2028) means Micron itself is building the glut that ends this cycle, so every quarter of current pricing strength is simultaneously raising future supply. Watch whether Q4 contract price increases hold ~20% QoQ; if they do while the stock stays 25% off highs, this is pure multiple compression and the asymmetry favors longs; the first negative sequential print flips the thesis to Morningstar's framework regardless of how good reported earnings look.

Key evidence (10)

FY26 (ended Sept 3, 2026): revenue $133.2B vs $37.4B FY25; GAAP EPS $74.33; Q4 revenue $54.2B beat by 7.5%, gross margin ~87%, $44B operating cash flow [T1 SEC/IR]
Q1 FY27 guidance $61.5B±1.5B revenue, EPS $38.15±1.00 — well above prior consensus $57B/$35.14 [T1 IR]
FY27 consensus $275B revenue / $176.15 EPS; 28 EPS raises vs 2 cuts in last 30 days; FY28 $206.33 [T4 consensus aggregators]
>75% of FY27 shipments committed via Strategic Customer Agreements; agreements covering >35% of revenue extend through 2030; supply-demand guided tighter through FY28 [T1 transcript]
DRAM and NAND contract prices each +~20% QoQ in Q3 2026 with further Q4 increases expected; server DRAM decelerating to 13-18% QoQ but positive [T3/T4 industry reporting]
FY27 capex ~$25B in H1 alone (Q1 ~$11.5B), implying ~$55B+ full year, construction-heavy, capacity landing late calendar 2028+ [T4 IR-derived]
Morningstar cut fair value $850→$700, modeling peak ~early 2028 and sharp 2029 downturn [T3]
Stock -16% in one week, -25% from $1,254.81 high; SOXX -11% same week; Bloomberg Oct 13: AI-trade cracks + CXMT technology advance fears [T2]
Forward P/E ~5.2-7.0x vs 10-13x FY22-FY25 historical normalized range [T4]
20d realized vol at 6th percentile of trailing year; RSI 25.6 — compressed and oversold [computed]

Catalysts and risks

Catalysts

Fiscal Q1 FY27 earnings (~mid-to-late December 2026): beat vs $61.5B/$38.15 guide and FY27 capex update
Q4 2026 DRAM/NAND contract price settlements (Nov-Dec): continuation of ~20% QoQ or first deceleration below ~10%
Hyperscaler 2027 capex guidance (Jan-Feb 2027 earnings season): AI-capex ROI narrative resolution
Any CXMT capacity/technology verification — confirmation or debunking of the October shock
Analyst target revisions: whether the $1,250-1,900 cluster holds or follows Morningstar down

Risks

Cyclical peak risk: 5x forward P/Es on memory stocks have historically preceded 50%+ drawdowns when pricing turns (2018, 2022 analogs)
~$55B FY27 capex compresses near-term FCF and seeds 2028-29 oversupply with fixed depreciation
CXMT/China supply shock is weakly sourced but, if real, changes the long-run industry structure
Beta 2.2-2.6 with 80% 1y realized vol — position sizing must assume -40% drawdowns are normal
Key consensus and capex figures rest on T4 aggregators, not primary filings

What to watch

DRAM contract price trajectory (TrendForce/DRAMeXchange monthly prints) — direction, not magnitude
December earnings: FY27 full-year capex number and any change to 'tighter through FY28' language
Hyperscaler AI capex commentary in January — the demand-side assumption under everything
Forward P/E vs estimate revisions: if estimates rise 10%+ while price stays flat, compression is nearing an extreme that historically resolves violently
SOXX relative strength: MU's idiosyncratic thesis needs it to stop trading as a SOXX proxy (corr 0.83)

Market snapshot when the call was made 2026-10-14

Price

Return 1w	-16.0%
Return 1m	-1.4%
Return 3m	+7.1%
Return 6m	+100.4%
Return 1y	+374.2%
From 52-week high	-24.7%
Realized vol, 20d	50.6%
SPY 3m / 1y	+2.5% / +16.1%
QQQ 3m / 1y	+3.3% / +21.1%

Technicals

MU technicals as of 2026-10-14: last close $914.13; 20d SMA 1,042.16 (bullish); 50d SMA 982.41; 12d EMA 1,010.80; 26d EMA 1,013.64; RSI 14 25.6 (oversold); 20d vol 46.4%; volume x1.0 vs 20d avg

Sources (30 · T1 = primary filings)

Earlier and later calls on MU

Aug 9, 2026Yes — but narrowly, and decelerating.Lean bullish
Aug 9, 2026Memory demand appears sustainable over the next 3-6 months, particularly for HBM and server DRAM, but Micron's extraordinary revenue and margin growth is not eqLean bullish

Want a fresh look at MU, or another name?

Queue a new deep-research run. Published to the feed when done.

Automated research opinion, published as-is on Oct 15, 2026. Not personalised advice. Figures use end-of-day closes.