Phase Space AI

Micron Trade Construction 2026-07-27

Micron Technology [MU]

Micron Technology, Inc. [MU] — Criteria Analysis & Trade Construction

Investing Hub Research | analysis 2026-07-27, migrated to the Criteria framework 2026-07-29 | framework v1.5.1

2026-07-29 MIGRATION NOTICE — read this before anything below it

This document has been migrated from the retired six-Gate / E[R]-vs-cash-hurdle framework onto the Criteria framework (references/criteria.md, references/valuation.md). Three things changed materially, and none of them is cosmetic:

1. The position verdict is gone. The prior version of this document ended in WATCHLIST. The memo no longer emits Long / Short / Watchlist / Avoid. It emits an analysis; whether that analysis justifies a position is a question about a particular book, and two books answer it differently. Sections 4 through 8 below have been rewritten accordingly — the evidence is kept, the verdict is deleted.
2. E[R] versus the 4.7% cash hurdle is gone, replaced by the two mandatory valuation outputs: the reverse-DCF implied-path test (§3A) and a 12-month target (§3B). Sensitivity is run over the exit multiple, never over scenario probabilities.
3. Gates are retired; every test is a named Criteria with a type (BINDING / MEASURED) returning PASS / FAIL / INDETERMINATE. A missing input is INDETERMINATE, never FAIL.

Pre-computed inputs are taken from reports/scan/MU_analysis.json (as-of 2026-07-28) and are used rather than recomputed, except where explicitly reconciled below. Price is the scan's $820.51 (2026-07-28), not the $900.20 / $884.75 used in the 2026-07-27 build.

---

0. PORTFOLIO CONTEXT — retained as context, no longer a test

portfolio_book.json as of 2026-07-27: no open positions; the book is 100% cash. Config: max single-name weight 5%, max gross 100%, pairwise-correlation disclosure threshold 0.60, drawdown ladder −5%/−10%/−15%, position hard stop −2% of book.

The 4.7% cash hurdle is retired and is not applied anywhere in this document. Under the current framework capital competition is resolved by slot competition in the strategy, not by a return threshold inside the memo. What remains relevant and is retained:

• Correlation (forward-looking, and material): the 2026-07-27 screen found ~20 of 47 long candidates in a single semiconductor/AI-hardware complex. MU would be the anchor of that cluster. An MU exposure and an NBIS exposure would not be two positions; they would be one factor bet in two tickers.
• Volatility: 252-day realised vol 78.0% (scan). Whatever a book does with this name, inverse-volatility sizing is the operative control.

1. FOUR CONCLUSIONS, KEPT SEPARATE

1. Fundamental conclusion. Micron's business is, right now, performing better than at any point in its history and better than the market expected a quarter ago. Revenue quadrupled year-over-year; gross margin is 84.6%; inventory is flat while revenue quadrupled (a genuine shortage signature); the balance sheet went to +$24.4B net cash while retiring $9.0B of debt. Nothing is deteriorating.
2. Expectations conclusion. The market expects a great deal, and is rapidly expecting more: FY2027 consensus EPS rose 53% in ninety days on 30 upward revisions and zero downward. Street price targets average ~$1,492–$1,569. The house model matches FY2026 consensus almost exactly but sits 22% below consensus FY2027 revenue and 42% below consensus FY2027 EPS.
3. Valuation conclusion. Under every method that assumes memory remains cyclical, the equity is worth materially less than $900.20 — normalised earnings power $474, probability-weighted DCF $479, and a price-to-book of 10.1x against a company that has never sustained more than ~3x. Under an explicit regime-change assumption (permanent revenue 4–5x FY2025 at permanently post-peak margins), the current price is defensible.
4. Portfolio conclusion. See the gates. A correct research view and a correct position are different things.

2. THE CRITERIA BLOCK (framework v1.5.1)

Every Criteria is scored. The memo blocks on none of them — BINDING/MEASURED describes how the long-only absolute-return strategy consumes the score, not a gate inside this document.

Nothing in this block is a rejection. The two BINDING Criteria both PASS on TTM-consistent inputs.

What the SCA disclosure does change — stated precisely, not rhetorically

It would be easy to let a striking new disclosure quietly re-rate the whole memo. It should not, and here is the arithmetic that keeps it honest:

• The floor is real but small — now verified by the explicit two-tranche build, not asserted. $100B of RPO over a five-year term is $20.0B/year. Three denominators, all live formulas in the workbook: 12.1% of the current $165.8B annualised run-rate; 10.3–15.8% of Base-case revenue across FY2026–FY2030; and 9.1% — one part in eleven — of the $220.8B of permanent revenue that $900.20 requires at 13x and a 48% post-contract margin. At the contractual floor the tranche generates $12.2B of gross profit a year against the $106.0B the price needs, i.e. one part in 8.7, and ~$7B of net income — a ~0.7% earnings yield on a $1,016.7B market capitalisation. It truncates the disaster; it does not underwrite the valuation.
• Where it does bite: the deep tail. Run the FY2023 analogue (merchant GM at the −9.1% trough) against Bear FY2028 revenue and the contracted tranche swings gross profit from −$7.2B to +$5.4B — a $12.6B truncation. A negative-gross-margin year is contractually impossible while the SCAs run. That, not the 5% lift to the modal Bear, is the real Tier 1 effect, and it is what justifies P(Bear) at 30% rather than 35%.
• Principle 5 holds — the floor rose 13x more than the ceiling. FY2028 gross-profit uplift from the contracted tranche: Bear +$4,860M, Base +$2,800M, Bull $0. On DCF value per share: Bear +$8.46 (+5.1%), Base +$4.60 (+1.1%), Bull +$0.65 (+0.1%). Because contracted margin is MAX(floor, MIN(merchant, ceiling)), the floor only lifts the tranche when the merchant market is below it — and in the Bull it never is. The regime-change reading survives the test that would have exposed it as narrative.
• The ceiling is real too, and nobody is discussing it. Micron's own words: when all planned SCAs are executed, "agreements with either fixed prices or price ceilings at or close to current CQ2 market prices are expected to be approximately 40% of our revenue." CQ2-2026 is the 84.6%-gross-margin quarter. Micron has contractually sold away the upside above record pricing on nearly half its book. The SCAs are a variance-reduction trade, not a level-raising one — and the Street's uniformly bullish read ("game-changing," "fundamental shift") has priced the floor while ignoring the cap.
• The mix finding cuts against the bulls. CDBU and MCBU — conventional DRAM — now earn 87% gross margins, above CMBU's 83%. The current profit engine is commodity pricing, which is the mean-reverting variable and precisely where CXMT's ~350k wafer-starts/month of end-2026 capacity lands. The "HBM franchise insulates Micron" argument is weaker than it looks, and the transcript data is what exposed it.

3. VALUATION — TWO OUTPUTS, BOTH MANDATORY

E[R] versus a cash hurdle has been deleted. The old §3 computed 0.30(+80.6%) + 0.48(+47.3%) + 0.22(−66.6%) = +32.2% for a short and ran its range across scenario probabilities — the precise defect the current framework replaces, because the probabilities could not change the answer while the exit multiple, which determined it, carried a point estimate.

3A. Implied-path test (reverse DCF) — the Valuation Criteria

THE DATA CORRECTION. Old EV/Sales 24.6x. New EV/Sales 10.25x.

Every valuation conclusion in the 2026-07-27 documents that rested on last-FY revenue is wrong and is superseded here. Note that Micron_Valuation_Analysis_2026-07-27.md §1 did use a TTM figure ($90.3bn, computed as FY25 Q4 + FY26 Q1–Q3) and reported 11.0x — it is the screen/scan path and the multiples that fed the gate scoring that used last-FY. The two paths disagreed by 2.2x and nothing reconciled them.

The correction runs in both directions, and this is the part that changes the verdict. The scan's revenue_cagr_demonstrated of 6.7% is a three-year CAGR computed on annual revenue ending FY2025 — the same stale endpoint. Measured to the same TTM endpoint as the revenue itself:

The FY2022 anchor is deliberately the prior cycle peak ($30.758bn), not the FY2023 trough ($15.540bn) — this is a peak-to-peak measurement, which is the correct base-rate framing for a deep cyclical and is the opposite of cherry-picking. The longest available window corroborates it at 28.0%.

The reverse DCF (assets/reverse_dcf.py, bisection on CAGR):

RESULT — what the price requires: a 22.9% five-year revenue CAGR. Margin = demonstrated − required = 32.6% − 22.9% = +9.7pp. → PASS On the uncorrected annual-FY growth measure the same run reads 6.7% − 22.9% = −16.1pp → FAIL. The entire difference between PASS and FAIL on this name is which revenue endpoint the growth measure uses.

Implied compression, as a number. Exit 22.6x EV/EBIT against today's 39.2x = −42.3%. On a sales basis the exit multiple is 22.6 × 26.10% = 5.90x EV/Sales against today's 10.25x = −42.5%. The two agree, which is a consistency check on the terminal margin.

Exit-multiple identification — declared honestly, per the growth-match rule. The scan's anchor is GROWTH_MATCHED on growth and its 15 constituents do bracket MU's growth. But 14 of the 15 are healthcare or software — THC 7x, ELV 14x, DGX 19x, MDT 20x, CRM 22x, CAH 22x, LH 22x, STE 22x, ZBH 23x, COO 23x, IQV 24x, BDX 25x, JNJ 30x, HAE 30x, KLAC 30x. Only KLAC is a semiconductor company. Restricting the same growth-matched screen to the semis/optical cohort leaves n = 1, below the minimum of 5, so on a sector-and-growth-matched basis MU's exit multiple is UNIDENTIFIED. The 22.6x is reported because it is the framework's number and it is growth-matched as the rule is written; it is not claimed to be business-matched, and the sensitivity below is the honest expression of that.

Sensitivity — over the EXIT MULTIPLE, never over probabilities. Range is the observed spread of the growth-matched anchor set:

The result is PASS across the entire anchor set except its single lowest member (a hospital operator at 6.8x). Terminal-margin cross-check: re-running at the semis cohort median of 18.45% instead of MU's own 26.1% raises the required CAGR and narrows the margin; the base case as specified uses the rule as written (max(own, cohort median)) and this cross-check is disclosed rather than buried.

Terminal value is well above 60% of EV, so the reverse DCF is the primary long-horizon output and the forward DCF in Micron_Valuation_Analysis_2026-07-27.md §3 is supporting evidence only.

3B. The 12-month target

Built per references/valuation.md: near-term consensus revenue, named product-cycle events, and a multiple anchored on MU's own trading history and percentile — never a peer median projected forward.

• Revenue base: FY2027 consensus revenue $238.82bn (38–41 analysts), sourced in Micron_Valuation_Analysis_2026-07-27.md §consensus. Micron's FY2027 runs Sept-2026 → Aug-2027, so this is very nearly exactly the revenue that will be trailing at the target date — which is what makes it the correct base for a trailing multiple.
• Named product-cycle events inside 12 months (each dated in the catalyst calendar, none invented): FY2026 Q4 print with the first formal FY2027 framing (~late Sept 2026); the 16 take-or-pay Strategic Customer Agreements, whose contractual gross-margin floor binds FY2028+; CXMT's conventional-DRAM capacity ramp toward ~350k wafer starts/month by end-2026.
• Multiple anchor — MU's own EV/Sales, daily, filed-date aligned (no look-ahead), 1,508 observations from 2020-07-27: full-history median 4.13x, p75 5.82x; trailing 3 years: p25 4.66x, median 5.82x, p75 10.60x, p90 17.12x. Current 10.25x = 74th percentile of the trailing 3 years (87th of the full history).

12-MONTH TARGET: $1,250 — +52.3% ABOVE SPOT.

The single most informative line here: the multiple required to merely hold spot on consensus FY2027 revenue is 3.79x EV/Sales — 63.0% below today's 10.25x, and below the 25th percentile of MU's own trailing three years. Today's price does not embed multiple expansion; it embeds multiple compression offset by revenue growth.

Stress on the revenue base — the house's own bearish number, not consensus. The 2026-07-27 build's house FY2027 base revenue is $185.6bn (−22.3% vs Street). On that base: hold-spot multiple 4.88x; own-p25 4.66x → $784 (−4.4%); own-median 5.82x → $976 (+18.9%). So even granting the house's bearish revenue in full, MU at its own median trailing multiple is above spot.

Sanity band against the external target. Street average ~$1,492–$1,569. The house 12-month target of $1,250 sits 16–20% below it. The divergence is explained and not tuned away: the Street's target embeds both consensus FY2027 revenue and a multiple above MU's own three-year median. This target uses consensus revenue with the name's own median multiple. The external target is used as a check on the output, never as a calibration target for the model.

Note on direction versus the prior build. This is the first MU target in this project that sits above spot. Calibration item B16 recorded 16 of 16 house targets below spot at a median 46.1% below Street. That pattern is a house view about the market, not a valuation, and it is not reproduced here.

4. WHAT THE ANALYSIS ESTABLISHES — no position verdict

This memo does not output a position. The prior version of this section read WATCHLIST — two-sided, not actionable. That verdict has been deleted, not softened: whether this analysis justifies a position is a question about a particular book — its slots, its existing factor exposure, its volatility budget — and the memo is strategy-agnostic. The evidence below is kept in full; the conclusion drawn from it belongs downstream.

What is established, stated as findings rather than a call:

1. The business is performing at the best level in its history and nothing is deteriorating. Revenue quadrupled year-over-year, gross margin reached 84.6% on the most recent print, inventory is flat while revenue quadrupled (a genuine shortage signature), and the balance sheet moved to +$21.7bn net cash while retiring $9.0bn of debt. Accruals −0.12 and F-score 8/9 corroborate rather than contradict.
2. The valuation objection in the 2026-07-27 build was substantially a data artifact. It rested on last-FY revenue of $37.4bn — 136% below the TTM figure of $88.3bn — which produced EV/Sales of 24.6x against an actual 10.25x. Correcting the endpoint on both the multiple and the demonstrated growth rate moves the Valuation Criteria from FAIL (−16.1pp) to PASS (+9.7pp).
3. What the price actually requires is 22.9% five-year revenue CAGR against 32.6% demonstrated peak-to-peak. That is a demanding but not extraordinary path, and it is achievable on the company's own record rather than on narrative.
4. The residual risk is a conventional-DRAM supply response, not a solvency or accounting problem. It is named, dated and monitorable (CXMT capacity milestones, monthly DRAM contract pricing, FY2027 revision counts), and the left tail is contractually truncated by 16 take-or-pay SCAs — worth roughly +$8.46 on the modal bear DCF and ~$1.7/share on the weighted target, which is materially less than the 2026-07-27 v1.4.0 build credited and is stated here as a correction to that build, not to the framework.
5. Momentum is a tailwind, not an obstacle. 12-1 momentum sits in the 98.4th cross-sectional percentile. The old framework failed the name for "fighting momentum" on a short framing and then carried that failure into the overall verdict. Under the current framework momentum is MEASURED and governs entry timing only; RSI-14 at 39.3 with price 32.5% off the high is a timing observation, and it is the only thing momentum is permitted to say.

What would refute the above, stated so it is falsifiable: FY2027 revision counts turning net negative from 30-up/0-down; the first sequential gross-margin decline; FY2027 revenue guided below ~$185bn at the late-Sept print; or CXMT capacity landing materially ahead of the ~350k wafer-starts/month end-2026 path. Each of these is dated in Micron_Catalyst_Calendar_2026-07-27.md.

5. MONITORABLE TRIGGERS (checkable, dated, pre-committed)

These are no longer conversion triggers for a position verdict — no verdict is emitted. They are the dated, falsifiable observations that would move the Criteria scores, and they are listed so that a book consuming this analysis can act on pre-committed evidence rather than on the emotion of the moment.

→ Evidence that would move the Quality and Valuation Criteria toward impairment:

Expectations (Quality Criteria — evidenced deterioration): - E1. Trailing-30-day EPS revisions for FY2027 turn net negative (currently 30 up / 0 down). This is the single cleanest early signal and is checkable monthly from Alpha Vantage. - E2. First sequential gross-margin decline reported, or FY2027 revenue guidance below ~$186B (the house Base case) at the FY26 Q4 print. - E3. DRAM contract pricing declines for two consecutive months (industry data), or CXMT announces volume production milestones on its post-IPO capacity.

New under v1.4.0 — SCA-specific triggers, checkable from disclosure the company has now committed to making: - E4 (the cleanest of all). RPO stops growing. Micron began disclosing remaining performance obligations in the May 2026 quarter ($5B at FQ3, ~$100B including post-quarter signings). RPO is now a quarterly, audited, contractual figure. If reported RPO fails to converge toward the claimed ~$100B by the FQ4 10-K, or declines thereafter, the entire business-model-transformation claim is falsified in a single line item. This did not exist as a checkable trigger before this revision. - E5. SCA count stalls below the "half or more of revenue" target, or a signed SCA is disclosed as amended, renegotiated or impaired. Take-or-pay contracts are only worth their counterparties' solvency. - E6 (bull-side). CDBU/MCBU gross margins roll over while CMBU holds. Because conventional DRAM is now the higher-margin business, this specific divergence — not aggregate margin — is the earliest read on whether the cycle is turning underneath the contracts.

Tape (Momentum Criteria — entry timing only, and it can never block or compel ownership): - T1. A weekly close below the 200-day moving average (currently ~$505 and rising) — the actual trend break, not the current pullback. - T2. Or a confirmed lower-high / lower-low structure on weekly bars over two-plus months, with the 50-day crossing below the 200-day.

→ Evidence that would strengthen the implied path: price approaching the normalised-earnings anchor (~$525, i.e. roughly −46% from here) while the HBM contract structure demonstrably holds — or if a second year of contracted, sold-out HBM at post-peak margins is disclosed with industry capacity discipline intact, which would legitimately move weight from the Bear scenario to the Bull and re-anchor the whole valuation.

6. SIZING AND VEHICLE ARITHMETIC (specified now, while calm — not a recommendation)

Sizing arithmetic, offered to the strategy rather than decided here. Volatility is High by a wide margin — 252-day realised vol 78.0% (scan) is nearly double the ">45% = High" threshold. Inverse-volatility sizing is the operative control and it caps this name tightly whatever view a book takes: on the sizing grid, High volatility with Medium conviction is ~1% of portfolio, and at 78% realised vol that 1% is a ceiling, not a starting point. The memo supplies the arithmetic; the book decides whether any of it is used.

(The prior text here concluded "round down to Watchlist". That conclusion is deleted — sizing is a constraint, not a verdict, and conflating the two is exactly what the framework now forbids.)

Vehicle — defined-risk spread, not naked premium (skill default). Live options data re-pulled at the revised $900.20 spot, 2026-07-27:

At 90% implied volatility, an at-the-money October put costs $150.82 — 16.8% of the stock price for eleven weeks. Buying naked premium here is close to the worst possible expression of a bearish view; the variance risk premium (Coval & Shumway 2001; Bakshi & Kapadia 2003) is at its most punitive exactly in this configuration.

Recommended structure on conversion: the October $800/$640 put spread. - Buy $800 put at the $100.03 ask; sell $640 put at the $41.46 bid → net debit $58.57 per contract ($5,857) - Max loss $58.57 (the debit) | Max gain $160.00 − $58.57 = $101.43 | Reward:risk 1.73:1 - Breakeven $741.43 — requires an 17.6% decline by expiration

The implied-vs-expected-move test (required before any naked premium): 93% IV over 81 days to expiration implies a one-standard-deviation move of ±43.8%. The house Base case implies −53% — but that is a twelve-month convergence, not an 81-day move, and the report has no basis for claiming a −44%+ move inside eleven weeks. The expected move does not exceed the implied move over the option's actual life; therefore naked long puts are not justified and the spread is the correct structure. Stating both numbers is what makes this auditable rather than a preference.

7. RISK MANAGEMENT PARAMETERS, IF A BOOK TAKES A POSITION

Governed by the portfolio-book dynamic risk protocol, pre-committed here: - Position hard stop: loss exceeding 2% of book → cut to half size immediately and fully re-underwrite before any re-add. For a short in a 12x-in-fifteen-months stock this is not theoretical. - Invalidation (thesis, not price): a reported sequential gross-margin expansion above 85% together with FY2028 revenue guidance above $250B would mean the regime-change case is winning — exit regardless of P&L. - Invalidation (price): a weekly close back above the prior high of $1,213.37 invalidates the tape thesis outright. - Gap discipline: if the stock gaps through the invalidation level on an earnings print, execute the exit per the pre-committed trigger; re-evaluate flat, never intraday and never while short a name with unbounded convex loss.

8. BOOK WRITE-BACK

The memo no longer writes a direction or a verdict to the book. It writes the analysis record in the §OUTPUT FORMAT block below, and the strategy decides what, if anything, to do with it. What is recorded:

ticker: MU | framework: v1.5.1 (Criteria) | analysis_date: 2026-07-27 | migrated: 2026-07-29
binding_criteria: Quality PASS, Valuation PASS (+9.7pp), Liquidity PASS
valuation_margin_pp: +9.7   (uncorrected annual-FY growth measure: -16.1)
target_12m: $1,250 (+52.3%)   implied_path_required_cagr: 22.9%   demonstrated: 32.6%
momentum_12_1_pctile: 98.4 (MEASURED - entry timing only)
vol_252d: 78.0%  ->  inverse-volatility sizing is the binding practical constraint
refutations: see §5

No entry is made in trade_recommendations.jsonl — the ledger records specific trade recommendations, and none is being made here. The Downside Criteria bear case and its probability are logged to the ledger on every name per criteria.md, and that logging is what earns the Criteria its promotion to BINDING once Brier-scored.

---

OUTPUT FORMAT (machine-readable analysis record — no position verdict)

FRAMEWORK: v1.5.1 (Criteria). NO POSITION VERDICT IS EMITTED. The prior record read
       "INVESTMENT DECISION: WATCHLIST (two-sided, not actionable)". It is deleted, not softened.
CRITERIA:
  Quality Criteria      (BINDING)  PASS  — COMPOUNDER; FY25 op margin 26.1%; accruals -0.12; F-score 8/9
  Valuation Criteria    (BINDING)  PASS  — required 5y CAGR 22.9% vs 32.6% demonstrated (peak-to-peak,
                                           TTM-consistent). MARGIN +9.7pp.
                                           On the uncorrected annual-FY growth measure: -16.1pp, FAIL.
  Liquidity Criteria    (BINDING)  PASS  — ~$927bn cap, 30-60m sh/day, deep quoted options
  Downside Criteria     (MEASURED) scored — permanent-loss cause: conventional-DRAM supply response (CXMT
                                           ~350k wspm end-2026). Impairment case ~$474/sh (-42%).
                                           NOT a going-concern case: +$21.7bn net cash, 16 take-or-pay SCAs.
  Catalyst Criteria     (MEASURED) PASS  — FY26 Q4 print ~late Sept 2026; first formal FY27 framing + RPO
  Momentum Criteria     (MEASURED) scored — 12-1 +928.8% = 98.4th pctile (Q5). RSI-14 39.3. 67.5% of 52w high.
                                           ENTRY TIMING ONLY. Never a selection veto.
  Peer Spread Criteria  (MEASURED) scored — vs SNDK 12.22x EV/S, WDC 54.6x EV/EBIT; MU 10.25x / 39.2x
  Consensus Criteria    (MEASURED) scored — house FY27 rev -22.3%, EPS -42.0% vs Street. Blocks nothing.
  Short Mech. Criteria  (MEASURED) scored — 1 of 2 legs (margin runway thin; growth accelerating, not decel.)
  Sub-sector Criteria   (MEASURED) Semiconductors - memory (DRAM/NAND/HBM)

DATA CORRECTION (the deliverable):
  Revenue basis  OLD last-FY FY2025 $37.378bn  ->  NEW TTM to 2026-05-28 $88.260bn  (+136%)
  EV/Sales       OLD 24.6x                     ->  NEW 10.25x
  Demonstrated CAGR OLD 6.7% (annual FY)       ->  NEW 32.6% (TTM-consistent, FY2022 cycle peak -> TTM)
  Valuation Criteria OLD FAIL (-16.1pp)        ->  NEW PASS (+9.7pp)

SPOT: $820.51 (2026-07-28)   EV: $904.95bn   Net cash: +$21.73bn   Shares: 1,129.4m   Vol(252d): 78.0%

IMPLIED-PATH TEST (reverse DCF; 5y, WACC 10%, terminal EBIT margin 26.10% = max(own, semis cohort median)):
  REQUIRED 5Y REVENUE CAGR: 22.9%   DEMONSTRATED: 32.6%   MARGIN: +9.7pp
  EXIT MULTIPLE: 22.6x EV/EBIT, basis GROWTH_MATCHED (n=15)
  IDENTIFICATION WARNING: 14 of the 15 anchors are healthcare/software; only KLAC is a semi. Restricted to
       the semis/optical cohort the growth-matched set is n=1 (<5) => UNIDENTIFIED on a sector basis.
  IMPLIED COMPRESSION: 22.6x vs 39.2x EV/EBIT = -42.3%   (= 5.90x vs 10.25x EV/Sales = -42.5%)
  SENSITIVITY OVER THE EXIT MULTIPLE (never over probabilities):
       6.8x -> req 56.2% (margin -23.6pp, FAIL) | 21.5x -> 24.1% (+8.5pp) | 22.6x -> 22.9% (+9.7pp)
       24.4x -> 21.0% (+11.6pp) | 30.4x -> 15.8% (+16.8pp)

12-MONTH TARGET: $1,250  (+52.3% vs spot $820.51)
  Base revenue: FY2027 consensus $238.82bn (38-41 analysts)
  Multiple: MU's OWN trailing-3y EV/Sales median 5.82x. Current 10.25x = 74th pctile of own 3y range.
  Band: own p25 4.66x -> $1,004 (+22.3%) | own p75 10.60x -> $2,261 (+175.5%)
  Multiple required merely to HOLD spot on that revenue: 3.79x (-63.0% vs today) - below own p25.
  On the HOUSE bearish FY27 revenue $185.6bn: own-median -> $976 (+18.9%); own-p25 -> $784 (-4.4%)
  Street average target ~$1,492-$1,569; house target sits 16-20% below it. Used as a CHECK, not a
       calibration target.

MONITORABLE REFUTATIONS: FY27 revisions turn net negative (from 30up/0down); first sequential GM decline;
       FY27 revenue guided below ~$185bn; reported RPO fails to converge to ~$100bn; CDBU/MCBU margins roll
       over while CMBU holds; CXMT capacity ahead of ~350k wspm.
THESIS IN ONE SENTENCE: Micron has contractually narrowed its own outcome distribution - 16 take-or-pay
       agreements put a gross-margin floor above any prior cycle peak under ~$20bn/yr of revenue - and the
       2026-07-27 build's "-47% overvalued" conclusion was substantially an artifact of valuing $88.3bn of
       TTM revenue as though it were the $37.4bn last fiscal year printed.