Samsung DRAM Lawsuit Proceeds
Samsung, SK Hynix and Micron failed to dismiss a supply-conspiracy lawsuit on November 14, 2026, but the court has not found them liable. The Samsung DRAM lawsuit concerns dynamic random-access memory, the temporary working memory processors use while running applications. Buyers allege the manufacturers coordinated restrictions on conventional memory supply while shifting toward artificial intelligence customers.
U.S. District Judge Noël Wise found the allegations sufficient to proceed in the Northern District of California. The central conflict is coordinated supply restriction versus independent business decisions. The manufacturers argue that each company responded rationally to market conditions and its own commercial interests.
The ruling distinguishes this complaint from an earlier memory case that failed in 2022. It puts the companies’ production decisions under continued legal scrutiny without ordering additional output, controlling prices or awarding compensation. For hardware buyers, the immediate change is procedural: a challenge to how suppliers managed the AI transition remains alive.
What the Samsung DRAM lawsuit actually changed
The court allowed the case to continue; it did not decide that the manufacturers conspired.
The dismissal order addresses a proposed class action brought by 17 consumers and small businesses from five states. They filed their complaint on June 25, 2026.
A proposed class action seeks to represent a broader group of buyers, subject to court approval. This ruling does not certify that broader group. It also does not establish which purchases qualify for any eventual recovery.
The defendants asked the court to dismiss the complaint before the case progressed further. Their argument centered on independent decision-making: similar companies can make similar choices without agreeing to coordinate.
That distinction matters in a concentrated market. Manufacturers face overlapping customers, demand forecasts and technology transitions. Matching decisions can reflect shared circumstances rather than an unlawful arrangement.
At this stage, the judge assesses whether adequately pleaded allegations support a plausible claim. The court generally accepts those factual allegations as true for that limited purpose. That procedural assumption does not convert the complaint into verified history.
Wise concluded that the buyers had alleged enough to clear that threshold. Her analysis considered the overall pattern of claimed conduct, including production restrictions and decisions affecting the availability of conventional memory.
The complaint describes parallel shifts toward high-bandwidth memory, or HBM, a specialized form of DRAM used with AI accelerators. It also challenges decisions to wind down older memory products and introduce customer-vetting practices.
Those allegations now remain part of an active dispute. The manufacturers retain the opportunity to contest the facts, explain their decisions and challenge the buyers’ interpretation of the evidence.
The order requires the defendants to answer by December 7, 2026. It schedules an initial case management conference for January 12, 2027, at 9 a.m.
Neither deadline establishes when the case will end. An answer sets out a defendant’s responses to allegations; a case management conference addresses how litigation will proceed.
The practical significance is therefore narrower than a headline about a successful lawsuit might suggest. Buyers preserved their opportunity to pursue the case. They have not obtained a finding that their alleged losses resulted from unlawful coordination.
For readers following memory availability, that leaves two separate questions. The litigation concerns whether supplier conduct crossed a legal line. Purchasing decisions still depend on actual stock, product compatibility and delivery commitments.
The 91.5% figure measures revenue, not available memory
The headline market-share figure describes a historical revenue snapshot, not current production volume or proof of wrongdoing.
The complaint places Samsung, SK Hynix and Micron’s combined share at 91.5% of global DRAM revenue in the third quarter of 2025. The order recounts those allegations.
The individual figures are 33.2% for SK Hynix, 32.6% for Samsung and 25.7% for Micron. They describe the market during that quarter, rather than establishing the companies’ positions in November 2026.
Revenue share measures sales value. It does not directly measure the number of chips produced, manufacturing capacity or the amount of conventional memory available to a particular customer.
Product mix also matters when interpreting revenue. A supplier selling more specialized memory can generate a different revenue profile from one selling more conventional products. The combined percentage cannot resolve those differences.
That makes “91.5% of DRAM” an incomplete description. Readers need the measurement, period and attribution to understand what the number actually supports.
The figure helps explain why the companies’ decisions attract attention. When purchasing options depend heavily on a small supplier group, overlapping changes can affect buyers across multiple distribution channels.
The order describes conventional memory reaching customers through computer manufacturers, data center operators and companies that assemble memory modules. Those channels connect semiconductor production decisions to finished hardware.
A module brand therefore does not necessarily represent an additional source of underlying chip production. A buyer can encounter several retail brands while remaining exposed to decisions by the same upstream manufacturers.
Still, concentration alone does not establish an agreement. The judge treated the fundamental market structure as neutral because it fits both sides’ explanations.
An oligopoly, a market dominated by a few large suppliers, creates conditions in which competitors closely observe one another. It also allows independent firms to recognize the same incentives.
That is why the lawsuit requires more than a market-share calculation. The buyers must connect their theory to conduct that supports an inference of agreement, rather than merely describing a concentrated industry.
For procurement teams, the useful implication is supplier visibility. Understanding which manufacturer sits behind a module can clarify exposure when evaluating alternative vendors.
Consider a business comparing replacement memory from different distributors. Separate quotations establish commercial options, but they do not automatically establish separate semiconductor supply chains.
That example does not prove a shortage or conspiracy. It shows why revenue concentration can matter operationally while remaining insufficient as legal evidence.
The Samsung DRAM lawsuit turns on that distinction. The percentage supplies context; the alleged sequence of business decisions supplies the contested theory of coordination.
The AI pivot creates pressure beyond AI hardware
The commercial tension concerns who receives manufacturing attention when AI demand competes with established memory customers.
The buyers’ theory connects several alleged decisions. They say the manufacturers restricted conventional supply, shifted toward HBM and reduced commitments to memory products still serving existing customers.
The defendants offer a different explanation. They argue that these choices reflected independent responses to demand, technology changes and their own business interests.
External market research provides relevant context without settling that dispute. In July 2025, TrendForce described suppliers shifting capacity toward higher-end products and announcing plans to end certain older product lines.
Its analysis of the DDR4 supply transition also described strong remaining demand and limited short-term alternatives. DDR4 is an older generation of conventional memory used in computers and servers.
That combination explains the pressure on established customers. A product can move down a supplier’s investment priorities while remaining necessary for hardware already deployed.
For example, a company maintaining an existing server fleet needs memory compatible with those systems. A newer memory generation does not automatically solve that requirement.
The operational decision is therefore broader than choosing whichever chip looks most advanced. Buyers must weigh keeping installed equipment running against replacing systems that depend on older components.
This is a practical illustration, not evidence about any named purchaser’s losses. It shows why a production transition can affect customers who are not directly buying AI accelerators.
The same distinction applies to developers planning local computing equipment. Their immediate concern is whether a workstation configuration remains obtainable and supportable, regardless of the legal explanation for market conditions.
Micron’s public strategy provides a concrete comparison within the supplier group. Its December 3, 2025 announcement described an exit from the Crucial consumer business.
The company said it intended to improve supply and support for larger strategic customers in faster-growing segments. It linked its decision to AI-driven demand for memory and storage.
The consumer business exit announcement also promised continued warranty service and support. Micron said enterprise products would remain available through commercial channels.
That announcement documents Micron’s stated rationale. It does not independently establish why Samsung or SK Hynix made their decisions, or whether the companies reached an agreement.
The Micron AI memory pivot nevertheless illustrates the commercial tradeoff facing buyers. A supplier can prioritize strategic customers while creating uncertainty for customers attached to a departing channel.
The manufacturers do not need identical product portfolios for their choices to affect the same market. Micron’s consumer-brand decision differs from Samsung’s and SK Hynix’s channel arrangements.
That difference deserves attention rather than compression into a single claim that every supplier abandoned consumers. The lawsuit alleges a broader pattern; each company’s actual conduct still requires individual examination.
Why the earlier dismissal did not end this case
The judge found this complaint materially different from the earlier case, while continuing to apply its legal framework.
The manufacturers relied on DRAM II, an earlier dispute involving the same supplier group. In March 2022, the Ninth Circuit affirmed dismissal of that lawsuit.
The earlier DRAM ruling concerned alleged supply restrictions during an earlier market cycle. The appeals court found insufficient additional allegations to support a plausible inference of conspiracy.
The legal issue was not whether competitors had behaved similarly. It was whether the complaint supplied enough context to suggest an agreement behind that behavior.
Courts describe such additional circumstances as “plus factors,” meaning facts that help distinguish alleged coordination from independent parallel conduct. Their importance depends on the particular allegations.
That framework remains central here. A complaint cannot establish an unlawful agreement simply by showing that competing manufacturers responded similarly to a market downturn.
Wise nevertheless concluded that the current allegations presented a different overall picture. Her reasoning emphasized the claimed sequence, risks and persistence of the challenged supply decisions.
The buyers allege that the manufacturers continued restricting conventional supply even as conditions changed. They also challenge decisions that allegedly made those restrictions harder to reverse.
That is the commercial logic behind the complaint. If one supplier could profit by expanding while others held back, refusing that opportunity requires an explanation.
The defendants maintain that independent explanations exist. A court allowing the complaint forward does not mean those explanations have been disproved.
The judge also separated stronger allegations from weaker ones. That treatment matters because it shows the decision was not a blanket endorsement of the buyers’ account.
For older DDR3 memory, the complaint did not describe matching conduct across all three manufacturers. Micron allegedly continued limited production while Samsung and SK Hynix retired their lines.
The court therefore excluded the DDR3 allegations from the remainder of its parallel-conduct analysis. DDR3 is an earlier conventional memory generation than DDR4.
By contrast, the complaint placed the companies’ DDR4 wind-down announcements within approximately April through June 2025. Wise found those allegations sufficient to describe parallel conduct.
She also considered the alleged customer-vetting practices significant within the broader theory. Buyers claim that similar procedures appeared at the same time and supported coordination.
The defendants dispute the inference, pointing to demand and capacity pressures. The ruling does not establish that customer information actually passed between competitors.
For anyone seeking the Samsung memory lawsuit explained, this is the essential distinction: the court evaluated a pattern of allegations rather than identifying a proven agreement.
The older precedent supplied the test. The newer complaint, in the judge’s assessment, supplied enough additional context to proceed under it.
What the buyers still have to establish
A plausible account of coordination remains an allegation until evidence supports it under the standards governing later proceedings.
The strongest caution comes from the ruling itself. Wise did not accept every claimed indicator of conspiracy, even while finding the complaint sufficient overall.
Public statements about supply and investment did not support the necessary inference in her analysis. Companies routinely discuss forecasts and strategy with investors.
The court also declined to rely on the alleged industry-conference meetings. Opportunities for competitors to meet do not, by themselves, establish what they discussed or agreed.
Market concentration received similarly limited treatment. A small supplier group can support the buyers’ account, but it also explains why independent companies monitor similar signals.
These limits prevent the case from becoming a simple narrative about suspicious appearances. Shared vocabulary, overlapping conference attendance and matching commercial pressures cannot substitute for evidence of unlawful coordination.
The defendants’ alternative account deserves continued scrutiny on its own terms. Manufacturers can face different contractual obligations, investment schedules and customer requirements.
Those considerations are analytical possibilities, rather than findings about these defendants. Establishing which factors actually drove particular decisions requires a factual record.
The same caution applies to claims about profitability. Comparing conventional memory with HBM requires attention to the relevant period, product and commercial commitments.
A favorable short-term market signal does not automatically establish that every supplier could immediately redirect production. Nor does invoking a technology transition automatically explain every challenged restriction.
The dispute therefore requires more precision than saying AI consumed all available memory. It also requires more precision than saying rising demand should have produced an immediate supply response.
The Micron AI memory pivot shows why both sides can point to the same event. The company presented a strategic allocation decision; buyers incorporate that decision into a larger coordination theory.
Neither interpretation becomes established merely through repetition. The question is what the surrounding evidence supports about independent conduct, agreement and resulting harm.
Customer-vetting allegations present another important boundary. Similar questions can have ordinary commercial purposes, while particular practices might also support the buyers’ theory when considered with other evidence.
The order permits that theory to continue. It does not establish that the manufacturers exchanged the collected information or used it to enforce an arrangement.
Likewise, the proposed class remains distinct from any individual buyer’s experience. A purchasing problem does not automatically establish eligibility for relief or demonstrate a common injury.
No settlement, damages award or new criminal conviction follows from this decision. The order also does not direct the manufacturers to restore a product line or expand capacity.
The Samsung memory lawsuit explained accurately is therefore a case about a surviving theory of coordinated supply restriction. Its outcome remains unresolved, and its commercial implications require separate observation.
Watch the answers, the January conference and supply decisions
The next useful signals concern the defendants’ formal positions, the litigation schedule and observable changes in conventional memory supply.
The first signal is the defendants’ answers, due December 7, 2026. Those filings should clarify which allegations each company admits, denies or otherwise contests.
Readers should look for differences between the manufacturers rather than assuming a shared motion means identical underlying conduct. Each company’s response matters to the alleged pattern.
Detailed independent explanations would strengthen the need to evaluate decisions company by company. Continued denials alone would not resolve the dispute or erase the ruling.
The relevant change would be additional specificity: dates, distinctions between products and explanations for decisions the complaint groups together. Those details can sharpen the competing accounts.
The second signal is the January 12, 2027 case management conference. Its significance lies in how the court organizes the proceedings.
Discovery, the formal exchange of relevant information during litigation, will matter if and as the case develops. The conference is not itself a trial.
A defined schedule would make the next procedural steps easier to track. A revised schedule would change expectations about timing without establishing that either side’s factual account is correct.
Readers should therefore distinguish scheduling developments from substantive decisions. A deadline extension does not demonstrate a weak complaint, and an active schedule does not establish liability.
The third signal is what manufacturers actually say and do about conventional memory production. Product wind-down notices, customer commitments and capacity announcements deserve attention alongside the docket.
The useful question is whether new information addresses the particular constraints alleged in the complaint. Broad statements about investment do not necessarily describe output for affected conventional products.
A commitment to produce more specialized memory also differs from a commitment to maintain an older product generation. The distinction matters for buyers supporting existing hardware.
Observable additions to relevant conventional supply would weaken an overly broad claim that suppliers never respond to customer demand. They would not, by themselves, resolve allegations about earlier conduct.
Continued constraints would reinforce buyers’ operational concerns. They would still not prove that an agreement caused those constraints.
For enterprise buyers, the appropriate response is to keep legal developments separate from procurement assumptions. A surviving lawsuit does not create additional stock or guarantee a future delivery date.
A purchase decision should rest on current specifications, confirmed availability and contractual commitments. The lawsuit supplies a reason to examine supplier explanations carefully, rather than a timetable for market relief.
The same approach helps AI teams assess infrastructure plans. Memory availability belongs in deployment planning, but unresolved litigation should not become an assumed source of future capacity.
Follow the Samsung DRAM lawsuit through those three signals rather than treating survival at dismissal as a forecast of the verdict. Read the manufacturers’ answers for concrete differences in their explanations. Check the January conference for the actual timetable, then compare supply announcements with products buyers can obtain. Keep dated quotations, allocation notices and compatibility requirements together when evaluating a hardware purchase. A searchable knowledge base can help teams preserve that decision history. The next useful question is specific: what new evidence explains the challenged supply decisions, and what verified change improves availability for the hardware your team needs?



