Skip to content
Sign in
LONGActiveMed1Y$MU$877.55Fri 23 Oct

Micron: priced like a cyclical at the top, contracted like a utility

By MenFem Editorial·Semiconductors·9 June 2026·Methodology·
ai-infrastructuresemiconductorsmemory
Share
Micron: priced like a cyclical at the top, contracted like a utility

The argument, taken apart

Turn it, hover a ring, a moon or a weight
One call, taken apart
Rings are the falsifier’s legs. Moons are the catalysts it named. Weights are its risks, longer for higher severity.
The claim

Priced like a peak-cycle memory-cyclical, contracted like a utility — a re-rating bet, not an earnings bet.

What would make it wrong · 1 leg
  1. Leg 1 · Not fired
    The contracts stop being contracts into 2027
    “Noncurrent customer contract liabilities rose from $568M to $12,895M.”
    read 9 Oct 2026
The window
9 Jun 2026 to 9 Jun 2027
38% gone · 226 days left
next print 24 Dec 2026
The reviews
Reviewed on 3 days · last 9 Oct 2026
3 Aug · 6 Sept · 9 Oct
Column = the claim, taller for more conviction · rings = falsifier legs, lit when firing · moons = catalysts · weights = risks, longer for higher severity
  • Leg 1, not fired: The contracts stop being contracts into 2027, read 9 Oct 2026: “Noncurrent customer contract liabilities rose from $568M to $12,895M.”
  • FQ3 2026 earnings
  • HBM4 16-Hi volume ramp
  • high severity: Memory-cycle reversal
  • medium severity: Margin mechanics reverse

Key Points

  • Entire CY2026 HBM output sold under multi-year fixed price-and-volume contracts.
  • Overtook Samsung into the #2 HBM slot; power-efficiency edge keeps it in every NVIDIA generation.
  • ~10x forward earnings prices a roll-over the contracts are designed to prevent — a re-rating, not earnings, bet.
Price at Call
$935.89
Micron

Micron is the cleanest expression of a simple mispricing: the market values it as a peak-cycle memory-cyclical at roughly ten times forward earnings, while the business is quietly becoming a contracted-capacity supplier. The whole of its 2026 high-bandwidth-memory output is sold under multi-year, fixed price-and-volume contracts — the structural inversion of the spot-priced boom-bust that has defined memory for forty years. Micron has overtaken Samsung into the number-two HBM slot, its power-efficiency edge keeps it qualified in every NVIDIA generation, and data-center is now a third of the company. The call is BULLISH but explicitly a re-rating bet, not an earnings bet: earnings are at a cycle high; the multiple is at a cycle low; if the contracts hold through the next capex wobble, the multiple re-rates. Conviction is MEDIUM and entry timing matters — this is the number-three-becoming-number-two in a three-player cartel, bought after a vertical run, and the same contracts that lower the trough do not abolish the cycle.

Research Log

Reviewed 2026-10-09 against the Sep 30 print. Tape: 935.89 → 1,057.76 (+13.0%). Falsifier NOT FIRED: noncurrent customer contract liabilities rose from $568M to $12,895M and customers paid $12,747M of deposits in cash in FY26, the strongest evidence yet that the contracts are contracts (smaller than the $22B we had assumed; the 10-K note is still to come). The re-rating has not happened: about 6.9x forward on the FQ1 guide, below the ~10x at entry. On this call's own terms that is a miss, not a break, and the gain so far is earnings, not the thesis. KEEP.

Reviewed 2026-09-06. Tape: 935.89 → 1,016.59 (+8.62%), worst weekly close −12.06%, peak-to-trough −27.42%, vs QQQ +6.65pp. Falsifier NOT FIRED. The 6 Sep dossier: CXMT in small-batch HBM3E (a bull nuance falsified), and the Chief Business Officer who owns the $100B contract book was moved to Senior Advisor on 26 Aug with no successor named. KEEP. Print confirmed 30 Sep (Micron IR, 26 Aug) — the calendar said 23 Sep and was wrong.

Down roughly 30% from the late-June high — while the guidance went up.

July alone took about 28.7% off the stock. Against that, Micron guides Q4 revenue to approximately $50B at roughly 86% gross margin.

Three pressures are doing the work, and none of them is a current-earnings problem. First, rivals are ramping HBM yields successfully, which threatens the pricing power the thesis depends on. Second, the market has started pricing the supply deficit closing into balance by late 2026 rather than 2027. Third, the China supply-side news flow: CXMT announced an $8.5B IPO on 15 July (Micron fell 7% that day), reports surfaced of Chinese progress on DUV lithography, and new US HBM export restrictions were floated.

This call's frame was "priced like a cyclical at the top, contracted like a utility." That frame is now precisely the axis of the argument, which is a good sign for the framing and an uncomfortable one for the position. The thesis breaks if contracted volumes reprice, or if the deficit closes materially earlier than 2027. Neither has shown up in reported numbers yet — but the drawdown says the market is willing to pay for that risk in advance.

Since this call

  1. Micron · 5 insiders, 13-16 Oct 2026: PRSU/RSU vesting (code M) and withholding (code F) at $989.04 on 13 Oct; Ray sold 6,328 shares on 14 Oct at $914-$947; Deboer sold 28,544.572 shares on 16 Oct at $900-$902.01. · sec.gov
  2. Micron · Six insiders filed Form 4s for 13-14 Oct 2026: RSU/PRSU grants, PRSU conversions, tax withholding at $989.04, and 10b5-1 sales by Ray on 14 Oct at $914 to $946. · sec.gov
  3. Micron · DRAM revenue $100.68bn and NAND revenue $31.79bn in fiscal 2026 (fiscal 2025: $28.58bn and $8.50bn); 1,131,423,212 common shares outstanding as of 2 Oct 2026. · sec.gov
  4. Micron · Fiscal Q4 2026 revenue $54.23bn, GAAP net income $37.70bn ($32.87 diluted EPS); FY2026 revenue $133.19bn, GAAP net income $84.97bn. Q1 FY2027 guidance: revenue $61.5bn +/- $1.5bn. · sec.gov
  5. Micron · Before (deep-dive-2026-07-07.md): LONG — MEDIUM-HIGH. The −13% pullback to ~$985 (from the $1,133 that anchored the prior NEUTRAL) is the change that flips the entry: the quality is unchanged and confirmed (clean books, $22B prepaid take-or-pay raising the floor, data-center run-rate >$100B, floor-pricing margins management says beat any prior peak), but the price now pays a fair-to-cheap ~9x forward on a raised trough instead of a full ~8x on the rip. The bull case is no longer fully in the price. Falsifier is unchanged (DRAM/HBM ASPs roll over OR CXMT floods the un-contracted ~60% OR HBM4 multi-sourcing compresses the premium). Prior dossier said 'own the dips, not the rip' — this IS the dip it was waiting for. After (deep-dive-2026-08-10.md): "LONG — MEDIUM (conviction DOWN one notch from the prior MEDIUM-HIGH, on wider risk, not a worse business). At $870 the entry is the best of any pass and the operating facts got BETTER (sold out through all of CY2027; SCA floors now quantified above the 62% all-time-peak GM and running to 2030; incumbent supply relief not landing until 2028-29). But the prior dossier's named falsifier stopped being hypothetical: CXMT was handed $8.6B on 2026-07-31 and has a Shanghai fab completing this half. It has NOT yet fired — DRAM spot rose 16.9% in July and Q3 contract prices are still up 13-18% QoQ, so the deceleration is demand-side, not a supply break. Two things the prior dossier got wrong are corrected here: it called the name regulatory-CLEAN when a US price-fixing class action against all three DRAM makers had already been filed (2026-06-30), and it never noticed FQ4'26 carries a 14th week that flatters the $50B guide by ~7.7%." · micron
  6. Micron · Before (deep-dive-2026-06-29.md): NEUTRAL / WATCHING — the business genuinely transformed (clean books, $22B prepaid take-or-pay raising the floor, data-center 61% of revenue), but at ~$1,133 (~8x PEAK EPS, upside capped at peak-price SCA ceilings, ~60% of revenue spot-exposed to CXMT, EV/EBITDA premium to the HBM leader) the bull case is in the price. Quality confirmed; entry unattractive — own the dips, not the rip. Diverges from the prior web-only dossier's BULLISH on the ENTRY, not the business. After (deep-dive-2026-07-07.md): LONG — MEDIUM-HIGH. The −13% pullback to ~$985 (from the $1,133 that anchored the prior NEUTRAL) is the change that flips the entry: the quality is unchanged and confirmed (clean books, $22B prepaid take-or-pay raising the floor, data-center run-rate >$100B, floor-pricing margins management says beat any prior peak), but the price now pays a fair-to-cheap ~9x forward on a raised trough instead of a full ~8x on the rip. The bull case is no longer fully in the price. Falsifier is unchanged (DRAM/HBM ASPs roll over OR CXMT floods the un-contracted ~60% OR HBM4 multi-sourcing compresses the premium). Prior dossier said 'own the dips, not the rip' — this IS the dip it was waiting for. · micron
  7. Micron · The dossier's stance shifted from "bullish" (deep-dive-2026-06-25.md) to "neutral" (deep-dive-2026-06-29.md). · micron
  8. Micron · Before (deep-dive-2026-06-25.md): The blowout FQ3 + the $100B take-or-pay contract book is the strongest evidence yet that the trough has been raised — but on the multiples that DON'T reset at peak earnings (EV/EBITDA ~32x vs 6.8x median; P/B ~16x), MU is NOT cheap; it is FAIRLY-to-RICHLY priced for a structural memory company and CHEAP only if you accept the contracts have permanently broken the cycle. The ~10x forward P/E is ~10x on peak EPS — the classic cyclical trap signature. BULLISH / MEDIUM (a re-rating + raised-trough bet, NOT a cheapness bet), 1Y. After (deep-dive-2026-06-29.md): NEUTRAL / WATCHING — the business genuinely transformed (clean books, $22B prepaid take-or-pay raising the floor, data-center 61% of revenue), but at ~$1,133 (~8x PEAK EPS, upside capped at peak-price SCA ceilings, ~60% of revenue spot-exposed to CXMT, EV/EBITDA premium to the HBM leader) the bull case is in the price. Quality confirmed; entry unattractive — own the dips, not the rip. Diverges from the prior web-only dossier's BULLISH on the ENTRY, not the business. · micron
  9. Micron · Before (deep-dive-2026-06-09.md): The market prices Micron as a peak-cycle memory-cyclical (~9–10× forward) while it becomes a structural HBM oligopolist on multi-year contracts — a re-rating bet, not an earnings bet, with peak-cycle entry risk. BULLISH / MEDIUM / 1Y. After (deep-dive-2026-06-25.md): The blowout FQ3 + the $100B take-or-pay contract book is the strongest evidence yet that the trough has been raised — but on the multiples that DON'T reset at peak earnings (EV/EBITDA ~32x vs 6.8x median; P/B ~16x), MU is NOT cheap; it is FAIRLY-to-RICHLY priced for a structural memory company and CHEAP only if you accept the contracts have permanently broken the cycle. The ~10x forward P/E is ~10x on peak EPS — the classic cyclical trap signature. BULLISH / MEDIUM (a re-rating + raised-trough bet, NOT a cheapness bet), 1Y. · micron

Bull Case

The number-two supplier in a three-player HBM cartel selling a fixed pie at rising prices on multi-year contracts, priced at ~10x forward for a roll-over the contracts are built to prevent.

Bear Case

Commodity memory at a cycle peak; 80% margins are partly write-down recovery plus an under-depreciated asset base that reverse together when the 2027 HBM4 ramp re-floods.

What would prove this wrong

The reason for holding this stops being true if the contracts stop being contracts — renegotiated, broken, or not renewed on comparable terms into 2027. That is the entire basis for 'contracted like a utility'; without it this is a peak-cycle memory name on a peak-cycle multiple and the market's read was right. Note what is NOT invalidation: the multiple failing to re-rate is a miss, not a thesis break. This call states plainly that it is a re-rating bet, and a bet that has not paid yet is not a bet that is wrong.

Catalysts

FQ3 2026 earningsEarnings

Record ~$33.5B revenue guide at ~81% gross margin; the print that tests the contract narrative.

HBM4 16-Hi volume rampProduct Launch

Share and pricing in the HBM4 generation against SK Hynix and Samsung.

Risk factors

Memory-cycle reversalHigh

All three players are expanding; the 2027 HBM4 ramp is the dated expiry of structural tightness.

Margin mechanics reverseMedium

Peak margins are partly write-down recovery and under-depreciation that unwind together when ASPs roll.

Related Calls

LONG$TSMActive

TSMC: own the bottleneck, not the brand

Own the bottleneck, not the brand: the chokepoint every AI chip depends on, at a discount to all of them.

Fabricates and CoWoS-packages every leading-edge AI accelerator — NVIDIA, AMD, Broadcom, Apple alike.

~22x forward earnings: a discount to NVIDIA, AMD and Broadcom, the names that cannot ship without it.

ai-infrastructuresemiconductorsfoundry
TSMC
High3YSemiconductorsUpdated 3d ago9 Jun
WATCH$AVGOActive

Broadcom: the right custom-silicon thesis at the wrong price

The right custom-silicon thesis at the wrong price — watching the most expensive, lowest-quality way to play it.

The named winner of 'custom silicon eats commodity inference' — AI revenue +143%, FY27 AI guided >$100B.

Trades at a premium to NVIDIA on every earnings multiple while earning a third of the ROE.

ai-infrastructuresemiconductorscompute
Broadcom
Med1YSemiconductorsUpdated 3d ago9 Jun
LONG$000660.KSActive

SK Hynix: the largest HBM rent, valued like a Korean cyclical

The largest slice of the HBM scarcity rent, valued like a Korean cyclical — with a Nasdaq listing as the re-rating catalyst.

~62% HBM share and ~two-thirds of NVIDIA's HBM4 — the largest rent-capture in the cluster.

72% operating margin and ~61% ROE, yet a single-digit forward multiple — rent in the earnings, not the multiple.

ai-infrastructuresemiconductorsmemory
SK Hynix
Med1YSemiconductorsUpdated 3d ago9 Jun

Conviction

Conviction History

Med→Med

3 Aug 2026

Reviewed against the Q2 2026 earnings wave (SK Hynix 29 Jul, Samsung 30 Jul, and the late-July hyperscaler prints). Guidance improved (~$50B Q4 revenue at ~86% GM) while the stock fell ~30%. New structural pressure noted (CXMT, mooted export limits, rivals ramping yields) but nothing yet in reported numbers. Held — no conviction change.

Weekly closes

entry 935.89

Sources

See all Semiconductors calls