CIP contamination cannot be unwound, and it was never the binding constraint — government-incentive netting is a bigger contaminant, and the honest band is 29-57%
Not a recommendation. No buy, sell, short, or allocation is expressed or implied anywhere below. This is disclosure forensics against a threshold the vault pre-registered on 2026-09-06.
The question
Verbatim from the backlog row: "Can Micron's construction-in-progress contamination be unwound — does any disclosure (capitalized interest, assets-not-yet-in-service, segment detail) separate building construction from tool installation, or is the shell-share threshold permanently a band rather than a number?"
This is open follow-up #5 of [[2026-09-06-vendor-capex-composition-memory-anchor]], which proposed shell_share = Δ(Land + Buildings + CIP) / Δ(total gross PP&E) as the supply-discipline gauge and flagged CIP contamination as its main precision limit. The item was gated on the FY2026 10-K existing.
Gate status: PASSED. Micron Technology (CIK 0000723125) filed its FY2026 Form 10-K on 2026-10-09, accession 0000723125-26-000023, period ended 2026-09-03 (SEC submissions API, retrieved 2026-10-14 01:02 ET). It is the primary source for everything below that is not explicitly labelled otherwise.
What we already know (from the vault)
- The metric and its pre-registered threshold. [[2026-09-06-vendor-capex-composition-memory-anchor]] set tripwire (1) at
shell_share≥ 45% TTM for two consecutive quarters, acquisitions stripped, against an FY2025 baseline of 31.1%; and tripwire (2) atequipment not placed into serviceup >50% YoY. The composite rule declares a composition trigger only on (1) AND (2). - The last reading was ambiguous and the vault said so. Through nine months of FY2026 the metric was 45.7% raw, 38.4-51.8% ex-Tongluo — "at the threshold and ambiguous, not through it" — while ENIS was up only +3.7%, clearly not triggered.
- The disclosure floor was already diagnosed correctly. [[2026-08-01-hyperscaler-capex-merchant-vs-inhouse-silicon-decomposition]] established that ASC 360 requires PP&E only "by major class," and that no finer split is obtainable from US GAAP filings. That holding survives this brief intact.
- The re-pricing rationale. [[2026-10-01-micron-fq4-fy2026-dram-hbm-tripwire-rescan]] raised this item to High because FY2027 capex was guided roughly double, with the increase skewed to construction aimed at late-CY2028 cleanrooms — making the readability of the composition signal load-bearing rather than cosmetic.
- The access gotcha.
ConstructionInProgressGrossreturns HTTP 404 on the XBRL companyconcept endpoint; the class table must be read from the filing itself or the R-files. Re-confirmed as still true this run (the FY2026 figures below were parsed from the primary document, not the API).
What the web says
Everything in this section is primary-filing evidence except the final bullet. "$M" throughout.
- Q1 answer, hard negative: the footnote is verbatim unchanged. FY2026 10-K Note 7 carries the same two footnotes as FY2025 and as the Q3 FY2026 10-Q. Footnote (2) to Construction in progress reads, in full: "Primarily includes building-related construction and tool installation." No dollar split, no sub-schedule, no roll-forward. The full class table, 2026-09-03 vs 2025-08-28: Land 420 / 420; Buildings 26,493 / 22,173; Equipment 93,954 / 79,934; Construction in progress 11,424 / 5,518; Software 1,885 / 1,651; gross 134,176 / 109,696; accumulated depreciation (70,866) / (63,106); net 63,310 / 46,590.
- Q3 answer, partial yes then hard negative. The "assets not yet in service" line exists and moved hard: footnote (1) to Equipment reads "Includes costs related to equipment not placed into service of $6.12 billion as of September 3, 2026 and $4.05 billion as of August 28, 2025" (FY2024: $3.10 billion, per the FY2025 10-K, accession 0000723125-25-000028, filed 2025-10-03). Segment detail, by contrast, is explicitly refused twice in the same document: Note 20 (Segment and Other Information) states "We do not identify or report internally our assets (other than goodwill) or capital expenditures by segment," and Item 2 repeats "We do not identify or allocate assets by operating segment, other than goodwill." There is no depreciation-commencement sentence in the PP&E policy that would constrain the split; the policy gives only straight-line lives (10-30 years buildings, 7 years production equipment, 3-5 years software).
- Q2 answer, negative on two independent grounds. Capitalized interest is disclosed separately: "Interest capitalized as part of the cost of property, plant, and equipment was $383 million, $321 million, and $225 million for 2026, 2025, and 2024" (FY2023: $208M). But the policy that governs it is class-blind by construction: "We capitalize interest on borrowings during the period of time we carry out the activities necessary to bring assets to the condition of their intended use and location." Tool installation is exactly such an activity, so the qualifying-asset base spans both legs of the contamination. And Micron does not disclose the capitalization rate numerically — "We utilize a weighted-average capitalization rate that is based on our consolidated debt portfolio" — so the relation cannot be inverted without assuming the rate.
- The arithmetic confirms the base is wider than CIP, which kills the capitalized-interest route empirically as well as textually (inference). Dividing FY2025's $321M by average CIP of $4,481M implies a capitalization rate of 7.16%. That is above the highest stated coupon on any instrument in Micron's portfolio (6.050%, the 2035 B Notes), which is a hard ceiling on any weighted average. So in FY2025 the qualifying-expenditure base was necessarily larger than CIP alone. FY2026 implies 4.52% on average CIP and 2.83% on average CIP+ENIS, bracketing a plausible rate from both sides. Capitalized interest therefore measures a base that includes tools-in-preparation, not shell. Labelled inference, built on disclosed numbers.
- The contaminant nobody had priced: PP&E is reported net of government incentives, and the netting is now large. Note 15: "Government incentives related to capital expenditures have reduced property, plant and equipment by $10.93 billion as of September 3, 2026, of which $6.43 billion pertained to 2026 expenditures" (FY2025: $5.04B cumulative, $3.11B in-year). The $24,480M gross-PP&E increase that is the denominator of
shell_shareis therefore understated by roughly $6.43B, and Note 15 says the conditions attach to "the construction of new manufacturing facilities, the purchase and installation of specialized tools and equipment" — both legs again, with no split. - The CHIPS agreements do separate the two, and are not on EDGAR. Note 15: "Funding will be based on the achievement of construction, tool installation, and wafer production milestones." Construction and tool installation are contractually distinct milestone classes with distinct disbursements. The direct funding agreements are not filed as exhibits — the FY2026 10-K exhibit index contains no funding agreement, only compensation plans and indemnification forms (exhibit index scanned this run). Total CHIPS grants up to $6.4B; a 35% federal investment tax credit (raised from 25% by the One Big Beautiful Bill Act); a non-binding New York term sheet for up to $5.5B; India/Gujarat incentives at 50% central + 20% state of project cost, ~$1.1B unrecognized.
- Geography is the one genuine, if coarse, cross-check Micron does give. Note 22, long-lived assets (net PP&E plus operating-lease ROU), 2026 vs 2025: Taiwan 27,613 / 18,965; U.S. 14,463 / 8,445; Singapore 12,345 / 10,669; Japan 7,247 / 7,038; Malaysia 1,184 / 1,124; India 609 / 449; China 473 / 544; Other 97 / 92; total 64,031 / 47,326. U.S. assets grew +$6,018M (+71%), and the U.S. is where the near-pure-shell projects sit: the Clay, New York fab reached "first concrete pour" in July 2026 with initial wafer output expected calendar 2030, and Idaho ID2 began construction in 2026 for initial output late calendar 2028.
- External benchmark, Tier C, for sanity only. Published fab-cost rules of thumb put equipment at roughly 65-80% of a greenfield fab's cost and the building at 20-35%; the Samsung Taylor disclosure of ~$6B construction against ~$11B equipment is the cleanest named datapoint (~35/65) (Quartz fab cost breakdown; Silicon Saxony). No source found any building-versus-equipment itemization for Micron's Clay or Boise projects — a targeted search of New York Green CHIPS and Onondaga County coverage returned project totals ($100B over four fabs, $51B for the first two) and a slip of first Clay output to Q3 2030, but no cost split (Yahoo Finance; NNY360 project guide).
Convergences and contradictions
- Convergence, and the direct answer: no disclosure separates shell from tool, and the three candidate routes each fail for a different, nameable reason. Capitalized interest fails because ASC 835-20 qualifying assets include equipment being prepared for use, Micron's own policy sentence says so, and the rate is unquoted. Assets-not-yet-in-service fails as a separator because it is attached to the Equipment line, not to CIP; it tells you about tools that have not entered installation, which is the complement of CIP's tool component rather than a measure of it. Segment detail fails because Micron twice states it does not allocate assets or capex to segments at all. The founder's prior from the parent brief — that CIP contamination makes the threshold a band — holds.
- Contradiction against the prior, and it is real: CIP's tool contamination is probably smaller than the parent brief assumed (inference). Micron parks tools awaiting service inside Equipment (footnote 1), not inside CIP. If tools kept their full carrying value in CIP once installation began, the ENIS disclosure would more naturally hang off CIP. The structure is weak-to-moderate evidence that CIP's "tool installation" element is installation and hook-up spend rather than the carrying value of the tools themselves, which would make CIP predominantly building. The footnote does not say this, and Micron could rebut it tomorrow. Labelled inference, medium confidence. It narrows the CIP term specifically.
- Contradiction that overwhelms the one above: a bigger contaminant was hiding in Note 15 the whole time. Even if CIP were 100% building,
shell_sharewould still be unreadable to the point, because gross PP&E is struck net of $6.43B of FY2026 government-incentive credits whose own allocation between construction and tooling is undisclosed. That single item is larger than the entire FY2026 increase in Buildings ($4,320M). Narrowing CIP therefore does not rescue the number. This is a correction to the parent brief's framing, not a confirmation of it: CIP was never the binding constraint. - Contradiction in the data itself, and it matters more than the measurement question: tripwire (2) has fired and tripwire (1) has cooled. ENIS went $4.05B → $6.12B, +51.1% YoY, through the pre-registered >50% line (FY2025 was +30.6%). Almost all of it landed in Q4: $4.20B at 2026-05-28 (Q3 FY2026 10-Q, accession 0000723125-26-000015, filed 2026-06-25, re-verified at primary source this run) to $6.12B eleven weeks later. Over the same quarter, shell spend decelerated — Q4-only implied
shell_shareis 35.3% ((Buildings +2,752 + CIP +489) / total +9,185), against 45.7% for the first nine months, and CIP added just $489M in Q4 after $5,417M in the nine months before. The full-year raw reading is 41.8%, below the nine-month figure. Tools are now stacking up and shell spend has paused — the exact inverse of the nine-month read.
Synthesis for RDCO
The direct answer is a band, and the defensible bounds are wider than the vault has been carrying. Start with the raw full-year number: Δ(Land + Buildings + CIP) = 0 + 4,320 + 5,906 = 10,226 over Δ gross PP&E of 24,480, or 41.8%. Then apply the two disclosed contaminants in the directions the disclosure permits. Tongluo ($1.80B, a wafer-fab acquisition whose own shell/tool mix is undisclosed) moves it to 37.2-45.1%. Government-incentive netting ($6.43B, allocation undisclosed) moves it to 33.1-53.9%. Taking both at their extremes gives the honest outer band: 28.9% to 57.2%. The CIP shell/tool question — the one this brief was commissioned to answer — sits inside those bounds and does not widen them further, because CIP enters the numerator whole either way; resolving it would only reallocate between Buildings and Equipment within an already-wider interval. That is the finding the queue item did not anticipate: unwinding CIP, even perfectly, would not deliver a number. A pre-registered 45% threshold cannot be adjudicated against a 28-point band, and the FY2026 reading straddles it on every variant.
So the indicator should be re-specified rather than re-measured, and the shape of the fix is already visible in the data. Three changes, in order of value. First, gross up for government incentives explicitly and report shell_share as a labelled interval with the netting assumption named, never as a point. Micron discloses the in-year incentive offset annually in Note 15, so the gross-up is mechanical; it belongs in the anchor's field list as gov_incentive_ppe_offset_fy alongside the existing columns. Second, demote shell_share from a trigger to context and promote ENIS to the primary. ENIS is a clean, single-line, uncontaminated disclosure of tools bought and not yet producing; it is not netted against incentives, not blended with buildings, and not an acquisition artifact. It is also the nearer-dated signal — crated tools are two to three quarters from bits, where a shell is three to six years from them. This run is the proof: ENIS gave an unambiguous +51.1% while shell_share gave three different answers depending on which contaminant you corrected. Third, add U.S. long-lived assets as the geographic shell proxy. Clay, New York has zero tools installed and no wafer output until calendar 2030; Idaho ID2 has none until late 2028. U.S. long-lived assets rising 71% to $14,463M is as close to a pure construction signal as Micron's disclosure gets, and it is annual, audited, and already in Note 22. This mirrors the demotion [[2026-10-06-sk-hynix-capex-intensity-reconciliation]] performed on capex/revenue: when a ratio's definitional band swamps its threshold, replace the ratio, do not keep refining it.
What would have to change for a number to exist, stated precisely, so this is never re-researched. Exactly one of four things. (a) Micron voluntarily disaggregates footnote (2) into building-related construction and tool installation — no rule compels it and the competitive incentive runs against it. (b) Micron files a PP&E movement schedule by class; IAS 16.73(e) requires one and ASC 360 does not, which is why SK Hynix under IFRS will be structurally better data than Micron the moment its first 20-F lands (~April 2027) — the vault should stop treating Micron as the reference vendor for this metric. (c) The CHIPS direct funding agreements become public with their milestone schedules; Note 15 confirms construction and tool-installation milestones are contractually distinct and separately funded, so the split exists on paper, though the agreements are not EDGAR exhibits and would have to come through a Commerce Department release or FOIA. (d) Micron begins allocating assets by segment, explicitly refused in two places in this filing, and not plausible while the CODM does not review it. Until one of those happens, the honest ceiling is a band, and the band is 29-57%, not the 38-52% the vault has been carrying.
The read on the thesis, kept narrow. Applying the pre-registered composite rule to the FY2026 print: tripwire (2) fires (+51.1%, through the >50% line); tripwire (1) does not (41.8% raw, band straddling 45%, and decelerating to 35.3% in Q4). The rule requires (1) AND (2), so no composition trigger, and no phase change. But the texture has inverted from the September read, and that is worth recording: nine months ago the finding was "dollars into concrete, tools not stacking." The full year says concrete paused in Q4 while tools began stacking, against FY2027 capex guided at ~$11.5B in Q1 and ~$25B in the first half with the second half "higher" — implying well over $50B net of incentives, roughly double FY2026's $30.71B gross. Buildings-then-tools is the normal ordering for capacity that switches on in 2027-2028, and ID1 is guided to first wafer output mid-CY2027. The composition indicator has not triggered; it has started behaving like a fab that is about to turn on.
Why this is in the vault
This closes open follow-up #5 of [[2026-09-06-vendor-capex-composition-memory-anchor]] with a negative answer at primary source, and in doing so corrects that brief's diagnosis: CIP contamination is real but is not the binding precision limit on shell_share — undisclosed allocation of government-incentive PP&E offsets is, and it is larger. It therefore changes what gets built in anchors/memory-vendor-capex/: add a gov_incentive_ppe_offset_fy field and a gross-up interval, demote shell_share from trigger to context, promote equipment_not_placed_into_service to the primary supply-discipline gauge, and add U.S. long-lived assets as the geographic shell proxy. It also records that tripwire (2) crossed its pre-registered >50% line on the FY2026 print while tripwire (1) did not, which the composite rule resolves to no trigger.
Open follow-ups
- Backfill ENIS across the 2017-18 and 2021-22 cycles and measure its lead on the ASP rollover. The >50% line has now fired once and has never been tested against a cycle turn. This is the same n=0 weakness the parent brief flagged for
shell_share, and ENIS is now the metric carrying the weight, so the backfill matters more than it did. - Check whether the CHIPS Act direct funding agreements, with their construction / tool-installation / wafer-production milestone schedules, are obtainable from the Department of Commerce. Note 15 confirms the split exists contractually and is separately funded; the agreements are not EDGAR exhibits. If a public or FOIA-able version carries dollar amounts per milestone class, it is the only known path to an actual number.
Related
- [[2026-09-06-vendor-capex-composition-memory-anchor]] — the parent brief; source of the
shell_sharemetric, the 45% threshold, and the follow-up this brief closes and partially corrects - [[2026-10-01-micron-fq4-fy2026-dram-hbm-tripwire-rescan]] — the re-scan that re-priced this item to High on the FY2027 capex doubling, now confirmed at primary source
- [[2026-08-01-hyperscaler-capex-merchant-vs-inhouse-silicon-decomposition]] — the ASC 360 "major class" disclosure floor reused here, and the never-quote-a-proxy-as-a-decomposition discipline
- [[2026-10-06-sk-hynix-capex-intensity-reconciliation]] — the precedent for demoting a ratio whose definitional band swamps its threshold
- [[2026-07-07-dram-hbm-phase2-phase3-early-signals]] — the five-row tripwire scoreboard this indicator feeds
- [[2026-06-28-chip-memory-cycle-phase2-phase3-indicators]] — the original threshold set and the "watch composition, not level" doctrine
- [[2026-09-15-dram-greenfield-fab-online-dates]] — the per-fab online register that dates Clay to 2030 and Idaho ID2 to late 2028
- [[2026-10-06-memory-fab-subsidy-share-2027-28-wave]] — the subsidy-insulated-capacity sizing that the $6.43B incentive offset bears directly on
- [[2026-05-18-memory-cycle-v1.1]] — the live thesis whose anchor set this re-specifies
Sources
Vault
~/rdco-vault/06-reference/research/2026-09-06-vendor-capex-composition-memory-anchor.md~/rdco-vault/06-reference/research/2026-10-01-micron-fq4-fy2026-dram-hbm-tripwire-rescan.md~/rdco-vault/06-reference/research/2026-08-01-hyperscaler-capex-merchant-vs-inhouse-silicon-decomposition.md~/rdco-vault/06-reference/research/2026-10-06-sk-hynix-capex-intensity-reconciliation.md~/rdco-vault/06-reference/research/2026-10-06-memory-fab-subsidy-share-2027-28-wave.md
Primary (Tier A — retrieved directly from SEC EDGAR, 2026-10-14)
- Micron 10-K FY2026, CIK 0000723125, accession 0000723125-26-000023, period 2026-09-03, filed 2026-10-09 — Note 7 PP&E class table and footnotes (1) and (2); Note 9 Debt; Note 15 Government Incentives; Note 20 Segment and Other Information; Note 22 Geographic Information; Item 2 Properties; MD&A capex guidance; exhibit index. https://www.sec.gov/Archives/edgar/data/723125/000072312526000023/mu-20260903.htm
- Micron 10-K FY2025, accession 0000723125-25-000028, filed 2025-10-03 — FY2024 ENIS ($3.10B), FY2023 capitalized interest ($208M), FY2025 incentive offset ($5.04B cumulative / $3.11B in-year)
- Micron 10-Q Q3 FY2026, accession 0000723125-26-000015, period 2026-05-28, filed 2026-06-25 — PP&E class table re-verified at primary source this run for the Q4-only derivation
- SEC submissions API, CIK 0000723125 — gate confirmation that the FY2026 10-K is on file
Web (Tier C — benchmark and negative-result context only; no load-bearing number depends on these)
- Semiconductor fab cost breakdown — Quartz
- Why are semiconductor factories so expensive — Silicon Saxony
- Micron delays construction at $100B New York megafab — Yahoo Finance
- Essential guide to Micron's chip project: timeline, hiring, construction and costs — NNY360
Searched and not found (a finding, not a gap in effort)
- No building-versus-equipment cost itemization for Micron's Clay, New York or Boise, Idaho projects in New York Green CHIPS coverage, Onondaga County reporting, or CHIPS award documentation reachable by search.
- No CHIPS direct funding agreement filed as an exhibit to any Micron filing — the FY2026 10-K exhibit index carries only compensation plans and indemnification forms.