Three memory makers, one shortage: the lawsuit that puts DRAM on trial

A federal antitrust complaint filed in the United States on 28 June 2026 accuses Samsung, Micron and SK Hynix of deliberately throttling DRAM production to keep memory prices elevated, according to a Telegram post by Euronews dated 29 June 2026, 08:14 UTC. The lawsuit, brought as a class action, treats the global memory shortage not as a market accident but as coordinated behaviour by the three firms that together supply the overwhelming majority of the world's DRAM.

The case lands while the market itself is signalling distress. A research note circulated by Jefferies and surfaced on X on 28 June 2026, 12:04 UTC, forecasts DRAM and NAND prices climbing 40–50% in the third quarter of 2026 and another 30–40% in the fourth quarter, with no meaningful relief expected before 2028. The plaintiffs are betting that what looks like a supply crunch will, in deposition, look like a cartel.

The claim, as the plaintiffs frame it

The complaint, as paraphrased in the Euronews wire and echoed by the X account @pirat_nation at 21:44 UTC on 28 June 2026, alleges that Samsung, SK Hynix and Micron acted in concert to "keep DRAM production artificially low." The legal hook is the American Sherman Act, which treats parallel conduct plus a plausible shared motive as grounds for civil liability once a price effect is visible. Memory pricing is unusually well-suited to this kind of case: there are only three credible suppliers of commodity DRAM at scale, pricing data is dense, and the inputs are largely fungible across vendors.

For the plaintiffs, that market structure is the tell. Three firms controlling the bulk of a market do not need to meet in a back room to coordinate output; industry conferences, public capacity guidance and mutual read-throughs of customer order books can do the same work. The complaint's premise is that the restraint is visible in the data.

Why this is harder than it looks

The defence playbook for an oligopoly of three is well-known and well-funded. The firms can argue — and historically have — that high prices simply reflect the cost of building new fabs, that capacity additions are running ahead of AI-driven demand, and that any production cuts were rational responses to the 2023–24 inventory glut that left the industry with billions of dollars of writedowns. Jefferies' own forecast, with relief not arriving until 2028, implies that the firms themselves do not need a conspiracy to make scarcity profitable; the demand backdrop does the work.

There is also a procedural hurdle. American class actions routinely struggle to prove injury-in-fact on a class-wide basis when buyers are diverse and prices pass through different channels. The defendants will press for individualised proof of damages, which can fragment a case into thousands of mini-trials. The plaintiffs' lawyers will need to lean on expert econometrics to draw a clean line between "oligopoly pricing" and "collusive pricing."

What it means for buyers, and for the chip cycle

If the case gains traction, the near-term effect on buyers is more legal discovery than lower prices. Discovery would expose the firms' internal forecasts, capacity plans and customer communications, which can be unflattering even where no crime occurred. Even a settlement would likely carry behavioural remedies: restrictions on how the three coordinate industry-wide capacity decisions, which in practice govern how an entire capex cycle unfolds.

For the chip cycle itself, the lawsuit arrives at an awkward moment. AI demand has pulled DRAM into the same strategic conversation as advanced logic. Memory is no longer a cyclical commodity bought by PC makers; it is a chokepoint for training and inference infrastructure. Prices that double in two quarters, as Jefferies projects, change the economics of building data centres and reshape who can afford to scale.

The stakes

If the plaintiffs' theory holds, the three firms have been levying a private tax on the AI build-out — and on every consumer electronics buyer downstream. If it does not, the case still does useful work: it forces transparency on capacity decisions that have, for two decades, been treated as commercially confidential. Either way, the complaint has put the global memory market on notice that the era in which three firms can manage a global shortage in private may be closing.

What remains genuinely uncertain is whether the litigation moves at the speed of the market it is challenging. Buyers paying 2026 prices will not be comforted by a 2028 ruling, and the Jefferies forecast implies the worst of the squeeze is still ahead. The case asks the courts to do in years what supply and demand will do in quarters — and that is a fight the plaintiffs will struggle to win on timing alone, even if they win it on the merits.

Desk note: Monexus frames this as an antitrust question first and a chip-cycle story second. The Western wire coverage tends to treat the shortage as a market fact; the lawsuit treats it as a choice. We are running both readings, with the plaintiffs' allegations clearly attributed, and we will track the firms' filings as they enter the public record.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://t.me/euronews/