If you have tried to buy a stick of RAM in 2026, you already know the punchline: memory has become shockingly expensive. A 32GB DDR5 kit that sold for around $149 last year now runs closer to $239, and server-grade contract prices have more than doubled. Now a group of buyers is arguing this was not just supply and demand — it was a coordinated scheme. On June 25, 2026, a federal class-action lawsuit accused the world’s three dominant memory makers — Samsung, SK Hynix, and Micron — of conspiring to keep DRAM artificially scarce and prices artificially high.
This article breaks down what the lawsuit actually claims, how much prices really moved, what the companies did with their production, and — the question that matters most — whether the case has any real chance of winning.
Disclaimer: This is an explainer for general information, not legal or investment advice. The allegations described here are unproven claims in an active lawsuit. All three companies deny wrongdoing.
What the lawsuit actually says
The complaint was filed in the U.S. District Court for the Northern District of California by 17 named plaintiffs — a mix of individuals and small businesses who buy memory. It alleges a violation of Section 1 of the Sherman Act, the core U.S. law against agreements that restrain trade.
The heart of the accusation is this: beginning around October 2022, the three manufacturers allegedly coordinated to restrict the supply of conventional, commodity DRAM — the standard memory in laptops, desktops, and ordinary servers — in order to push prices up. According to the suit, the mechanisms included:
- Coordinated production cuts on mainstream DRAM
- Shifting capacity toward high-margin High-Bandwidth Memory (HBM) for AI servers
- Exiting product lines (for example, older DDR3/DDR4 modules)
The plaintiffs claim this coordinated tightening drove commodity DRAM prices up by roughly 700% over four years — the figure that gave the episode its nickname online: the “RAMpocalypse.”
How much did DRAM prices actually rise?
The lawsuit’s “700%” is the plaintiffs’ framing over a long window. The independently reported numbers are less dramatic but still stunning:
| Period | Reported price movement |
|---|---|
| 2025 (year to date) | ~50% increase |
| Q4 2025 | Additional ~30% (Counterpoint) |
| Q1 2026 contract | +55–60% conventional DRAM; server DRAM +60%+ (TrendForce) |
| Q3 2026 | Samsung reportedly seeking up to 70% server memory hikes |
| Specific example | 32GB DDR5 module: $149 → $239 (~60%) |
Contract pricing for DDR5 has, by some accounts, surged more than 100%. However you slice it, memory has roughly doubled in a very short time — an almost unheard-of move for a commodity component.
The “collusion” question: what the companies actually did
Here is where the story gets more complicated than a simple villain narrative. The three companies did, in fact, reduce the supply of commodity DRAM. That part is not really in dispute. The dispute is why.
The three makers control roughly 90% of the global DRAM market:
| Company | DRAM revenue share (Q1 2026) |
|---|---|
| Samsung | 38% |
| SK Hynix | 29% |
| Micron | 22% |
What changed is where that capacity went. Instead of pumping out cheap commodity DRAM, all three prioritized HBM — the specialized, stacked memory that AI accelerators like Nvidia’s GPUs depend on. HBM carries far higher margins. The results are visible:
- SK Hynix says its memory capacity is essentially “sold out” for 2026.
- Micron exited the consumer memory market entirely, redirecting output to enterprise and AI customers.
So the same fabs that would have made your laptop’s RAM are now making memory for AI data centers instead. TrendForce estimates memory supply will grow 23% in 2026 while demand grows 35% — a structural shortage on its face.
The defendants’ argument writes itself: this is not a conspiracy, it is a rational, independent response to the biggest demand shock in the industry’s history. Why sell low-margin commodity chips when every wafer of HBM you can make is spoken for?
The plaintiffs’ counter: three companies, moving in near-perfect lockstep, quietly starving a market they collectively control 90% of, is exactly what illegal coordination looks like.
Does the lawsuit have a real chance? The precedent cuts both ways
This is the section most headlines skip. The honest answer: the case faces an uphill battle, and history is the reason.
The bad precedent (for plaintiffs): 2018
In 2018, these same three companies were hit with a nearly identical class action alleging DRAM price-fixing. A federal judge dismissed it, and in 2022 the appeals court affirmed, ruling that the behavior reflected “conscious parallelism” rather than collusion.
That legal phrase is the whole ballgame. In a market with only three players, everyone can watch everyone else. If all three independently decide to cut supply because it is individually profitable, they will move together — and moving together, by itself, is not illegal. To win, plaintiffs must show an actual agreement, not just parallel behavior.
The good precedent (for plaintiffs): the 2000s
There is a reason the plaintiffs are trying anyway. In the 2000s, the U.S. Department of Justice won a criminal price-fixing case against DRAM makers — because it had direct evidence: emails and explicit agreements. Samsung paid roughly $300 million and SK Hynix about $185 million, and several executives went to prison.
So the pattern is clear:
- Direct evidence (a “smoking gun” email or agreement) → plaintiffs can win.
- Only parallel conduct and circumstantial timing → courts dismiss it as conscious parallelism.
What to watch next
Legal observers expect the defendants to file a motion to dismiss soon. That motion is the case’s first and most important gate:
- If dismissed (the 2018 outcome repeating), the lawsuit effectively ends and market impact is minimal.
- If it survives and reaches discovery — where plaintiffs can subpoena internal communications — the risk profile changes entirely, because discovery is exactly where a 2000s-style smoking gun could surface.
What it means for investors and buyers
For investors: Wall Street is, so far, treating this as a legal overhang rather than a fundamental threat. Micron — the only primarily U.S.-listed defendant — is up roughly 300% year to date, driven by the AI memory boom. The market’s implicit bet is that the case gets dismissed like its 2018 predecessor. The real risk is not a verdict years away; it is the possibility that discovery unearths something. (Again: not investment advice.)
For buyers and PC builders: the lawsuit does nothing for your wallet in the short term. Even the plaintiffs are not claiming prices will fall soon. With HBM demand still outrunning supply and the makers openly prioritizing AI customers, elevated DRAM prices look likely to persist well into 2026 and possibly beyond.
The bottom line
Did Samsung, SK Hynix, and Micron rig the RAM market? The uncomfortable truth is that both stories are consistent with the facts. The companies really did cut commodity supply, and prices really did explode. But an AI-driven gold rush is a perfectly innocent explanation for the exact same behavior — and in 2022, a court already accepted that explanation.
Unless the plaintiffs can produce direct evidence of an actual agreement, the smart money says this case follows its 2018 predecessor to dismissal. The one thing that could change everything is discovery. Until then, the “RAMpocalypse” is best understood not as a proven conspiracy, but as the collision of an AI boom and a market so concentrated that healthy competition and quiet collusion look almost identical from the outside.
FAQ
Why is RAM so expensive in 2026? Primarily because Samsung, SK Hynix, and Micron shifted manufacturing capacity toward high-margin HBM for AI servers, tightening the supply of ordinary DRAM. A lawsuit alleges this was coordinated; the companies say it is a response to AI demand.
Who is being sued and where? Samsung, SK Hynix, and Micron, in a class action filed June 25, 2026 in the U.S. District Court for the Northern District of California, under Section 1 of the Sherman Act.
Will the lawsuit lower memory prices? Not in the near term. Even if it succeeds, any remedy would take years, and the underlying HBM-driven shortage is expected to keep prices high through 2026.