Pian Chen* and Rui Huang** | July 7, 2026
I. Introduction
On June 25, 2026, Bathaee Dunne LLP filed a class action, Garciaguirre et al. v. Samsung Electronics Co., Ltd. et al., in the U.S. District Court for the Northern District of California (No. 3:26-cv-06345), on behalf of plaintiffs (14 individuals and three small PC businesses) against three dominant DRAM makers, Samsung, SK Hynix, and Micron, which together hold roughly 90% of the global DRAM market. The complaint alleges a violation of Section 1 of the Sherman Act and seeks class certification, an injunction, and treble damages.[1]
The complaint alleges the defendants used an AI-driven shift toward High Bandwidth Memory (HBM) as cover to restrict output of commodity DRAM (e.g., DDR3 through DDR5 and LPDDR5X), while carving out HBM itself from the relevant product market. The complaint alleges that the coordinated supply restriction began around 2022, when demand was weak and prices were falling — the moment at which, plaintiffs say, the defendants chose to cut rather than compete. The headline figure is striking: commodity DRAM prices have risen roughly 700% between the third quarter of 2024 and the first quarter of 2026, within a proposed class period running from 2022 to 2026.[2]
As background, these three defendants faced a nearly identical Section 1 case in this same courthouse in 2018 (In re DRAM Indirect Purchaser Antitrust Litigation, No. 4:18-cv-02518).[3] That case was dismissed by the district court in 2020 as the plaintiffs’ eight plus factors did not meet Section 1’s bar. The dismissal was affirmed by the Ninth Circuit in 2022, holding the conduct was “more likely explained by lawful, unchoreographed free-market behavior” in a concentrated industry than by agreement.[4]
Parallel output reduction in a concentrated industry can reflect tacit coordination or independent rational behavior facing similar business conditions. In this article, we explain what plus factors are and evaluate the plus factors in the 2026 DRAM complaint by checking each firm’s stated rationale against publicly disclosed earnings, capacity, and losses, rather than relying on market structure or timing of supply cuts alone. Based on public information, we think the defendants’ commodity DRAM supply decisions and AI-driven strategic shifts to HBM were consistent with independently disclosed business events on a coherent timeline, therefore economically defensible as parallel, individually profit-motivated conduct rather than behavior that requires agreement or coordination.
The rest of this article proceeds as follows. Section II introduces the case and its theory of harm. Section III summarizes the legal standard the complaint must clear, explains three categories of plus factors and why the 2018 case against these same defendants failed. Section IV addresses market structure, the weakest of the plus factors. Section V, the core of our economic analysis, tests whether the defendants’ conduct is explicable as independent, individually rational profit-maximization, addressing the complaint’s strongest factual claims in turn. Section VI addresses the complaint’s traditional conspiracy evidence, the category on which it is thinnest. Section VII concludes.
II. The Theory of Harm
The plaintiffs allege that Samsung, SK Hynix, and Micron agreed — tacitly or expressly — to keep conventional commodity DRAM artificially scarce to inflate prices, in violation of Section 1 of the Sherman Act. The novel feature is the alleged mechanism: the complaint contends that the defendants used a coordinated, publicly announced pivot of manufacturing capacity toward HBM — the stacked DRAM that feeds AI accelerators — as cover to curtail output of older commodity products (e.g., the DDR3, DDR4 and DDR5 modules in computers, the LPDDR memory in phones, and the server memory in data centers). By steering wafers to HBM, the argument runs, the three defendants could collectively tighten commodity DRAM supply while pointing to surging AI demand as an innocent explanation. The resulting harm is framed as classic supply restriction: less commodity DRAM, higher prices, borne by PC builders, repair shops, and consumers.
The complaint’s theory has two parts. The first is not seriously in dispute: all three defendants restricted commodity DRAM output beginning in 2022, cutting production within roughly a six-month window. The second part is the contested claim: that the shift toward HBM cannot fully account for the resulting DRAM shortage, so that some portion of the restriction reflects coordination rather than an independently rational reallocation of capacity.
The price data itself is likewise not in dispute. Per the complaint, conventional DRAM contract prices rose 171.8% year-over-year through Q3 2025, another 50% quarter-over-quarter in Q4 2025, and 93–98% quarter-over-quarter in Q1 2026 — a compounded increase of approximately 697% — while retail DDR5 module prices roughly tripled or quadrupled over the same period.[1] A price run-up of this size could be consistent with a coordinated restriction, or with a shared shock that the three firms independently and simultaneously responded to. Section V tests which explanation the public record supports.
III. The Legal Standard: Plus Factors and Why the 2018 Case Failed
This is the analytical heart of the case. Parallel conduct alone — even conscious parallelism — does not violate Section 1. Three rational firms in an oligopoly can each independently make the same move without any agreement, and that is lawful. To get from “they all did the same thing” to “they agreed,” plaintiffs must plead “plus factors”: circumstantial evidence that is contrary to independent action and points toward a meeting of the minds.
Courts and commentators have recognized more than twenty distinct plus factors, which can be grouped into three broad categories:[5][6]
- Market structure / motive to conspire: A market structure that is conducive to collusion, which benefits defendants.
- Actions against unilateral self-interest: Conduct that would be irrational unless the firm had advance assurance that its rivals would go along.
- Traditional conspiracy evidence: Such as inter-competitor communications, information exchanges, or opportunities to collude (a recognized, non-economic category of plus factor).
The analysis is holistic: no single factor is required, and each additional factor reinforces the inference from the others. Direct evidence of communication is the most powerful, but plaintiffs rarely have it at the pleading stage. That leaves the two economic categories, and conduct against self-interest is usually the more probative. Market structure shows only that collusion would be feasible. Conduct that makes sense only with advance assurance that rivals will follow points toward an actual agreement.
This is the second time these three defendants have faced a parallel-conduct theory in this courthouse, and the 2018 case’s failure sets the bar this complaint must clear. Under Bell Atlantic Corp. v. Twombly, 550 U.S. 544 (2007), parallel conduct alone cannot support a Section 1 claim; plaintiffs must plead plus factors that make an actual agreement more plausible than independent, conscious parallelism. The 2018 class action alleged similar parallel production cuts and offered eight plus factors — (1) price signaling; (2) complex, simultaneous, and historically unprecedented decreases in capital investment; (3) supply cuts against the firms’ self-interest; (4) public statements encouraging supply cuts; (5) changed conduct between the start and end of the class period; (6) information exchanges among the defendants regarding future supply and demand; (7) high market concentration; and (8) the firms’ prior criminal convictions for price fixing. The district court dismissed the case in 2020, rejecting all eight factors individually and holistically. The Ninth Circuit affirmed the district court’s decision in 2022, holding the conduct was “more likely explained by lawful, unchoreographed free-market behavior” in a concentrated industry than by agreement.[4]
The current complaint is built to fill that gap. Its wager is that the HBM capacity shift supplies what the 2018 case lacked — a specific, shared mechanism tying the parallel DRAM supply restrictions together — reinforced by the near-simultaneous customer-and-order “policing” by all three firms in January 2026, Samsung’s counter-cyclical output cuts, and Micron’s shutdown of its profitable Crucial consumer brand. Sections IV through VI test whether that mechanism and the surrounding conduct clear the bar.
IV. Market Structure
Samsung, SK Hynix, and Micron together hold roughly 90% of the DRAM market, a share the complaint alleges has held for years, including before the AI and HBM boom.[1] Entry is effectively foreclosed: a single modern fab costs $15–20 billion, takes years to build, and cannot readily be repurposed for other semiconductor types — so coordination, if it occurred, could be both feasible and profitable.[1]
These market structure facts are not in dispute, and they explain why parallel conduct could be sustained: no entrant or fringe producer can add capacity fast enough to compete the shortage away. But high concentration does not automatically mean coordination. The 2022 Ninth Circuit panel weighed nearly identical concentration figures — there, roughly 96% combined share — and still affirmed dismissal, reasoning that firms in a concentrated, interdependent market are simply “cognizant of — and reacting to — similar market pressures.”[4] We treat market structure the same way here: necessary context for the analysis in Section V, not evidence in itself.
V. Economic Analysis of Firm Conduct
The complaint’s evidence for coordination rests on four factual claims. The first two claims involve conduct from late 2022 and early 2023, before the AI boom — so any innocent explanation must come from ordinary industry cycles, and it does. The latter two involve conduct from 2024 onward, and the explanation lies in HBM itself: a technology-gated product growing explosively. First, the synchronized cuts: SK Hynix and Micron throttled commodity production within three weeks of each other in late 2022, before the AI-driven demand shift existed. Second, Samsung’s reversal: Samsung, which had held to a documented counter-cyclical strategy through the 2008–2009 and 2019 downturns, broke that pattern in April 2023 — the hardest fact in this set to explain. Third, the capex shift: from 2024 onward, all three defendants’ capital expenditure shows a reallocation toward HBM. Fourth, Micron’s Crucial brand and the Stargate project: Micron announced shutting down its Crucial consumer brand in December 2025, roughly nine weeks after Samsung and SK Hynix committed substantial DRAM wafer capacity to OpenAI’s Stargate project.
Each claim reduces to the same question: did the defendants forgo DRAM output that was individually profitable to produce, in a way explicable only by coordination? Our approach is the same across all four: check each firm’s stated or implied rationale against publicly disclosed business facts — earnings, capacity, and losses — rather than against market structure or timing alone. Our conclusions, developed below, are these: the first two claims are explained by an ordinary, severe industry downturn that predates any AI rationale, and the latter two are explained by the rise of HBM itself — a technology-gated, exponentially growing product that each firm had independent reasons to pursue. In short, each of the four claims has an individually rational explanation that requires no inference of coordination.
V.1 The Synchronized Cuts in Late 2022
SK Hynix and Micron cut commodity production within three weeks of each other in late 2022. The decisions were driven by weak demand and significant financial losses at the time, and the firms’ contemporaneous explanations made no reference to HBM or AI. SK Hynix’s October 26, 2022 release stated that “supply will continue to exceed demand for the time being” and announced a capital-spending cut of more than 50% for the following year; Micron followed on November 16, reducing wafer starts by roughly 20% and describing “bold and aggressive steps to reduce bit supply growth” to limit inventory.[1] What motivated them is well documented and ordinary: a severe, industry-wide demand correction following the post-pandemic PC and smartphone inventory build-up — the same kind of downturn the industry has experienced roughly every two years for decades. A three-week gap between SK Hynix and Micron independently reacting to the same publicly visible glut, each citing its own inventory position, is not evidence of coordination; it is what independent firms watching the same falling demand and prices look like.
Could the synchronized cuts nonetheless reflect each firm counting on the others to follow? The economics say the cuts did not need that assumption. By late 2022, prices were falling toward cost (within a quarter they fell below it) and inventories were swelling. Every additional wafer meant selling at a deeper loss. Cutting output in those conditions is individually rational even if no rival follows — the same logic that leads any firm to stop making a product it can only sell at a loss. And the record shows there was no followership to count on: Samsung, the market leader, resisted cutting and did not move until April 2023, more than five months later. SK Hynix and Micron held to their cuts anyway. Even if each firm also anticipated that rivals facing the same losses would eventually reach the same conclusion, anticipation is not agreement. Firms in a concentrated market lawfully base decisions on the expected reactions of competitors — that is conscious parallelism, and the Ninth Circuit has held it does not violate Section 1.[4] What plaintiffs must show is advance assurance, and nothing in the late-2022 record supplies it.
V.2 Samsung’s Reversal: Deviation from Its Historical Strategy
The complaint leans on Samsung deviating from its historical playbook. In prior memory gluts — most vividly the 2008–2009 “chicken game” — Samsung kept investing and producing into a collapsing market, driving DRAM prices down (prices fell from several dollars to well under $1). This was painful but strategically successful for Samsung: weaker, higher-cost rivals could not survive the bleed. Germany’s Qimonda went bankrupt in 2009, Japan’s Elpida filed for bankruptcy in 2012 (later absorbed by Micron), and the industry consolidated from more than twenty DRAM makers into today’s three — with Samsung emerging as the share leader.[7] So historically, the counter-cyclical strategy helped Samsung: it used downturns as a weapon to take share and force consolidation.
That history is what makes Samsung’s April 2023 production cut notable: a firm that had repeatedly used downturns to expand and outlast weaker rivals instead chose to cut alongside SK Hynix and Micron. Plaintiffs frame that as out of character and against unilateral self-interest. The defense response may be that the market was already a stable three-firm oligopoly (the weak rivals were already gone), so “discipline” rather than a “share war” was the rational strategy by 2023 — a change in conduct explained by a changed market structure, not an agreement.
Plaintiffs appear to have anticipated this defense: the complaint alleges that Samsung’s reversal “is rational only if Samsung had assurance that its rivals would likewise restrain rather than exploit Samsung’s sacrifice.”[1] That deserves a real answer, and public records predating this litigation by more than three years supply one. On March 28, 2023 — ten days before Samsung’s reversal — Micron reported a fiscal Q2 operating loss of $2.3 billion and a gross margin of -32.7%, and told the market it would maintain flat annual bit share, cutting price to defend that share rather than cede it.[8] Samsung’s traditional strategy of producing through downturns depends on rivals eventually giving up; by March 2023, Micron had just said publicly, in an independently verifiable disclosure, that it would not retreat. Samsung cutting ten days after that disclosure is consistent with the lawful conscious parallelism the Ninth Circuit described: one firm adjusting to a rival’s public commitment, not agreeing with it.[4] Micron is the third-largest supplier, not the market leader — but the economics here do not depend on who moves first. Samsung’s war-of-attrition strategy pays off only if rivals eventually retreat; once Micron publicly committed to defending its bit share whatever the price, the attrition war had no foreseeable end, and cutting output became the individually rational choice.
V.3 The Capex Shift: Is the Move from DRAM to HBM Rational Firm Behavior?
The capex shift from 2024 onward raises a further economic question. HBM is the scarce, high-growth product at the center of the AI boom. If HBM is more attractive than commodity DDR3/DDR4, then three profit-maximizing firms would each, independently, reallocate scarce wafer capacity toward it — no agreement required. That is precisely the “lawful, unchoreographed free-market behavior” the Ninth Circuit credited when it affirmed dismissal of the earlier 2018 DRAM case.[4] On this view, curtailed DDR3/DDR4 output is a byproduct of everyone chasing the better product, not a conspiracy.
What does the margin data tell us? Industry commentary put HBM gross margins around 60% as HBM manufacturing capacity was being ramped up in 2023–2024. Commodity DRAM, by contrast, was deeply unprofitable through the 2022–2023 glut — SK Hynix posted its first quarterly operating loss in a decade in late 2022, and Samsung’s chip division ran billions in losses through 2023 — before memory margins recovered across 2024 as AI demand took hold. It makes economic sense for each firm to switch from DRAM that suffered great losses to HBM that earned high margins during this period.
The HBM growth has been extraordinary: Micron’s cloud memory business unit grew from $1.87 billion to $13.52 billion in two fiscal years,[9] and SK Hynix’s HBM revenue more than doubled year-over-year in FY2025.[10] Participation in that growth is also technology-gated in a way commodity DRAM is not: each HBM generation requires firm- and customer-specific qualification that can be — and has been — failed. Samsung’s HBM3E qualification with Nvidia failed publicly in April 2024 and did not pass until September 2025,[11] and Micron’s early HBM4 samples ran behind its rivals’ in early 2026 before it was confirmed as a qualified supplier for Nvidia’s Vera Rubin platform in June 2026.[12]
In contrast, commodity DRAM does not carry an equivalent technology gate: its foundational standard, JEDEC’s DDR5 specification, was jointly developed by all three defendants as JEDEC members and published in July 2020, and capital spending there buys more wafers of an already-shared specification — with all three firms having lived through what that additional supply tends to produce.[13] The conclusion follows: moving from cyclical, commoditized DRAM to high-growth, differentiated HBM was a rational strategic decision that each of the three firms would make individually — and in the new AI age, no firm could afford to cede that ground to its rivals.
V.4 The Margin Inversion, Crucial Brand, and Stargate Project
By late 2025–2026, industry news suggests DDR5 margins caught up to or even surpassed HBM — a striking margin inversion in which the shortage itself made “outdated” commodity DRAM extraordinarily profitable.[14] One caveat matters for understanding this inversion: capacity moved to HBM is not easily moved back. HBM is several times as wafer-intensive per bit as DDR5 — the complaint itself alleges it requires more than twice the silicon wafer area per gigabyte[1] — and depends on specialized stacking and packaging tooling with long lead times, so wafers committed to HBM when its margins were highest could not be quickly redirected when commodity margins later caught up. Because firms cannot forecast demand with certainty or switch capacity between products easily in the short run, the later margin inversion is at least as consistent with ordinary business uncertainty as with a scheme.
One development, though, invites closer scrutiny: Micron’s decision to wind down its Crucial direct-to-consumer brand just as commodity DRAM margins were catching up to and even surpassing those on HBM. The timing is notable. In October 2025, Samsung and SK Hynix reportedly signed letters of intent to supply OpenAI’s Stargate project with up to 900,000 DRAM wafers per month. If fully implemented, that volume would commit on the order of 40% of global DRAM output to a single customer. Two months later, in December 2025, Micron announced it would wind down Crucial by February 2026. With conventional DRAM scarce and highly profitable, Micron could have leaned harder into its consumer channel. Instead it exited, citing AI and data-center demand. Micron’s own disclosure described the exit as reallocating supply “to our larger, strategic customers in faster-growing segments,”[15] and its long-term cloud agreements carry floor-price terms that management says deliver “a very robust gross margin” — comparable to or better than what Crucial earned.[16] On that record, the Crucial exit looks like a shift between two comparable revenue streams, not a sacrifice of margin. It is consistent with an independently rational decision; only direct evidence from discovery — communications among the defendants about the timing — could show otherwise.
VI. Traditional Conspiracy Evidence
This is the category on which the complaint is thinnest: no direct communication among the defendants is alleged anywhere in its 118 pages. Two categories of evidence stand in for that gap, and neither closes it.
The first is the near-simultaneous January 2026 “policing” of the market: all three makers, at roughly the same time, began vetting customers and pressing buyers on who they would resell to, how much they truly wanted, and whether their stated demand was real — with one customer quoted saying that “the three companies” had all grown stricter and were asking the same questions at the same time.[1] Identical behavior observed by a customer is the closest the complaint comes to coordination language, but it remains parallel conduct as seen from the outside, not communication between the defendants themselves.
The second is the firms’ criminal history from the 1998–2005 DRAM cartel, and here the record needs a correction. SK Hynix pleaded guilty and paid $185 million in April 2005, Samsung pleaded guilty and paid a $300 million fine in October 2005, and both pleas covered conduct between April 1999 and June 2002.[17] Micron did not plead guilty: it participated in the cartel but escaped prosecution as the first conspirator to report the conspiracy and cooperate under the DOJ’s corporate leniency program, and the only individual charge touching the company was a regional sales manager’s guilty plea to obstruction of justice.[17] That history is not evidence of communication over the current cuts.
VII. Conclusion
The 2026 DRAM class action asks a court to infer a Sherman Act Section 1 agreement from the three firms’ parallel supply decisions: an alleged pivot to HBM used as cover to keep commodity DRAM scarce while prices climbed roughly 700% (Section II). The legal bar the complaint must clear was set by Twombly and illustrated by the failure of the nearly identical 2018 case against these same defendants: parallel conduct becomes actionable only with plus factors that make an agreement more plausible than independent, profit-motivated conduct (Section III).
Market structure cannot carry that burden by itself. The defendants’ roughly 90% combined share and prohibitive entry barriers explain why a parallel restriction is feasible, but the Ninth Circuit weighed nearly identical concentration in 2022 and still affirmed dismissal — high concentration is context, not evidence (Section IV).
Checked against the defendants’ own disclosures (Section V), all four claims the 2026 DRAM complaint rests on have individually rational explanations that require no inference of coordination. The synchronized cuts trace to an ordinary demand correction that predates the AI opportunity entirely. Samsung’s reversal traces to a competitor’s independently disclosed loss and public commitment to defend its market position, ten days before Samsung acted. The capex shift to HBM and Crucial’s closure both trace to a technology-gated, exponentially growing opportunity that Samsung, SK Hynix, and Micron have each separately, and verifiably, pursued.
The complaint’s traditional conspiracy evidence adds little (Section VI): no direct communication among the defendants is alleged in its 118 pages, the January 2026 customer vetting is parallel conduct observed from the outside, and the 2005 criminal history reaches only two of the three defendants — Micron did not plead guilty.
Taken together, the 2026 complaint is a more serious attempt than its 2018 predecessor: the price run-up is larger, the market just as concentrated, and the alleged mechanism more specific. But on the public record, each building block of the case is at least as consistent with three firms independently chasing the same extraordinary AI opportunity as with an agreement. Whether the complaint survives dismissal will turn on whether its plus factors, viewed holistically, push the inference of agreement across Twombly’s line; what would ultimately decide the matter is the evidence only discovery can reach — the defendants’ internal records and firm-level wafer and margin data.
This commentary is based on the complaint and publicly available information. The allegations are unproven, and nothing here constitutes legal advice. As with our NAND study[18], the analysis is necessarily limited by the public record: evidence of private communications among the firms is not accessible through public sources, and were such evidence to emerge, it could materially change the assessment.
Sources
[1] Complaint and docket, Garciaguirre et al. v. Samsung Electronics Co., Ltd. et al., U.S. District Court, N.D. Cal., No. 3:26-cv-06345 (filed June 25, 2026) (via PacerMonitor / Justia Dockets). https://dockets.justia.com/docket/california/candce/5:2026cv06345/472931
CourtListener docket: https://www.courtlistener.com/docket/73532135/garciaguirre-v-samsung-electronics-co-ltd/
The complaint itself: https://stream-ai.s3.us-west-1.amazonaws.com/class-action-complaints/6-25-26/Garciaguirre_et_al_v_Samsung_Electronics_Co_Ltd_et__candce-26-06345__0001.0.pdf
[2] Luke James, “Samsung, SK Hynix, and Micron sued over alleged DRAM price fixing amid record memory costs — lawsuit claims coordinated HBM shift was cover to curtail DDR3 and DDR4 production,” Tom’s Hardware (June 29, 2026). https://www.tomshardware.com/tech-industry/samsung-sk-hynix-and-micron-sued-over-alleged-dram-price-fixing-amid-record-memory-costs
[3] In re Dynamic Random Access Memory (DRAM) Indirect Purchaser Antitrust Litigation, No. 4:18-cv-02518-JSW (N.D. Cal.), consolidated amended class action complaint (named plaintiffs). https://www.courtlistener.com/docket/6379330/
[4] In re Dynamic Random Access Memory (DRAM) Indirect Purchaser Antitrust Litigation, No. 21-15125 (9th Cir. Mar. 7, 2022). https://cdn.ca9.uscourts.gov/datastore/opinions/2022/03/07/21-15125.pdf
[5] Paul, Weiss, “Ninth Circuit Analyzes Parallel Conduct Allegations and Does Not Find ‘Something More’ Suggesting an Antitrust Conspiracy” (client memorandum), Mar. 9, 2022. https://www.paulweiss.com/media/gdnbsgx0/ninth_circuit_analyzes_parallel_conduct_allegations_and_does_not_find_something_more_suggesting_an_antitrust_conspiracy.pdf
[6] Christopher R. Leslie, “Plus Factors in Price-Fixing Litigation,” Columbia Law School Blue Sky Blog (Nov. 24, 2021). https://clsbluesky.law.columbia.edu/2021/11/24/plus-factors-in-price-fixing-litigation/
[7] “DRAM maker Elpida files for bankruptcy,” Computerworld (Feb. 2012); industry histories of the Qimonda (2009) and Elpida (2012) bankruptcies and DRAM consolidation. https://www.computerworld.com/article/1549504/dram-maker-elpida-files-for-bankruptcy.html
[8] Micron Technology, Inc., fiscal Q2 2023 results, Form 8-K Exhibit 99.1 and earnings-call transcript (Mar. 28, 2023). https://investors.micron.com
[9] Micron Technology, Inc., Form 10-K filings (FY2024 and FY2025), revenue by technology and business-unit disclosures. https://investors.micron.com
[10] SK Hynix, quarterly and FY2025 earnings releases; Counterpoint Research coverage of Q1 2023 DRAM results. https://news.skhynix.com
[11] KED Global, reporting on Samsung passing Nvidia’s HBM3E qualification tests (Sept. 2025). https://www.kedglobal.com
[12] Industry reporting on Micron’s HBM4 qualification for Nvidia’s Vera Rubin platform (June 2026), including coverage of Nvidia’s GTC Taipei announcement (June 1, 2026). https://finance.yahoo.com/markets/stocks/articles/micron-hbm4-certification-nvidia-puts-151437687.html
[13] JEDEC, “JEDEC Publishes New DDR5 Standard for Advancing Next-Generation High Performance Computing Systems” (July 14, 2020). https://www.jedec.org
[14] TrendForce, “Tight DRAM Supply to Boost DDR5 Contract Prices — Profitability in 2026 Expected to Surpass HBM3e” (Oct. 29, 2025); Micron, Samsung, and SK Hynix quarterly earnings disclosures. https://www.trendforce.com/presscenter/news/20251029-12758.html
[15] Micron Technology, Inc., press release announcing the wind-down of the Crucial consumer brand (Dec. 3, 2025). https://investors.micron.com
[16] Micron Technology, Inc., fiscal Q3 2026 earnings-call prepared remarks (June 24, 2026). https://investors.micron.com
[17] U.S. Department of Justice, “Samsung Agrees to Plead Guilty and to Pay $300 Million Criminal Fine for Role in Price Fixing Conspiracy” (Oct. 13, 2005); U.S. Department of Justice, “Korean Company—Hynix—Agrees to Plead Guilty to Price Fixing and Agrees to Pay $185 Million Fine for Role in DRAM Conspiracy” (Apr. 21, 2005), https://www.justice.gov/archive/opa/pr/2005/April/05_at_207.htm; on Micron’s cooperation under the DOJ corporate leniency program, see “Inside the history of DRAM price-fixing lawsuits,” Tom’s Hardware (2026), https://www.tomshardware.com/pc-components/dram/samsung-sk-hynix-and-micron-face-a-third-dram-price-fixing-lawsuit.
[18] Pian Chen and Rui Huang, “Supply Discipline in the NAND Market: Any Antitrust Concerns?” Nutcracker Economics (Apr. 22, 2026). https://nutcrackereconomics.com/nand-supply-restriction/
About the Authors
* Pian Chen is Founder and Lead Economist at Nutcracker Economics. She has over 15 years of experience in litigation consulting and government oversight, with deep expertise in antitrust economics and financial fraud investigation. She previously held senior positions at leading litigation consulting firms and served as Associate Director of Economic Modeling at the Public Company Accounting Oversight Board (PCAOB). She holds a PhD in Agricultural and Resource Economics from UC Davis.
** Rui Huang is Principal Economist and Data Science Expert at Nutcracker Economics. She has over 20 years of experience in antitrust economics, applied econometrics, and data science, with expertise spanning causal inference, experimentation design, and marketing ROI measurement. She spent over eight years leading science teams at major technology companies, served as a Staff Economist at the U.S. Department of Justice Antitrust Division, and was a tenure-track professor at the University of Connecticut with 10+ peer-reviewed publications. She holds a PhD in Economics from UC Berkeley.
