The 2023 DRAM trough stopped at the high-cost producer's firm-level cash breakeven, not at cash cost per bit, and that is what sizes the CXMT modifier
The question
Verbatim: "What was the big three's (Samsung, SK hynix, Micron) DRAM cash-cost floor at the 2023 trough, and would CXMT sit above or below it at a 10% vs a 30% cost-per-bit gap?"
Context: [[2026-09-12-cxmt-ddr5-cost-per-bit-yield-parity-tripwire]] recommends moving ChangXin Memory Technologies' (CXMT's) cost leg out of the Phase-2 tripwire AND-condition and re-filing it as a Phase-3 severity modifier. The 2023-trough cash-cost floor for dynamic random-access memory (DRAM) is the number that sizes that modifier. This brief sizes a framework parameter. It is not a position recommendation and contains no price target or buy/sell call.
Two definitions, kept separate throughout, because conflating them is the main way this question goes wrong.
- Cash cost of goods sold (cash COGS). Reported cost of goods sold, less inventory write-downs, less the share of depreciation and amortisation (D&A) sitting in cost of goods sold. This is the quantity a published "cost per bit" figure is comparable to, because cost-per-bit models cover fab output, not corporate overhead.
- Firm-level cash operating cost. Cash COGS plus research and development (R&D) plus selling, general and administrative expense (SG&A). This is the quantity that decides whether a maker is actually burning cash, and therefore whether it has a reason to cut wafer starts.
What we already know (from the vault)
- No absolute cash cost per bit exists anywhere in the vault. A full-text sweep of
01-projects/investingand06-reference/researchfor "cash cost / below cash / cost per bit" returns only the CXMT lineage ([[2026-07-03-cxmt-dram-capacity-memory-oligopoly-risk]], [[2026-07-07-dram-hbm-phase2-phase3-early-signals]], [[2026-09-08-cxmt-ymtc-bit-share-phase3-threshold]], [[2026-09-12-cxmt-ddr5-cost-per-bit-yield-parity-tripwire]]), and every entry there is relative. This brief supplies the first absolute dollars-per-gigabit ($/Gb) derivation. - The only prior use of "cash cost" in the vault is qualitative, and it is the exact premise this brief tests: relative cost matters "at the trough... when prices fall toward the big three's cash cost and a 30% disadvantage means losses, while a 10% disadvantage means rough breakeven" [[2026-09-12-cxmt-ddr5-cost-per-bit-yield-parity-tripwire]].
- The >30% CXMT gap is single-sourced. Every 2026 repetition traces to one SemiAnalysis 1Q26 estimate; no second house has published an independent cost-per-bit number and there has been no update since 1Q26 [[2026-09-12-cxmt-ddr5-cost-per-bit-yield-parity-tripwire]]. The 10% figure originated as a proposed threshold in [[2026-07-03-cxmt-dram-capacity-memory-oligopoly-risk]] and was never derived. Both are treated here as given scenario inputs, not as findings.
- CXMT's capital is not price-disciplined: roughly 77% capital-expenditure intensity on state capital, against Micron at about 21% and SK hynix about 11% [[2026-07-21-memory-maker-fab-financing-supply-side]]. That figure is a 2026 analyst estimate (Futurum, as carried in the source note), used here as a standing structural fact rather than a 2023 measurement. CXMT's bit-share-to-capacity-share ratio is forecast flat at about 0.7 through 2027, meaning it closes essentially none of its density gap [[2026-09-08-cxmt-ymtc-bit-share-phase3-threshold]].
- The vault's 2023 record is thin and the Micron gross-margin series has a 2023 hole. The DRAM-inventory-days anchor series jumps from the quarter ended 2022-12-01 (gross margin 21.9%) to 2025-08-28 [[2026-09-07-memory-supplier-inventory-days-dio-anchor]]. The
memory-cycle-v1.1phase-history notes do carry "Micron Q2 FY23 GM -32.7%" as the down-cycle confirmation, so the fact is in the anchors; the brief-level series is what is missing. The table below fills the four 2023 quarters from primary filings.
What the web says
Micron's fiscal-2023 income statement and cash flow, hard disclosure (SEC XBRL
companyconceptapplication programming interface, CIK 0000723125, as tagged in the FY2023 Form 10-K and the FY2023 Form 10-Qs). Quarterly figures are differenced from the company's own fiscal-year-to-date tags; no estimate is involved. "FQ" denotes Micron's fiscal quarter; its fiscal year ends in late August or early September.Fiscal quarter ended Revenue $M COGS $M Gross profit $M Gross margin Inventory write-down $M Total D&A $M R&D + SG&A $M Operating cash flow $M 2022-12-01 (FQ1'23) 4,085 3,192 +893 +21.9% 0 1,921 1,100 943 2023-03-02 (FQ2'23) 3,693 4,899 -1,206 -32.7% 1,430 1,942 1,019 343 2023-06-01 (FQ3'23) 3,752 4,420 -668 -17.8% 401 1,956 977 24 2023-08-31 (FQ4'23) 4,010 4,445 -435 -10.8% 0 1,937 938 249 2023-11-30 (FQ1'24) 4,726 4,761 -35 -0.7% 0 1,915 1,108 1,401 Full fiscal 2023: revenue $15,540M, COGS $16,956M, gross profit -$1,416M, operating loss -$5,745M, inventory write-downs $1,831M, D&A $7,756M, operating cash flow +$1,559M. Also hard-disclosed in the FY2023 10-K: $382M of period costs from fabrication-facility underutilisation, and wafer-start reductions of about 20% announced at FQ1'23 and about 25% at FQ2'23, covering both DRAM and NAND flash memory (Micron FY2023 10-K; quarterly cut percentages via StorageNewsletter FQ2'23 coverage).
Korean peers, hard disclosure. SK hynix FY2023 revenue of 32,766bn Korean won (KRW) with an operating loss of KRW 7,730bn (-24% margin); 4Q23 returned to a KRW 346bn operating profit, "marking the first quarter of profit following four straight quarters of losses" (SK hynix newsroom, Jan 2024, explicitly preliminary, Korean International Financial Reporting Standards). Samsung's Device Solutions division posted FY2023 sales of KRW 66.59tn and an operating loss of KRW 14.88tn, with memory-business sales of KRW 44.13tn; company-wide depreciation was KRW 35.53tn (Samsung 4Q23 investor-relations deck; press release). Neither maker discloses DRAM-only cost or cash cost.
The trough dates and its depth, analyst (TrendForce). DRAM industry revenue bottomed in calendar 1Q23 at US$9.663bn (-21.2% quarter on quarter, third consecutive decline), then 2Q23 US$11.43bn, 3Q23 US$13.48bn, 4Q23 US$17.46bn. TrendForce's own headline for the period is "Contract Prices Bottom Out in Q3" (1Q23, 2Q23, 3Q23, 4Q23).
Why the fall stopped, verbatim and analyst-sourced. "After suppliers reduce production, there's a diminished inclination to drop prices, which means that contract prices are stabilizing" (TrendForce, 2Q23 release). And: "facing negative gross margins on DRAM products, most manufacturers are resistant to further price reductions"; "Samsung initiated significant production reductions starting in September to diminish its surplus of older inventory. These cuts are projected to hit 30% by the fourth quarter" (TrendForce, Oct 13 2023). Separately, the 1Q23 release forecast 2Q23 utilisation falling to 77% at Samsung, 74% at Micron and 82% at SK hynix; that is a forecast, not a reported outcome.
DRAM-only operating margins at the trough, analyst estimate (TrendForce 2Q23 release): Samsung -24% improving to -9%, SK hynix -50% to -2%, Micron -55.4% to -36%. Micron was the worst of the three in both quarters, but which peer was best rotated: the Micron-to-best-peer spread was 31.4 points of revenue in 1Q23 (Samsung) and 34 points in 2Q23 (SK hynix), while the Micron-to-SK hynix spread in 1Q23 was only 5.4 points.
Price anchors, analyst. DDR4 8Gb average fixed transaction price was US$1.81 in March 2023 (DRAMeXchange, via Tom's Hardware); the same series peaked at US$4.10 in July 2021 and fell below US$2.00 in January 2023 (Asiae, Jan 2023). TrendForce's published forward bands from that point were a 10-15% DRAM average-selling-price (ASP) decline in 2Q23 (iconnect007 reprint) and a 3-8% DDR4-specific decline in 3Q23 (TechPowerUp reprint, Jul 2023, search-result extract; article body not fetched).
Flagged gap, not retried: no free source reachable in this run publishes the actual monthly 2023 contract-price low. TrendForce's
/price/dram/contractand/price/dram/spotreturn HTTP 500, their historical price tables are paywalled, and the Internet Archive CDX endpoint was offline. The Q3-2023 low used below is therefore a chained derivation off the March 2023 print, not a print of its own.
Convergences and contradictions
- Convergence on the date and shape of the trough: 1Q23 revenue low, Q3 2023 price bottom, 4Q23 inflection. Micron's filings, SK hynix's four-loss-quarter run and TrendForce's quarterly series agree, from three independent reporting regimes.
- The 2023 "below cost" narrative is true at the gross-margin line and false at the cash-cost-per-bit line. Micron printed -32.7% gross margin, Samsung's Device Solutions lost KRW 14.88tn, SK hynix lost KRW 7.73tn. Those are full-cost losses and they are mostly depreciation. Strip depreciation and the inventory write-down out of Micron's cost of goods sold and cash COGS runs 44-70% of revenue across the four trough quarters (derivation below). On that measure Micron never came close to cash cost per bit.
- But the firm-level test tells a different and more decision-relevant story, and this is the finding. Adding R&D and SG&A, Micron's firm-level cash operating cost reached 84-89% of revenue in FQ3'23 and 88-93% in FQ4'23, and its reported operating cash flow in FQ3'23 was $24M on $3,752M of revenue, which is cash breakeven to the nearest rounding. The trough did reach a cash floor. It reached the firm's cash floor, roughly 30 points of revenue above its cash cost per bit.
- Contradiction with the parent brief's premise. [[2026-09-12-cxmt-ddr5-cost-per-bit-yield-parity-tripwire]] assumes a trough where "prices fall toward the big three's cash cost," making a 30% gap a loss and a 10% gap a breakeven. The 2023 evidence does not support that geometry on a per-bit basis. It partially supports it on a firm basis, which is a different comparison and one the single-sourced per-bit gap figure is not directly commensurable with.
Synthesis for RDCO
The floor, with its derivation shown. Cash cost per bit is not disclosed by anyone, so it has to be backed out. The one derivation that avoids needing bit volumes is: cash cost per bit = realised ASP x (cash COGS / revenue). Micron's cash COGS is its tagged cost of goods sold less the tagged inventory write-down less the cost-of-goods-sold share of tagged D&A. That last term is an assumption, taken here as 85-95%, because Micron does not split D&A between cost of goods sold and operating expense. On that assumption, cash COGS runs 44.0-49.2% of revenue in FQ2'23, 57.6-62.8% in FQ3'23, 65.0-69.8% in FQ4'23 and 62.2-66.3% in FQ1'24. FQ2'23 is biased low because removing a $1.43bn write-down removes a real period charge, even a non-cash one, from a quarter whose output it related to. FQ3'23 and FQ4'23 are biased the other way, because they sold previously written-down inventory at an artificially low carrying value. The least-contaminated single read is FQ4'23, which took no write-down and is the quarter nearest the Q3-2023 price bottom: 65-70% of revenue, blended DRAM and NAND.
Converting that to DRAM only, and then to dollars per gigabit. Micron's DRAM weight is derived by pairing TrendForce's Micron DRAM revenue ($2.95bn in calendar 2Q23, $3.075bn in 3Q23) with Micron's hard-disclosed quarterly totals over the nearest fiscal quarters, giving roughly 77-79%; the periods are misaligned by about a month, so call it 77%. Solving the blend b = w·x + (1-w)·n for the DRAM-only ratio x, with b = 65-70%, w = 0.77, and a NAND cash-COGS ratio n assumed in the range 80-100% of revenue (NAND was the worse business in 2023, and that assumption is this derivation's softest input), gives x = 55-67% of realised DRAM price. Applying that to the chained Q3-2023 DDR4 8Gb low of $1.42-1.58, or $0.177-0.198/Gb, gives Micron's DRAM cash cost of goods sold at the 2023 trough of roughly $0.097-0.132 per Gb, or $0.78-1.06 per GB. Two caveats sit on that number. It is the high-cost producer's floor: Samsung's and SK hynix's are lower, and the TrendForce operating-margin spread of 31-34 points cannot be decomposed into cash cost versus depreciation intensity versus product mix from public data, so this brief does not put a number on theirs. And applying a Micron-wide revenue ratio to a DDR4 commodity contract price biases the floor down, because Micron's realised DRAM ASP was higher than DDR4 8Gb (TrendForce's 2Q23 release notes DDR5 held "a significant proportion" of shipments at a relatively stable ASP).
Where CXMT sits, on the band rather than a midpoint. On the literal question CXMT sits above the floor at both gaps by construction; the magnitudes are small in absolute terms, roughly $0.010-0.013/Gb at a 10% gap and $0.029-0.040/Gb at a 30% gap. The decision-relevant comparison is CXMT's cash COGS against the price the 2023 trough actually cleared at. Carrying the full band through: at a 10% gap CXMT's cash COGS is $0.107-0.146/Gb against a $0.177-0.198/Gb trough price, a cash-COGS margin of 18-46%. At a 30% gap it is $0.127-0.172/Gb, a margin of 3-36%. Because required-price-fall-to-breakeven and cash margin are the same quantity, those are also the further declines below the 2023 low that would put CXMT under water on a per-bit basis: 18-46% at a 10% gap, 3-36% at a 30% gap. So the per-bit answer is above the floor and below the trough price across the whole band in both scenarios, but the 30% case comes within about 3 points of breakeven at the pessimistic end while the 10% case keeps at least 18 points everywhere. That is where the 10-versus-30 distinction actually lives, and it is a statement about the band's tail, not its middle.
The firm-level answer, which is the one that predicts behaviour. Micron cut wafer starts twice, and the first cut of about 20% was announced at FQ1'23 when its gross margin was a positive 21.9%, long before any cash test. By FQ3'23 it was at operating-cash breakeven ($24M) and, with hard-disclosed fiscal-2023 capital expenditure of $7,676M against $1,559M of operating cash flow, deeply free-cash-flow negative. So the 2023 price floor was behavioural: it was set by discretionary supply cuts taken at negative gross margin and firm-level cash breakeven, roughly 30 points of revenue above cash cost per bit. Carrying Micron's own overhead load of 23-26% of revenue across to CXMT as a scenario assumption, CXMT's firm-level cash cost at 2023 trough prices would be 87-123% of that price at a 30% gap (cash-negative over most of the band, positive only at its optimistic end) and 78-108% at a 10% gap (cash-positive over most of the band, negative only at its pessimistic end). That is scenario arithmetic on an assumption CXMT's public filings cannot yet support, since its only reported quarters are boom quarters at 79-88% gross margin ([[2026-09-12-cxmt-ddr5-cost-per-bit-yield-parity-tripwire]], third-party, TrendForce), and it is labelled as such.
How to re-file the cost leg, concretely. Three changes. First, the modifier should bite in both scenarios, because at neither gap does a 2023-style trough reliably force CXMT off a per-bit cash basis. Second, it should bite harder at 10% than at 30%, but the difference is in the tail rather than the central case, so the ~10% threshold should be retired as a binary trigger: the number was proposed without derivation, and the derivation now available shows no step change there. Third, and most useful: key the modifier to CXMT's bit share at the time of rollover multiplied by (1 minus the probability CXMT cuts), and let the cost gap enter only as a prior on that probability. Those priors are unanchored Ray judgment, not derived: roughly 0.1 at a 10% gap, roughly 0.25-0.35 at a 30% gap, on the reasoning that a state-funded producer at 77% capex intensity cuts only when overhead-inclusive cash losses become politically expensive. That keys the modifier to two things the vault can already observe, the bit share from [[2026-09-08-cxmt-ymtc-bit-share-phase3-threshold]] and a cut-or-no-cut event now that CXMT is listed, rather than to a single-sourced cost estimate not refreshed since 1Q26. The cost gap still matters, but for a Phase-2 question rather than a Phase-3 one: at either gap CXMT can take share on a positive per-bit cash margin at the big three's trough prices, which is a statement about share-gain velocity, not trough depth.
One precedent worth recovering before anyone leans on this. 2023 is a trough that stopped at a firm's cash breakeven. The cycle that plainly went below cash cost on a per-bit basis is 2008-09, and what it produced was exits rather than a price floor (Qimonda's insolvency, Elpida's path to it). That is background knowledge in this brief, uncited, and item four of the follow-ups below turns it into a second and much harsher anchor for the modifier.
Why this is in the vault
It supplies the missing parameter for the specific change [[2026-09-12-cxmt-ddr5-cost-per-bit-yield-parity-tripwire]] proposed to tripwire row 5a in [[2026-07-07-dram-hbm-phase2-phase3-early-signals]], and it fills the 2023 hole in the Micron gross-margin and cash-cost series that [[2026-09-07-memory-supplier-inventory-days-dio-anchor]] left open, with primary filing data the anchor files can ingest directly.
Open follow-ups
- Micron does not split D&A between cost of goods sold and operating expense in its tagged data. A 10-K footnote or segment disclosure that does would retire the 85-95% assumption and tighten the cash-COGS band by roughly 5 points of revenue, which is the single largest controllable source of width in the floor.
- A verified monthly 2023 DDR4 8Gb contract-price series, to replace the chained Q3-2023 low ($0.177-0.198/Gb) with a hard print. Distinct from the existing 2017-18 series follow-up; the 2023 gap is a paywall plus a dead TrendForce price endpoint, not an archive problem.
- Does CXMT, now listed, disclose capacity utilisation, wafer starts, or operating-expense ratios quarterly? Utilisation is the observable for the "does CXMT cut" term that this brief argues should carry more weight than the cost gap; the operating-expense ratio is what would replace the Micron-like overhead assumption in the firm-level scenario.
- Recover the 2008-09 trough's price-versus-cash-cost gap and the exits it caused. It is the only DRAM cycle that plainly cleared below cash cost per bit, and it would give the severity modifier a second anchor far harsher than 2023 provides.
- Is there a public bottom-up cost model (TechInsights, Yole, or similar, non-paywalled) giving absolute $/Gb cash cost for a named DRAM node? That would replace this brief's margin-residual derivation with an independent one and test the $0.097-0.132/Gb band, including the DDR4-versus-realised-ASP downward bias flagged above.
- Did the big three's 2023 wafer cuts fire at a consistent firm-level cash level rather than a consistent gross margin? Micron's first cut came at +21.9% gross margin and its cash breakeven came two quarters later, so gross margin looks like the wrong trigger variable. If operating-cash-flow-near-zero is the real cut trigger, that is both more usable as a Phase-3 depth marker and observable quarterly.
- Were Samsung and SK hynix above or below firm-level cash breakeven in 2023? Neither discloses DRAM-only cost, but their consolidated cash-flow statements would settle whether the behavioural floor was common across the three or specific to the high-cost producer.
Related
- [[2026-09-12-cxmt-ddr5-cost-per-bit-yield-parity-tripwire]] — the parent brief whose severity-modifier proposal this sizes
- [[2026-07-03-cxmt-dram-capacity-memory-oligopoly-risk]] — origin of the underived ~10% threshold
- [[2026-07-07-dram-hbm-phase2-phase3-early-signals]] — the canonical tripwire table containing row 5a
- [[2026-09-08-cxmt-ymtc-bit-share-phase3-threshold]] — the bit-share leg this modifier should key to
- [[2026-09-07-memory-supplier-inventory-days-dio-anchor]] — the anchor series with the 2023 gap this brief fills
- [[2026-09-09-2018-capacity-announce-asp-rollover-lag]] — announce-to-ASP lag, the timing counterpart to this depth question
- [[2026-07-21-memory-maker-fab-financing-supply-side]] — the 77% state capex intensity that breaks the behavioural floor
- [[2026-09-29-phase3-memory-cycle-rotation-basket]] — the drawdown base rates a severity modifier would scale
- [[2026-10-01-micron-fq4-fy2026-dram-hbm-tripwire-rescan]] — current phase call (Phase 2, 1 of 5 fired)
- [[2026-05-18-memory-cycle-v1.1]] — the executable thesis whose phase triggers this parameter feeds
- [[phase-history-notes]] — the
memory-cycle-v1.1anchor notes carrying the -32.7% down-cycle confirmation
Sources
Vault
06-reference/research/2026-09-12-cxmt-ddr5-cost-per-bit-yield-parity-tripwire.md06-reference/research/2026-07-03-cxmt-dram-capacity-memory-oligopoly-risk.md06-reference/research/2026-07-07-dram-hbm-phase2-phase3-early-signals.md06-reference/research/2026-09-08-cxmt-ymtc-bit-share-phase3-threshold.md06-reference/research/2026-09-07-memory-supplier-inventory-days-dio-anchor.md06-reference/research/2026-09-09-2018-capacity-announce-asp-rollover-lag.md06-reference/research/2026-07-21-memory-maker-fab-financing-supply-side.md06-reference/research/2026-09-29-phase3-memory-cycle-rotation-basket.md06-reference/research/2026-10-01-micron-fq4-fy2026-dram-hbm-tripwire-rescan.md01-projects/investing/theses/2026-05-18-memory-cycle-v1.1.md01-projects/investing/anchors/memory-cycle-v1.1/phase-history-notes.md
Hard disclosure
- SEC XBRL
companyconceptAPI, CIK 0000723125 (Micron), conceptsCostOfGoodsAndServicesSold,GrossProfit,DepreciationDepletionAndAmortization,InventoryWriteDown,ResearchAndDevelopmentExpense,SellingGeneralAndAdministrativeExpense,OperatingIncomeLoss,NetCashProvidedByUsedInOperatingActivities,PaymentsToAcquirePropertyPlantAndEquipment— https://data.sec.gov/api/xbrl/companyconcept/CIK0000723125/us-gaap/ - Micron FY2023 Form 10-K — https://www.sec.gov/Archives/edgar/data/723125/000072312523000054/mu-20230831.htm
- SK hynix 2023 / 4Q23 results (preliminary) — https://news.skhynix.com/en/sk-hynix-reports-fourth-quarter-2023-financial-results/
- Samsung Electronics 4Q23 and FY2023 investor-relations deck — https://images.samsung.com/is/content/samsung/assets/global/ir/docs/2023_4Q_conference_eng_01.pdf
- Samsung Electronics 4Q23 / FY2023 press release — https://news.samsung.com/global/samsung-electronics-announces-fourth-quarter-and-fy-2023-results
Analyst estimate
- TrendForce 1Q23 DRAM revenue and 2Q23 utilisation forecast — https://www.trendforce.com/presscenter/news/20230525-11688.html
- TrendForce 2Q23 DRAM revenue, DRAM-only operating margins, DDR5 shipment mix — https://www.trendforce.com/presscenter/news/20230824-11805.html
- TrendForce Oct 2023 on negative DRAM gross margins and Samsung's 30% cut — https://www.trendforce.com/presscenter/news/20231013-11880.html
- TrendForce 3Q23 DRAM revenue, "contract prices bottom out in Q3" — https://www.trendforce.com/presscenter/news/20231204-11942.html
- TrendForce 4Q23 DRAM revenue — https://www.trendforce.com/presscenter/news/20240305-12060.html
- TrendForce 2Q23 ASP decline band of 10-15%, via iconnect007 — https://iconnect007.com/article/135733/decline-in-dram-asp-narrows-to-1015-in-2q23-with-no-end-in-sight/135736/ein
- TrendForce 3Q23 DDR4 decline band of 3-8%, via TechPowerUp (search-result extract) — https://techpowerup.com/310864/dram-asp-decline-narrows-to-0-5-for-3q23-owing-to-production-cuts-and-seasonal-demand
- DDR4 8Gb fixed price March 2023, via Tom's Hardware — https://www.tomshardware.com/news/dram-prices-dropped-20-in-q1-2023-will-keep-falling-in-q2
- DDR4 8Gb July 2021 peak and sub-$2 January 2023, via Asiae — https://www.asiae.co.kr/en/article/2023013116061036730
- Micron FQ2'23 wafer-start cut detail, via StorageNewsletter — https://www.storagenewsletter.com/2023/03/30/micron-fiscal-2q23-financial-results/
- SemiAnalysis 1Q26 CXMT cost-per-bit estimate (the single source for >30%; not fetched in this run, carried from the parent brief) — https://newsletter.semianalysis.com/p/chinas-cxmt-is-set-to-challenge-dram
Paywalled or unreachable, flagged and not retried
- TrendForce historical DRAM contract and spot price tables (
/price/dram/contract,/price/dram/spot): HTTP 500; historical series paywalled - Internet Archive CDX: offline during this run, so no Wayback recovery of the 2023 price table
- Silicon Analysts DDR4 price history: 2023 contract values paywalled