Pian Chen and Rui Huang | August 21, 2026
DRAM prices are up roughly 700% and the headlines are talking about a $154 billion market. So how large could the damages in the 2026 DRAM class action actually be? Much smaller than those numbers suggest.
I. Introduction
In our July 7, 2026 blog article, we evaluated the plus factors alleged in Garciaguirre et al. v. Samsung Electronics Co., Ltd. et al., the class action accusing the three dominant DRAM makers — Samsung, SK Hynix, and Micron — of coordinating a restriction of commodity DRAM supply behind the cover of the AI-driven shift to High Bandwidth Memory.[1][2] That article asked whether the complaint’s evidence could support the inference of an agreement under Section 1 of the Sherman Act. This companion article asks the question that matters for valuation: how large could the recoverable damages be? Whether U.S. antitrust law reaches particular commerce is for the courts to decide. The economist’s task is scenario analysis — sizing the potential damages under each plausible legal outcome.
Our conclusion is that the potential damages could be much smaller than the headlines suggest. The complaint alleges that conventional DRAM contract prices rose roughly 700 percent between late 2024 and early 2026.[1] The 700 percent price jump reflects an imbalance between demand and supply. Shifting capacity to HBM in response to AI demand does not, by itself, violate the Sherman Act. The overcharge, if any, must be estimated with a defensible damages model and reliable but-for prices.
This article does not build a damages model. Its focus is the step that comes before any model: how much of the commerce at issue is protected by U.S. antitrust law. DRAM is a globally traded product, and most of it is bought and sold outside the United States. A federal statute — the Foreign Trade Antitrust Improvements Act of 1982 (FTAIA) — draws the boundary between the commerce U.S. antitrust law protects and the commerce it leaves to other countries’ laws, and that boundary limits the recovery base. Where the boundary sits is a live circuit split that the Supreme Court is now being asked to resolve in NHK Spring Co. v. Seagate Technology LLC.[3] A recent amicus brief by the International Center for Law & Economics (ICLE) frames the economic stakes, and we draw on it below.[4] On top of the FTAIA boundary, every claim in this case is an indirect-purchaser claim, so any recovery also depends on pass-through — a challenging empirical analysis that is likely to further reduce the damages.
The rest of this article proceeds as follows. Section II describes where DRAM is actually bought and sold. Section III explains the boundary of the Sherman Act. Section IV describes the circuit split — the Seventh Circuit’s Motorola decision and the Ninth Circuit’s recent negotiation-situs rule — and the economic logic at issue. Section V applies the framework to the DRAM complaint, sorting the commerce into two buckets with very different prospects. Section VI examines the pass-through question that runs through both buckets. Section VII sets the potential damages against the headlines and concludes.
II. Where DRAM Is Bought and Sold
DRAM is manufactured in a handful of fabs concentrated in South Korea, Japan, Taiwan, China, and the United States, and it is sold worldwide. From the fabs, DRAM reaches U.S. buyers through two very different channels — and the difference carries legal weight. The first channel is plainly covered by U.S. antitrust law; whether the law reaches the second is an open legal question — the circuit split described in Section IV — so it remains, for now, a question mark.
One channel runs into the United States itself: DRAM sold into this country as memory modules moving through U.S. distributors and retailers to consumers, system builders, and repair shops, and as server memory sold to U.S. data-center operators. The named plaintiffs sit in this channel. All seventeen — fourteen individuals and three small PC-building businesses — allege retail purchases of standalone, branded memory modules (Corsair, Crucial, Kingston, TeamGroup) from major U.S. retailers; the three businesses bought modules to build into custom computers for their customers.[1] None bought from the defendants directly. A module is itself a DRAM-containing product: the defendants’ chips reach module brands first and arrive at the plaintiffs only through at least one intermediate sale. This is an indirect-purchaser case from top to bottom.
The second channel is larger and never touches the United States at the point of sale. The majority of commodity DRAM by volume is purchased where electronic devices are assembled — by original equipment manufacturers (OEMs), contract manufacturers, and module makers operating principally in Asia. Consider a laptop bought in a U.S. store: its DRAM was typically sold abroad — the transaction, in the legal sense, occurred overseas, and the buyer was a foreign device maker — and the chips entered the United States only inside the finished product.
How large is each channel? Worldwide DRAM industry revenue in 2025 was roughly US$154 billion, summing TrendForce’s quarterly estimates.[5] That figure includes high-bandwidth memory (about 20 percent of DRAM revenue in 2024, and growing), so commodity DRAM — the product at issue — is smaller, though still enormous.[5] Direct U.S. imports of memory chips are a tiny slice: roughly US$2–3 billion a year (US$2.2 billion in 2021 per a USITC study; about US$2.8 billion in the latest twelve months), or about 2 percent of world memory-chip imports.[6][7]
The country breakdown tells the same story. Nearly half of the world’s memory-chip imports flow into China and Hong Kong, with Malaysia, Singapore, India, and Japan — the assembly hubs — next; Taiwan’s fast-growing DRAM exports go overwhelmingly to Mainland China, Hong Kong, and ASEAN, not to the United States.[7][8] The analogous LCD litigation points the same way: only about 1 percent of the price-fixed panels Motorola bought were purchased directly in the United States; 42 percent arrived embedded in finished phones; and 57 percent never reached U.S. commerce.[9] However measured, the second channel dwarfs the first.
III. The Boundary of the Sherman Act
The two channels’ very different prospects trace to a single statute. The FTAIA sets the general rule: the Sherman Act does not apply to foreign commerce. That general rule has a boundary and an exception — and they map onto the two channels. The boundary: import commerce — goods sold into the United States, like the first channel — is never excluded because it is domestic commerce, not foreign. The exception: foreign conduct — like the sales in the second channel — is nonetheless covered if it has a “direct, substantial, and reasonably foreseeable effect” on U.S. commerce, and that effect “gives rise to” the plaintiff’s claim — the domestic-effects exception.[10] The contrast, then: the first channel is protected by the Sherman Act; whether the exception brings the second channel within its reach is a legal question the courts will decide. This section traces where the boundary came from; the next describes the fight over where it sits.
The statute codified a boundary the courts had spent seven decades drawing. In American Banana Co. v. United Fruit Co. (1909), the Supreme Court held that the Sherman Act stopped at the U.S. border: conduct abroad was governed by the law of the country where it occurred, so a scheme carried out in Central America was beyond the Act’s reach even though its victim was American.[11] That territorial rule did not survive the growth of global commerce. In United States v. Aluminum Co. of America (1945), Judge Learned Hand held that the Sherman Act reaches agreements made abroad if they were intended to affect U.S. commerce and did so — the “effects test,” which moved the boundary from the location of the conduct to the location of the harm.[12]
The effects test then needed a limit of its own, and Congress supplied one in 1982: the FTAIA confines the Sherman Act’s foreign reach to conduct with direct, substantial, and reasonably foreseeable U.S. effects. The Supreme Court confirmed that limiting purpose in F. Hoffmann-La Roche Ltd. v. Empagran S.A. (2004), rejecting claims by foreign vitamin purchasers injured in foreign transactions.[13] Foreign harms from foreign transactions belong to foreign law — even when the same cartel also harms Americans.
One procedural point magnifies the stakes. Courts historically treated the FTAIA’s requirements as jurisdictional, so a foreign-commerce claim could be tested and dismissed at the threshold. Most circuits now treat them as substantive elements of the claim. The shift traces to Arbaugh v. Y&H Corp. (2006), which told lower courts to stop treating statutory limits as jurisdictional unless Congress clearly ranked them so.[14] The en banc Seventh Circuit reclassified the FTAIA’s requirements as elements of the claim in Minn-Chem, Inc. v. Agrium Inc. (2012), and most circuits have followed.[15] The practical consequence: elements are decided on the merits, so whether U.S. law even reaches the commerce at issue may not be resolved until after years of full antitrust discovery.[4] For defendants, that converts a legal boundary into a costly factual battle; for plaintiffs, it means surviving the motion to dismiss does not settle how much commerce is actually in the case.
IV. The Circuit Split: Motorola and NHK Spring
Whether U.S. law reaches the second channel is where the circuits have split, and the split is why the Supreme Court is being asked to step in: identical foreign purchases are outside the Sherman Act under the Seventh Circuit’s rule but inside it under the Ninth Circuit’s, and only the Supreme Court can settle which rule governs. The answer matters for the DRAM case directly — it will determine whether the second, far larger channel can enter the damages base at all.
The Seventh Circuit’s Motorola Mobility decision (2015) is the leading application of these principles to a component cartel that reaches most consumers only inside finished products. Motorola’s foreign subsidiaries bought price-fixed LCD panels abroad, built them into phones, and sold many of the phones into the United States. Judge Posner assumed the cartel had a direct, substantial, and reasonably foreseeable effect on U.S. commerce; the claims failed on the domestic-effects exception’s second requirement: the overcharges were paid abroad, by the foreign subsidiaries — the direct purchasers of the panels — so the injury was a foreign injury, and the cartel’s U.S. effects were not what gave rise to those claims. Whatever harm later reached U.S. phone buyers was derivative of that foreign injury. In the court’s words, “U.S. antitrust laws are not to be used for injury to foreign customers.”[9]
The Ninth Circuit recently went the other way on materially similar facts. In the Hard Disk Drive Suspension Assemblies litigation,[16] Seagate’s Thai and Singaporean affiliates bought price-fixed components abroad, for delivery abroad — but the master supply agreement and quarterly price negotiations took place in the United States. The Ninth Circuit held that the domestic situs of the negotiations was enough to bring those wholly foreign purchases within the Sherman Act.[17] The defendants have petitioned the Supreme Court for review. Briefing on the petition was completed this summer, and the ICLE amicus brief supports the petitioners.[3][4]
The economic argument in the ICLE brief is worth stating precisely, in the brief’s own words. “Stripped of labels, an antitrust injury is an overcharge above the otherwise competitive price,” and the FTAIA’s effects requirement is “in economic terms, a requirement that this overcharge be borne in American commerce.” The cognizable effect, on this view, is an overcharge paid in a market — defined as a product market and a geographic market. The location of the bargaining table is another matter: “[t]he arbitrary locations of the negotiators during either in-person or multi-location electronic negotiations is, in economic terms, irrelevant.” The same overcharge, paid by the same Asian buyer for the same goods delivered to the same Asian port, does not become an American injury because the meeting happened in San Jose rather than Singapore. The brief warns, “[a] rule that makes liability turn on an economically irrelevant variable is not a limiting principle. It is an invitation to manipulate that variable.”[4]
V. Applying the Framework to the DRAM Complaint
Where does the June 2026 DRAM case sit in this landscape? Sorting the DRAM commerce into the FTAIA’s two categories produces two buckets with very different prospects.
Bucket one: DRAM sold into the United States
Bucket one is the DRAM sold into the United States: memory modules sold into U.S. channels — where all the named plaintiffs bought — and chips the defendants’ U.S. subsidiaries sold directly to U.S. OEMs and data-center operators. That commerce sits comfortably within the Sherman Act. An overcharge paid by a U.S. purchaser in a U.S. transaction is a domestic injury under any circuit’s rule. On this slice of commerce, the FTAIA poses no obstacle; the plaintiffs’ challenge remains the one we analyzed in the companion article — proving that an agreement, rather than parallel unilateral conduct, caused the price increases.
Bucket two: DRAM sold abroad and embedded in imported devices
The far larger volume of commerce — DRAM sold to OEMs abroad and imported inside finished devices — is where the geography bites. Under Motorola’s logic, the overcharge on that DRAM was paid abroad by foreign purchasers, and those purchasers’ injuries belong to foreign law. U.S. consumers who later bought the devices face a second, independent barrier: the Illinois Brick indirect-purchaser rule, which reserves federal damages claims to direct purchasers precisely to avoid tracing overcharges down distribution chains.[18] A consumer who buys a laptop is at least two steps removed from the DRAM maker — an indirect purchaser twice over. State indirect-purchaser statutes permit what federal law forecloses — the route this complaint takes.
The named plaintiffs all sit in the first bucket — standalone modules bought at U.S. retail. But the proposed classes are broader, covering indirect purchasers of conventional DRAM both as standalone modules and as a component inside finished devices such as laptops, desktops, phones, game consoles, and servers.[1] The mismatch is itself a certification issue: because every named plaintiff sits in bucket one, defendants will argue that the named plaintiffs cannot adequately represent the class members in bucket two — the device purchasers, whose DRAM was sold abroad and whose claims face both the legal question described in Section IV and a longer pass-through chain.
Where does the FTAIA fit? First, the defendants’ U.S. subsidiaries — Samsung Semiconductor and SK Hynix America, both in San Jose — sell directly to major U.S. OEMs and data-center operators; those sales are import commerce, inside bucket one, and beyond the FTAIA’s reach entirely.[1] Second, the complaint alleges that contract prices are negotiated quarterly. To the extent those negotiations happened in the United States, the Ninth Circuit’s Seagate rule could sweep even bucket two — chips sold abroad to foreign device makers — into the Sherman Act. That is the negotiation-situs fight described in Section IV. The case is pending in the Northern District of California, where Seagate currently governs; the real uncertainty is the Supreme Court, whose decision in NHK Spring — if it takes the case — will set the rule.
VI. The Pass-through Question
Potential damages are, at bottom, the product of two components for each channel: the volume of commerce U.S. antitrust law reaches, and the share of the overcharge that was passed through to U.S. purchasers. Sections II through V sized the first component; this section turns to the second.
The pass-through question runs through the entire supply chain, and it matters for both buckets. Because every claim in this case is an indirect-purchaser claim, even bucket one requires pass-through analysis: the overcharge on the defendants’ chips must be traced through module brands and retailers to the consumers and PC builders who sue. Bucket two adds more links to the same chain: from chips sold abroad, through OEM device prices, to U.S. retail shelves. In either case, pass-through is an empirical matter, not an assumption. It depends on the competitive structure of each link in the supply chain (component makers, module brands and OEMs, distributors, retailers), on contract terms and inventory cycles, and on demand conditions in the downstream market.
Pass-through rates can range from near zero to above one hundred percent, and they can differ by product and by period. Apple is a good example to explain this point. Apple raised Mac and iPad prices in June 2026, and Tim Cook told investors in July 2026 that Apple had “reluctantly raised prices” in the face of what he called “a 100-year flood on memory pricing” — announcements the complaint itself cites as evidence of the squeeze.[1][19][20] The economics behind the shift is telling. A year ago, memory accounted for roughly 10 percent of a flagship iPhone’s bill of materials — small enough that Apple could, and apparently did, absorb DRAM price increases rather than pass them on. By the third quarter of 2026, memory had become the single largest cost component at roughly 34 percent, with industry projections above 40 percent for the first half of 2027 — large enough that absorption gave way to visible price increases.[21] Notably, the increases so far have come on Macs and iPads, not iPhones. Pass-through, in other words, has differed across products, over time, and with the strategic choices of a single firm — which is precisely why it must be measured, not assumed.
Measuring them requires third-party transactional data from each level of the supply chain, and it is a genuinely hard econometric problem. Any damages figure that treats the headline DRAM price increase as if it were borne dollar-for-dollar by U.S. purchasers has skipped the hardest and most consequential step. If this case ever reaches the merits phase, the pass-through analysis will be the main battle fought between the parties’ economic experts.
VII. Damages Could Be Much Smaller Than the Headlines Suggest
As Section I previewed, neither the headline 700% price increase nor worldwide revenue is the base for a damages estimate. The recoverable base is built bucket by bucket — and it shrinks at every step.
- Bucket one: DRAM sold into the United States — is the case’s floor. It is legally secure but economically modest. Direct U.S. memory-chip imports run in the low single-digit billions of dollars a year; the damages classes cover five states; and impact must still be traced through module brands and retailers to the buyers.[6][7]
- Bucket two: DRAM sold abroad and embedded in devices — is the case’s ceiling. It enters the damages base only if the Ninth Circuit’s negotiation-situs rule survives Supreme Court review (or the domestic-effects exception is satisfied); only for the five damages states; and only to the extent pass-through from foreign chip sales to U.S. shelf prices is proven. The state-law workarounds to Illinois Brick are what make these claims possible at all.
The economist’s product here is a scenario analysis, not a verdict: the legal rulings select the scenario, and the data determine its size. For attorneys and litigation funders evaluating this case — or any component cartel case with a global supply chain — the practical lesson is to size the cognizable U.S. commerce before valuing the claim.
Two pieces of economic work belong in that evaluation, and they belong to different stages. The first is industry research locating the commerce — how much commodity DRAM was sold into the United States versus abroad, and through which channels — which, as Section II illustrates, can be approximated from trade statistics and commercial industry data at the outset. The second, if data become available, is the pass-through analysis quantifying how much of any foreign overcharge reached U.S. buyers; because pass-through requires transactional data from each level of the supply chain, it is merits-phase work rather than an early-stage screen. Together with the plus-factor analysis from our companion article, these analyses define both dimensions of the case’s risk: whether liability can be established, and how much of the harm a U.S. court can compensate. A case can be strong on one dimension and thin on the other. This one, on the public record, faces real questions on both.
In short, U.S. antitrust law does not compensate every injury a global cartel inflicts — it compensates injuries to U.S. commerce. Where the courts draw that line is for the courts to decide. The economist’s job is to size the damages under each plausible ruling — and the questions underneath every scenario are the same: where was the overcharge paid, and how much of it passed through to U.S. buyers? Those are questions economists, not pleadings, will answer — if the data to answer them become available.
Sources
[1] Complaint and docket, Garciaguirre et al. v. Samsung Electronics Co., Ltd. et al., No. 3:26-cv-06345 (N.D. Cal., filed June 25, 2026), https://www.courtlistener.com/docket/73532135/garciaguirre-v-samsung-electronics-co-ltd/.
[2] Pian Chen and Rui Huang, “The 2026 DRAM Antitrust Class Action: An Early-Stage Economic Evaluation,” Nutcracker Economics (July 7, 2026), https://nutcrackereconomics.com/the-2026-dram-antitrust-class-action-an-early-stage-economic-evaluation/.
[3] NHK Spring Co., Ltd. v. Seagate Technology LLC, No. 25-1358 (U.S., petition for certiorari filed June 4, 2026), docket: https://www.supremecourt.gov/search.aspx?filename=/docket/docketfiles/html/public/25-1358.html;
Brief in Opposition filed July 22, 2026, https://www.supremecourt.gov/DocketPDF/25/25-1358/417066/20260722144604980_NHK%20v.%20Seagate%20-%20BIO%20-%20Final.pdf.
[4] Brief of Amicus Curiae International Center for Law & Economics Supporting Petitioners, NHK Spring Co. v. Seagate Technology LLC, No. 25-1358 (U.S. July 15, 2026), https://www.supremecourt.gov/DocketPDF/25/25-1358/416607/20260715204339320_25-1358%20Amicus%20Brief%20Intl%20Ctr%20Law%20Economics.pdf.
[5] TrendForce, quarterly global DRAM industry revenue releases for 2025 (1Q25: ~US$27.0 billion; 2Q25: ~US$31.6 billion; 3Q25: ~US$41.4 billion; 4Q25: ~US$53.6 billion) and “Memory Industry Revenue Expected to Reach Record High in 2025” (July 22, 2024) (HBM at ~20 percent of DRAM revenue in 2024), https://www.trendforce.com/presscenter/news/20250603-12603.html; https://www.trendforce.com/presscenter/news/20250902-12694.html; https://www.trendforce.com/presscenter/news/20251126-12802.html; https://www.trendforce.com/presscenter/news/20260226-12937.html; https://www.trendforce.com/presscenter/news/20240722-12228.html.
[6] Lin Jones et al., “U.S. Exposure to the Taiwanese Semiconductor Industry,” USITC Working Paper (Nov. 2023) (U.S. memory-chip imports of ~US$2.2 billion in 2021, about half from Taiwan), https://www.usitc.gov/publications/332/working_papers/us_exposure_to_the_taiwanese_semiconductor_industry_11-21-2023_508.pdf.
[7] Global Trade Analytics (GTAIC), “Electronic Memories: Top-40 Importing Countries” (HS 854232, latest twelve months: China/Hong Kong ~US$59.4 billion, ~47 percent of top-40 imports; United States ~US$2.77 billion), https://gtaic.ai/market-reports/electronic-memories-breakdown-top-40-importing-countries-world-in-2026.
[8] Taiwan Ministry of Finance, Statistical Bulletin No. 20 (2025), Memory Exports and Imports, https://service.mof.gov.tw/public/Data/statistic/bulletin/114/2025_20_Memory_Exports_and_Imports.pdf.
[9] Motorola Mobility LLC v. AU Optronics Corp., 775 F.3d 816, 820 (7th Cir. 2015), https://law.justia.com/cases/federal/appellate-courts/ca7/14-8003/14-8003-2015-01-12.html.
[10] Foreign Trade Antitrust Improvements Act of 1982, 15 U.S.C. § 6a, https://www.law.cornell.edu/uscode/text/15/6a.
[11] American Banana Co. v. United Fruit Co., 213 U.S. 347 (1909), https://supreme.justia.com/cases/federal/us/213/347/.
[12] United States v. Aluminum Co. of America, 148 F.2d 416 (2d Cir. 1945), https://law.justia.com/cases/federal/appellate-courts/F2/148/416/1503668/.
[13] F. Hoffmann-La Roche Ltd. v. Empagran S.A., 542 U.S. 155 (2004), https://supreme.justia.com/cases/federal/us/542/155/.
[14] Arbaugh v. Y&H Corp., 546 U.S. 500 (2006), https://supreme.justia.com/cases/federal/us/546/500/.
[15] Minn-Chem, Inc. v. Agrium Inc., 683 F.3d 845 (7th Cir. 2012) (en banc), https://law.justia.com/cases/federal/appellate-courts/ca7/10-1712/10-1712-2012-06-27.html.
[16] In re Hard Disk Drive Suspension Assemblies Antitrust Litigation, No. 3:19-md-02918 (N.D. Cal.), https://www.courtlistener.com/docket/16307295/in-re-hard-disk-drive-suspension-assemblies-antitrust-litigation/.
[17] Seagate Technology LLC v. NHK Spring Co., Ltd., No. 24-4470 (9th Cir. Jan. 8, 2026), https://cdn.ca9.uscourts.gov/datastore/opinions/2026/01/08/24-4470.pdf.
[18] Illinois Brick Co. v. Illinois, 431 U.S. 720 (1977), https://supreme.justia.com/cases/federal/us/431/720/.
[19] “Tim Cook Says Apple Price Increases Are ‘Unavoidable’ Due to Memory Costs,” MacRumors (June 17, 2026), https://www.macrumors.com/2026/06/17/apple-increasing-prices/.
[20] Tim Cook, remarks on Apple’s Q3 FY2026 earnings call (July 30, 2026), as reported by MacRumors, https://www.macrumors.com/2026/07/30/tim-cook-on-apple-price-increases/.
[21] TrendForce bill-of-materials estimates, as reported by Notebookcheck (Aug. 10, 2026), https://www.notebookcheck.net/TrendForce-says-iPhone-18-Pro-s-bill-of-materials-cost-rises-almost-40-as-memory-prices-surge.1364498.0.html.
About the Author
* 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.
