Five-year take-or-pay memory agreements exist, pricing is dominantly a floor-and-ceiling band, a fixed-price subset exists, and no filing ties a fixed-price agreement to HBM or a hyperscaler
The question
Verbatim: "Does a 3+ year fixed-price hyperscaler HBM offtake agreement actually exist as of September 2026? Resolve from primary sources (10-Q/10-K purchase-commitment disclosures, earnings-call transcripts), not trade press."
Context: this is open follow-up #2 of [[2026-09-09-hbm-plateau-hypothesis-backtest-variant]], which named it as the single fact that most moves the plateau prior. The vault carried an unreconciled three-way split: [[2026-05-18-hbm3e-hbm4-capacity-timeline-phase-b-end]] said a 3+ year fixed-price hyperscaler offtake "has not happened," [[2026-05-18-memory-cycle-v1.1-v2-honest-rerun]] cited "multi-year offtakes" as plateau evidence, and trade press claimed five-year long-term agreements (LTAs). This brief resolves what filings and company-published management remarks actually say, and marks precisely where they say nothing.
Date-framing note. The question was filed 2026-09-09 and framed "as of September 2026." The disclosure that resolves the structure and term conjuncts predates it: Micron's FQ3 FY2026 10-Q (filed 2026-06-25) and FQ3 prepared remarks (2026-06-24). Three sources postdate the framing and are used as confirmation and extension: Micron's FQ4 prepared remarks (2026-09-30), Micron's FY2026 10-K (filed 2026-10-09), and SK hynix's Q3 2026 results release (2026-10-28). So the vault's September contradiction was resolvable at the time on structure and term; nobody had read the 10-Q. The counterparty conjunct was not resolvable in September: the first category-level naming of cloud service providers anywhere in the primary record is SK hynix on 2026-10-28, and SK hynix's prior release (2026-07-29) said only "around 10 customers, including key strategic partners," with no five-year term.
What we already know (from the vault)
- The capitulation test was written as a fixed-price test. [[2026-05-18-hbm3e-hbm4-capacity-timeline-phase-b-end]]: "The single fact that would force capitulation to the bull case (long-dated): any hyperscaler signing a 3+ year HBM offtake at fixed price. That has not happened." It was written against a structure the industry then largely declined to use, which is why it needs amendment rather than a verdict.
- The plateau variant's exit legs were already found non-functional, and its evidentiary weak point was named as this exact leg ([[2026-09-09-hbm-plateau-hypothesis-backtest-variant]]): "the plateau case leans hardest on the leg with the weakest evidentiary chain."
- Tripwire #1 (ASP deceleration) has fired once and its basket was audited. [[2026-10-01-micron-fq4-fy2026-dram-hbm-tripwire-rescan]] scored it fired; [[2026-10-03-dram-asp-series-basket-consistency-tripwire-1]] cleared it as a real firing on a consistent conventional-DRAM basket, and incidentally surfaced a TrendForce release titled "Long-Term Agreements Cap Price Increases" without pursuing what the LTAs actually were.
- Supply-side Phase-3 triggers remain cold and late-dated. [[2026-09-15-dram-greenfield-fab-online-dates]] puts the first real greenfield step at H2 2027; [[2026-10-06-memory-fab-subsidy-share-2027-28-wave]] re-dates the subsidy-insulated tail to 2029-31; [[2026-09-08-cxmt-ymtc-bit-share-phase3-threshold]] finds no reachable CXMT DRAM bit share triggers Phase 3 against strong demand.
- The vault has never had a primary-source contract-structure anchor. Every existing anchor is physical (inventory days, capex composition, fab online dates, bit share) or price-series based. Contract architecture was an assumed input, not a measured one.
What the web says
Primary sources only. The Micron 10-K quotes, the Micron 60%-concentration figure and the Nvidia Note 10 quote were each re-verified directly against the raw filings by the synthesising agent rather than accepted on an extraction report.
- Micron discloses take-or-pay multi-year customer agreements whose majority pricing is banded. FY2026 10-K (filed 2026-10-09, fiscal year ended 2026-09-03), Item 1 Business "Marketing and Customers," repeated in Note 1 and Note 16: "We also have strategic customer agreements structured as take-or-pay agreements, with binding commitments for specific contractually enforceable volumes over the multi-year contract terms. Pricing for our contracts is either fixed or periodically negotiated, with the majority of the strategic customer agreements having pricing that is subject to minimum and maximum bands. ... Under the strategic agreements, our customers have provided financial security in the form of cash deposits and letters of credit." Note the phrase "either fixed or periodically negotiated": the filing discloses that a fixed-price subset exists without sizing it. The exact string "fixed price" appears nowhere; the term of art is "minimum and maximum bands." Item 1A risk-factors these agreements explicitly, warning they "could also constrain our available supply and limit our flexibility to respond to changes in market conditions."
- The quantum is large and newly disclosed. FY2026 10-K Note 16: remaining performance obligations (RPO) "approximately $134 billion, of which $13 billion has been recognized as contract liabilities ... Nearly all of the deposits are scheduled to be repaid between 2029 and 2031. ... The remaining performance obligations are expected to be fully recognized within the next five years. As of August 28, 2025, our remaining performance obligations were not material." Balance sheet: noncurrent customer contract liabilities $12,895M versus $142M a year earlier; cash deposits received in FY2026 were $12.75B, reported as a financing inflow; $7B of customer letters of credit, accessible only on "specified events of default or breach." Both Micron sources state the RPO basis: it is computed on committed volumes and minimum pricing and excludes agreements lacking a determined pricing framework, so it is a floor-case figure rather than an expected-revenue figure.
- The term is five years, from management's own mouth. Micron FQ3 FY2026 prepared remarks (company-published PDF, 2026-06-24), Sanjay Mehrotra: "Typically, these agreements have a five-year term, from calendar 2026 through the end of calendar 2030. Automotive agreements generally have a three-year term. The 16 signed agreements represent roughly 20% of our DRAM volume and a third of our NAND volume over this period. These SCAs include four very large customers and three medium-sized customers." FQ4 prepared remarks (2026-09-30): 26 agreements signed, "over 35% of our revenue through 2030," some now extending into 2031, and RPO "approximately $150 billion" stated "to date" as of 2026-09-30 (Mark Murphy, CFO), against the 10-K's $134B as of the 2026-09-03 balance-sheet date. Management does not explain the gap between the two figures; the measurement-date difference plausibly reflects agreements signed after fiscal year-end, but that is this brief's inference, not a disclosed reconciliation.
- Micron's own pricing split is quantified only partially and only forward-looking, and the product scope is memory-wide. FQ4 prepared remarks: "Three-quarters of this estimated revenue has a defined pricing framework, a majority of which have pricing bands with floor and ceiling prices. The remaining quarter of this SCA revenue expectation has pricing negotiated periodically based on market prices." FQ3 prepared remarks give the ceiling's anchor and the fixed-price residual: "The largest agreements generally have a ceiling price for existing products at the current CQ2 market price, and a floor price through the term of the agreement. Several SCAs, which account for a modest portion of the SCA-related revenue, include either fixed prices or have no price bands associated with them where pricing will be subject to market conditions." Also FQ3, forward-looking: "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" — a figure that deliberately merges fixed with ceilinged and therefore cannot size the fixed-price slice. On scope, Mehrotra, FQ3: the agreements "provide committed DRAM, including HBM as appropriate, and NAND supply to our customers over a multi-year time horizon" — so HBM can sit inside the five-year agreements, but no source quantifies how much does. Separately, FQ4: "We have completed agreements for the vast majority of our calendar 2027 HBM bit supply with significant price increases year over year." That sentence states which calendar-2027 supply is covered; it does not state how long those agreements run, how their prices are struck, or whether they sit inside the strategic agreements. Management also never uses the words "sold out" in any FQ3 or FQ4 press release, prepared-remarks PDF or earnings deck.
- SK hynix corroborates the five-year term, describes range pricing, and is silent on fixed pricing. Q3 2026 results release (news.skhynix.com, 2026-10-28): "The company has currently signed LTAs with 14 major customers, including leading global cloud service providers (CSPs) ... which are typically five years"; coverage is "more than 50% of total production capacity for 2027 and 2028"; the mechanism is "by introducing a mechanism that reflects market conditions within a price range agreed upon with customers ... while allowing flexible responses to market changes." The release never mentions fixed pricing, take-or-pay or customer deposits, so it neither confirms nor denies them. SK hynix became SEC-registered in July 2026 (CIK 0002120882, Nasdaq: SKHY); its 424B4 prospectus (2026-07-10) still carries the pre-LTA risk factor "Specific quantities and pricing are typically determined through mutual agreement at the time of purchase," and its last balance sheet (2026-06-30) shows zero noncurrent customer contract liabilities and no RPO disclosure.
- Samsung confirms multi-year memory agreements exist and withholds every term. 2026 Half-year Business Report (English, company-published), Note 13(C): "the Company has entered into multi-year agreements with multiple global customers for the supply of memory semiconductor products. The total order value, delivery schedules, and other contractual terms are subject to change based on customer demand and business developments." The long-term-contracts section adds: "Specific terms and conditions of these agreements have been omitted as they constitute confidential business information." Its only quantified long-dated contract is foundry, not memory: Tesla, 2025-07-26 to 2033-12-31, US$16,544M.
- No hyperscaler discloses a memory commitment, and Nvidia's memory commitment is explicitly adjustable. Across 14 filings (Microsoft, Alphabet, Amazon, Meta most recent 10-K plus every subsequent 10-Q through 2026-11-01), the terms HBM, high-bandwidth memory, DRAM, NAND, offtake and fixed price return zero hits; the sole "memory" occurrence is Amazon boilerplate flagging "resource and supply volatility, including for memory chips" as a risk to results, with no commitment disclosed against it. Nvidia Q2 FY2027 10-Q (filed 2026-08-26), Note 10 Commitments and Contingencies, immediately below the commitments table: supply commitments rose "from $119 billion last quarter to $279 billion as of July 26, 2026. These supply commitments are for our data center infrastructure systems, primarily memory and manufacturing facilities ... in certain instances, these agreements may be cancelable, rescheduled, or adjustable for our business needs prior to placing firm orders." The table concentrates $267B of the $279B in FY2027-FY2029. No price term appears, the FY2026 10-K says long-term supply agreements "have increased our product costs," and the Q1 FY2027 10-Q carries an unqualified "Supply availability affecting memory, and other components, as well as rising prices, may drive the prices for data center buildouts higher."
Convergences and contradictions
- Converge: the floor-and-ceiling band is the pricing form that actually got used, on both the sell side and, for unspecified components, the buy side. Micron sells under "minimum and maximum bands"; SK hynix sells within "a price range agreed upon with customers"; and Alphabet buys under long-term supply agreements running "through 2032" where "Certain of our long-term supply agreements contain fixed volume commitments with variable pricing between minimum and maximum rates and/or the ability to partially reduce committed volumes" (Q3 2026 10-Q, Note 10). Two caveats keep this from being stronger than a pattern. Alphabet's filing covers "technical infrastructure and inventory components" and never mentions memory, so it is not memory evidence. And these parties are not independent observations: Alphabet's banded agreements could be the very contracts Micron or SK hynix are reporting from the other side.
- Contradict: the trade-press LTA claim was right on duration, overstated on "sold out," and silent where it mattered most. The five-year duration in the TweakTown and Macrostream citations carried by [[2026-09-09-hbm-plateau-hypothesis-backtest-variant]] is now corroborated by Micron and SK hynix management directly, so the vault should stop treating duration as a trade-press-only claim. The "sold out for all of 2027" framing is not supported: the nearest primary wording is "the vast majority of our calendar 2027 HBM bit supply." And no trade-press source characterised the pricing mechanism at all, which is the conjunct that decides the thesis.
- Contradict and unresolved: both vault documents need amendment, and the counterparty question cannot be closed from primary sources. [[2026-05-18-hbm3e-hbm4-capacity-timeline-phase-b-end]]'s capitulation test demanded a fixed price and therefore could not fire in the state it was meant to detect, which is the same defect [[2026-09-09-hbm-plateau-hypothesis-backtest-variant]] diagnosed in the plateau variant's exit legs; [[2026-05-18-memory-cycle-v1.1-v2-honest-rerun]]'s "multi-year offtakes" phrase turns out to be corroborated and should be annotated with its now-known source rather than retired. On counterparties: SK hynix names the category itself ("leading global cloud service providers"), but Micron names nobody. No counterparty appears in any Micron filing, no agreement is filed as a material-contract exhibit, and Note 21 reports "No customer accounted for 10% or more of total revenue in 2026." Micron's "four very large customers" against "approximately 60% of our total revenue ... concentrated in the data center end market" (FY2026 10-K, Item 1A) makes hyperscalers the obvious inference. It remains an inference, and the vault should not record it as disclosed fact.
Synthesis for RDCO
Verdict on each conjunct, separately. (a) Does a long-term agreement exist? Yes, confirmed, high confidence. Micron's FY2026 10-K discloses take-or-pay agreements with "binding commitments for specific contractually enforceable volumes," backed by $12.75B of customer cash deposits received in FY2026 and $7B of third-party letters of credit, with $134B of RPO where a year earlier RPO was "not material." That is an audited balance sheet with a new line item, not a press release. SK hynix and Samsung independently confirm multi-year memory agreements in their own primary disclosures. (b) Is the term 3+ years? Yes, confirmed, high confidence. Micron management: "typically ... a five-year term, from calendar 2026 through the end of calendar 2030," automotive at three years, some extending into 2031. SK hynix: "typically five years." The 10-K corroborates independently, since RPO is "expected to be fully recognized within the next five years" and deposits are "scheduled to be repaid between 2029 and 2031." (c) Is the price fixed? Split answer. (c1) The dominant structure is a band, not a fixed price: high confidence. The majority of Micron's agreements price within minimum and maximum bands, and SK hynix prices within an agreed range that tracks the market. (c2) A fixed-price subset is disclosed to exist, and is not sizeable from any primary source: existence moderate-to-high confidence, magnitude unknown. The confidence is not higher because every source sentence names fixed pricing only inside an "either fixed or banded/negotiated" disjunction rather than describing a fixed-price agreement on its own; the strongest single wording is the 10-K's RPO exclusion sentence, which treats "fixed pricing" as its own category. Micron's filing says pricing is "either fixed or periodically negotiated"; the FQ3 10-Q says "Pricing for most agreements is either fixed, or is subject to minimum and maximum pricing"; FQ4 says RPO covers agreements whose pricing "can be either a fixed price or subject to a pricing floor and ceiling." The only bound available is FQ3's "a modest portion of the SCA-related revenue," and even that merges fixed-price agreements with no-band agreements. So the full question, read as a conjunction, is not confirmable from primary sources and is also not refuted. No source ties a fixed-price agreement to HBM specifically, or to a named or category-identified hyperscaler counterparty, and two further conjuncts the question assumed are themselves unresolved: whether and how much HBM volume sits inside any strategic agreement is not disclosed, and the calendar-2027 HBM book's term and pricing form are not disclosed either. "The filings do not disclose this" is the honest answer to the fixed-price-HBM-hyperscaler conjunction, and it is not the same as "it does not exist."
What this does to the plateau prior: it moves it up, modestly, for a mechanism the plateau thesis never named, and it truncates the thesis's upside at the same time. The plateau case as carried through four vault documents rested on demand being contractually locked at a fixed price. It is not locked that way. What exists instead is better-targeted than the thesis asked for and smaller in scope: a price floor on a growing share of revenue, with real enforcement behind it. Micron's statement that "We expect gross margins from our strategic customer agreements with price bands, even at floor pricing levels, to yield gross margins meaningfully above our peak quarterly margins in any past cycle" (FY2026 10-K, Item 7 MD&A; the FQ4 remarks carry a similar sentence in Murphy's voice) is the load-bearing economic claim in this brief, because the historical memory-cycle crash is an ASP and margin event and a contractual floor is precisely a left-tail truncation on ASP. It is also a management expectation, not an audited outcome, and it has never been tested through a demand deceleration. Deposits and letters of credit make the floor collectible rather than aspirational. But the same instrument caps the right tail: ceilings anchored at calendar-Q2-2026 market prices mean the covered share cannot participate in further ASP escalation. The honest characterisation is therefore variance reduction, not direction, on a bounded and disclosed share: over 35% of Micron revenue through 2030, over 50% of SK hynix 2027-28 capacity. The next downturn looks shallower and later for the covered portion and fully spot-exposed for the rest, which is a materially different claim from "the cycle is broken." I would not attach a numeric probability shift to this. The defensible statement is that the plateau prior moves from "resting on an unsupported claim" to "resting on a partially supported, bounded and as-yet-untested mechanism," while the uncovered residual still carries the full historical cycle risk. The 2029-2031 deposit-repayment window is now a datable cash event, and it coincides with the subsidy-insulated capacity tail that [[2026-10-06-memory-fab-subsidy-share-2027-28-wave]] re-dated to 2029-31.
The tripwire consequence is a measurement problem and it needs an amendment. Tripwire #1 tracks DRAM contract-ASP deceleration off third-party series, and [[2026-10-03-dram-asp-series-basket-consistency-tripwire-1]] did the work of pinning it to one consistent conventional-DRAM basket. Band pricing breaks that series' relationship to realized revenue going forward: as agreement coverage grows, the published contract-price series increasingly measures the uncovered residual rather than blended realized ASP, and it will overstate both the rise and the eventual fall. The TrendForce headline the vault already had in hand, "Long-Term Agreements Cap Price Increases," was reporting exactly this and the vault read past it. The amendment is not to retire Tripwire #1 but to re-key it: a residual contract-price decline is now necessary but no longer sufficient for the ASP leg of a Phase-3 call, and the sufficiency condition should be a decline in blended realized ASP, which is computable from Micron's own revenue and bit-shipment disclosures. No other v1.1 tripwire changes, since inventory days, capex composition, greenfield online dates and bit share are all physical and indifferent to contract form. The 05-18 bull-capitulation test should be amended rather than deleted, and the candidate replacement is a three-part condition: banded-floor coverage above a named share of maker revenue, and collectible security behind it (deposits plus letters of credit, as Micron now discloses and SK hynix does not), and disclosure tying HBM volume specifically into those agreements. Noting honestly that on this brief's own reasoning a banded floor with a ceiling may not deserve to satisfy a bull capitulation test at all, since it caps upside as well as downside; that is a judgment for the quarterly review, not something this brief should settle.
The methodological finding is that contract architecture is now a disclosed, machine-readable quarterly series, and the vault should treat it as a new anchor rather than an anecdote. Four of these did not exist a year ago and are now recurring line items: Micron's RPO, its noncurrent customer contract liabilities, its agreement count and revenue-coverage percentage, and its letters-of-credit total. The natural cross-maker test is whether SK hynix's LTAs carry the enforceability Micron's do, but the timing of that test needs care: the RPO disclosure requirement in IFRS 15 is an annual financial-statement requirement, and quarterly reporting under IAS 34 and K-IFRS 1034 generally does not compel it. So absence of an RPO disclosure in SK hynix's 2026-11-14 quarterly business report would be close to uninformative, and the real tests are its FY2026 annual business report (approximately March 2027) and its first Form 20-F (approximately April 2027). What the November quarterly can show is the deposit and contract-liability balances, which speak to customer financial security rather than to price terms. One standing caution on the maker comparison: Micron is a US registrant under full SEC disclosure obligation while Samsung is not and SK hynix only became one in July 2026, so part of the disclosure gap is regulatory rather than commercial, and the 2027 annual filings are what separate the two.
Why this is in the vault
It closes open follow-up #2 of [[2026-09-09-hbm-plateau-hypothesis-backtest-variant]], amends the "no 3+ year fixed-price offtake" capitulation test in [[2026-05-18-hbm3e-hbm4-capacity-timeline-phase-b-end]] as mis-specified rather than merely stale, and supplies the Tripwire #1 re-keying that the memory-cycle v1.1 anchor set needs before the next quarterly review can read a contract-price series as a Phase-3 input.
Open follow-ups
- Compute Micron's blended realized DRAM ASP from its own disclosures and compare it to the TrendForce conventional-DRAM contract series over the same quarters. This is the test of how far the published series has already diverged from realized pricing, and the quantitative half of the Tripwire #1 re-keying. Micron reports revenue and bit-shipment growth by technology quarterly, so the series is constructible from primary sources.
- Read SK hynix's FY2026 annual business report (approximately March 2027) and first Form 20-F (approximately April 2027) for a remaining-performance-obligation disclosure and any non-current customer contract liability. Annual filings are where IFRS 15's RPO requirement actually binds, so this is the first clean test of whether the five-year LTAs have banded pricing in the accounting sense. The 2026-11-14 quarterly report is worth reading for deposit balances only.
- Establish whether the fixed-price-only share of Micron's agreements is ever separately disclosed. Management has only ever reported fixed and ceilinged together. If a future filing or remark splits them, conjunct (c2) becomes answerable with a magnitude instead of an existence statement.
Related
- [[2026-09-09-hbm-plateau-hypothesis-backtest-variant]]
- [[2026-05-18-hbm3e-hbm4-capacity-timeline-phase-b-end]]
- [[2026-05-18-memory-cycle-v1.1]]
- [[2026-05-18-memory-cycle-v1.1-v2-honest-rerun]]
- [[2026-10-03-dram-asp-series-basket-consistency-tripwire-1]]
- [[2026-10-01-micron-fq4-fy2026-dram-hbm-tripwire-rescan]]
- [[2026-09-25-hbm-dram-phase-2-3-transition-q3-2026]]
- [[2026-09-08-cxmt-ymtc-bit-share-phase3-threshold]]
- [[2026-09-15-dram-greenfield-fab-online-dates]]
- [[2026-10-06-memory-fab-subsidy-share-2027-28-wave]]
- [[2026-09-07-memory-supplier-inventory-days-dio-anchor]]
- [[2026-09-06-vendor-capex-composition-memory-anchor]]
- [[2026-06-28-chip-memory-cycle-phase2-phase3-indicators]]
- [[2026-07-07-dram-hbm-phase2-phase3-early-signals]]
- [[2026-09-26-financing-led-capital-cycle-phase3-precedent]]
Sources
Research notes trail: every source consulted, with reachability
Micron Technology (CIK 0000723125), SEC filings via EDGAR, curl with declared user agent, all HTTP 200
- FY2026 10-K, accession 0000723125-26-000023, filed 2026-10-09, FY ended 2026-09-03. https://www.sec.gov/Archives/edgar/data/723125/000072312526000023/mu-20260903.htm — the primary document. Item 1 "Marketing and Customers" (take-or-pay, bands, financial security); Item 1A (the strategic-agreement risk paragraph and the 60% data-center concentration figure, printed p. 33); Item 7 MD&A "Strategic Customer Agreements" (floor-pricing margin expectation) and "Liquidity" ($12.75B deposit inflow); Note 1 (deposit return mechanics); Note 10 Commitments ($10.8B, Micron as buyer — not customer commitments); Note 16 ($134B RPO, $13B contract liabilities, 2029-31 repayment, $7B letters of credit, five-year recognition, minimum-pricing basis); Note 21 (no customer at 10%). Load-bearing quotes re-verified against the raw filing.
- FQ3 FY2026 10-Q, accession 0000723125-26-000015, filed 2026-06-25, period 2026-05-28. https://www.sec.gov/Archives/edgar/data/723125/000072312526000015/mu-20260528.htm — first appearance in the filing record; carries "Pricing for most agreements is either fixed, or is subject to minimum and maximum pricing," the CQ2-2026 ceiling anchor, and the $22B/$18B expected-deposit figures, all three dropped from the 10-K. RPO then approximately $5B.
- FY2025 10-K, accession 0000723125-25-000028, filed 2025-10-03. https://www.sec.gov/Archives/edgar/data/723125/000072312525000028/mu-20250828.htm — the baseline diff and the direct contradiction: "Due to volatile industry conditions, our customers are generally reluctant to enter into long-term, fixed-price purchase contracts." Deleted from FY2026. Zero hits for take-or-pay, strategic customer, customer deposit, price band, letters of credit.
- 8-K earnings releases: 0000723125-26-000018 (2026-09-30, FQ4) and 0000723125-26-000013 (2026-06-24, FQ3), EX-99.1 both reachable.
Micron company-published management remarks (investor-relations CDN, curl, HTTP 200)
- FQ3 FY2026 Prepared Remarks, 2026-06-24. https://s25.q4cdn.com/621799436/files/doc_financials/2026/q3/Q3-FY26-Prepared-Remarks.pdf — five-year term, 16 agreements, the 20%-DRAM/one-third-NAND volume split (as of 16 agreements, superseded in scope by FQ4's 26), "including HBM as appropriate," ceiling/floor mechanics, the "modest portion" fixed-price residual, the forward-looking approximately-40% figure, four very large / three medium customers.
- FQ4 FY2026 Prepared Remarks, 2026-09-30. https://s25.q4cdn.com/621799436/files/doc_financials/2026/q4/Q4-FY26-Prepared-Remarks.pdf — 26 agreements, over 35% of revenue through 2030, three-quarters/one-quarter pricing split, extension into 2031, calendar-2027 HBM bit supply, approximately $150B RPO on a to-date basis.
- Q3 and Q4 FY2026 earnings decks, same CDN path, reachable; carry dedicated "Strategic Customer Agreements" slides.
- investors.micron.com returned HTTP 403 to every curl user agent tried; Playwright cleared it and was used only for link discovery.
- Not available: Micron publishes no official Q&A transcript. The only Q&A wording is third-party (Motley Fool, which discloses possible inaccuracy and LLM involvement). No Q&A wording is relied on anywhere in this brief, including on the counterparty question, which rests instead on the filings naming no counterparty, filing no exhibit, and Note 21's 10% statement. No paywall was encountered or circumvented.
SK hynix (CIK 0002120882, Nasdaq: SKHY), SEC-registered since July 2026
- Form 424B4 IPO prospectus, 2026-07-10. https://www.sec.gov/Archives/edgar/data/2120882/000119312526299963/d32785d424b4.htm — zero hits for take-or-pay, price floor, fixed price, long-term agreement, customer deposit, RPO, offtake, minimum purchase. The four "multi-year supply agreements" hits run the opposite direction (SK hynix as buyer of raw materials). Risk-factor quote re-verified against raw HTML.
- Form 6-K, 2026-08-18. https://www.sec.gov/Archives/edgar/data/2120882/000119312526354777/d147827d6k.htm — English translation of the Korean semi-annual business report for H1 2026. Section 7 "Major Contracts" lists only a Rambus cross-license and the Intel NAND acquisition; Note 29 Commitments has no customer-side commitment; noncurrent customer contract liabilities zero at all four balance-sheet dates through 2026-06-30. Its own cover states the translation was "reformatted, rearranged or summarized," so absence here is weaker evidence than absence in the Korean original.
- All 30 Forms 6-K plus one 6-K/A through 2026-10-29 swept: zero hits on the full term list. The Q3 2026 earnings 6-K (2026-10-28) is a numbers-only table.
- Q3 2026 results release, 2026-10-28. https://news.skhynix.com/en/q3-2026-business-results/ — 14 LTA customers "including leading global cloud service providers (CSPs)," "typically five years," over 50% of 2027-28 capacity, "a price range agreed upon with customers." Zero hits for "fixed."
- Q2 2026 results release, 2026-07-29. https://news.skhynix.com/en/q2-2026-business-results/ — "around 10 customers, including key strategic partners"; no CSP wording, no five-year term, no price-mechanism language. Establishes that the CSP naming is a late-October event.
Samsung Electronics (CIK 0000879316 exists but has filed no periodic report since 2015; a DART-only issuer)
- 2026 Half-year Business Report, English, company-published, 323 pp. https://images.samsung.com/is/content/samsung/assets/global/ir/docs/2026_Half_Interim_Report.pdf — Note 13(C) (multi-year memory agreements, "subject to change"); Section 4F "Long-term contracts" (terms withheld as confidential; Tesla foundry contract to 2033-12-31, US$16,544M). Zero hits for take-or-pay, price floor, fixed price, long-term agreement, RPO, minimum purchase, letter of credit. Advances received US$2.224B at 2026-06-30 (+70% in six months); consolidated contract liabilities KRW 23,592,176M (+68.8%), but the note's own footnote makes that a broad aggregate ("includes advances received, accrued expenses, other current liabilities and others") never linked to the memory agreements.
- Q3 2026 earnings conference presentation, English. https://images.samsung.com/is/content/samsung/assets/global/ir/docs/2026_3Q_conference_eng.pdf — the only long-term-contract mention is a forward intent bullet with no term, volume or pricing. "Unearned revenue & other advances" fell from KRW 4,466.5bn (2026-06-30) to KRW 3,810.2bn (2026-09-30).
Nvidia (CIK 0001045810), the intermediary test, all HTTP 200
- Q2 FY2027 10-Q, filed 2026-08-26. https://www.sec.gov/Archives/edgar/data/1045810/000104581026000075/nvda-20260726.htm — the $279B "primarily memory" commitment and the cancelable/reschedulable/adjustable characterization, both in Note 10 Commitments and Contingencies directly below the maturity table (attribution verified against the raw filing, not MD&A).
- FY2026 10-K, filed 2026-02-25. https://www.sec.gov/Archives/edgar/data/1045810/000104581026000021/nvda-20260125.htm — "long-term supply agreements and capacity commitments, which have increased our product costs"; customers "can generally cancel, change, or delay product purchase commitments with little notice ... and without penalty."
- Q1 FY2027 10-Q, filed 2026-05-20. https://www.sec.gov/Archives/edgar/data/1045810/000104581026000052/nvda-20260426.htm — the rising-memory-price risk factor.
- Zero hits across all three for HBM, high-bandwidth memory, DRAM (the single apparent hit is a substring of "dramatically"), offtake, take-or-pay, fixed price, price protection, volume commitment, backlog. Micron, SK Hynix and Samsung are each named once or twice in the 10-K's Item 1 "Manufacturing" with no contract terms attached, and are absent from both 10-Qs including the one disclosing the memory-centric commitment.
Hyperscaler demand-side test, 14 filings, all HTTP 200, all zero-hit on memory commitments
Microsoft (CIK 0000789019): 10-K 2026-07-29, 10-Q 2026-10-28. Alphabet (CIK 0001652044): 10-K 2026-02-05; 10-Qs 2026-04-30, 2026-07-23, 2026-10-29 (the Q3 filing carries the Note 10 banded-pricing language quoted above: https://www.sec.gov/Archives/edgar/data/1652044/000165204426000084/goog-20260930.htm). Amazon (CIK 0001018724): 10-K 2026-02-06; 10-Qs 2026-04-30, 2026-07-31, 2026-10-30; plus the FY2024 10-K and Q3 2025 10-Q as baselines establishing that the "memory chips" phrase is new. Meta (CIK 0001326801): 10-K 2026-01-29; 10-Qs 2026-04-30, 2026-07-30, 2026-10-22. Amazon's affirmative disclaimer, 10-K Item 1A: "We do not have long-term arrangements with most of our suppliers to guarantee availability of merchandise, content, components, or services."
Not reachable
- DART (dart.fss.or.kr) original Korean filings were not read directly; englishdart.fss.or.kr returned HTTP 302 to a zero-byte response, confirming thin English coverage. The DART open API requires a key we do not hold and was not attempted. All Korean-statutory content here comes from the issuers' own English translations, which is a real limitation, flagged above.
- SK hynix and Samsung Q3 2026 Korean business reports do not exist yet (both due approximately 2026-11-14). SK hynix's first Form 20-F is not due until approximately April 2027.
- SK hynix does not furnish its earnings presentation or call transcript to the SEC, and no company-published transcript was located; the newsroom release carried the needed pricing language.
Vault
~/rdco-vault/06-reference/research/2026-09-09-hbm-plateau-hypothesis-backtest-variant.md— the parent; open follow-up #2 is this question verbatim~/rdco-vault/06-reference/research/2026-05-18-hbm3e-hbm4-capacity-timeline-phase-b-end.md— the capitulation test amended here~/rdco-vault/06-reference/research/2026-10-03-dram-asp-series-basket-consistency-tripwire-1.md— the consistent conventional-DRAM basket, and the unpursued "Long-Term Agreements Cap Price Increases" headline~/rdco-vault/06-reference/research/2026-10-01-micron-fq4-fy2026-dram-hbm-tripwire-rescan.md— Tripwire #1 scored as fired~/rdco-vault/06-reference/research/2026-09-25-hbm-dram-phase-2-3-transition-q3-2026.md— the five-row tripwire scoreboard this brief amends~/rdco-vault/06-reference/research/2026-10-06-memory-fab-subsidy-share-2027-28-wave.md— the 2029-31 subsidy tail coinciding with the deposit-repayment window~/rdco-vault/06-reference/research/2026-09-08-cxmt-ymtc-bit-share-phase3-threshold.md— the China-entrant leg~/rdco-vault/06-reference/research/2026-09-15-dram-greenfield-fab-online-dates.md— supply-side Phase-3 staircase~/rdco-vault/06-reference/research/2026-09-26-financing-led-capital-cycle-phase3-precedent.md— the financing-led-turn precedent work~/rdco-vault/01-projects/investing/backtests/2026-05-18-memory-cycle-v1.1-v2-honest-rerun.md— the "multi-year offtakes" citation, now annotated rather than retired~/.claude/skills/investing-backtest-thesis/SKILL.md— harness contract (skill, not a vault doc)