Memory Shortages Reshape AI Hardware Marketplace
A California class action filed June 25, 2026 (*Garciaguirre v. Samsung Electronics*) accuses Samsung, SK hynix, and Micron, who together control roughly 90% of the global DRAM market, of coordinating a shift toward high-bandwidth memory for AI accelerators as cover to curtail DDR3 and DDR4 production, driving commodity RAM prices up an alleged 700% over four years. The same AI-driven memory crunch pushed Apple and Microsoft to raise device prices this week, according to CNBC, while smaller hardware makers like GoPro warn the cost spike threatens their survival. A nearly identical 2018 price-fixing suit against the same three companies was dismissed and upheld on appeal in 2022 for lacking evidence of an actual agreement; this complaint leans on the AI-driven HBM pivot as the new evidence the earlier case lacked. For practitioners, tightening DDR supply and hyperscaler-favored allocation directly raise the cost of on-prem and edge AI hardware.
The detail practitioners should sit with here isn't the price spike itself, it's the mechanism the new lawsuit alleges: that the memory industry's pivot to AI-oriented high-bandwidth memory (HBM) is being used as cover to shrink supply of the ordinary DDR3 and DDR4 modules that power everyday laptops, workstations, and edge devices. If that theory holds up in court, it reframes today's DRAM shortage as a direct, structural side effect of the AI buildout rather than a temporary supply-chain hiccup.
What happened
Samsung, SK hynix, and Micron were sued on June 25, 2026 in the U.S. District Court for the Northern District of California; 17 plaintiffs, 14 individuals and three small PC businesses including Troy's Computers and My Florida PC, accuse the three memory makers of illegally coordinating to restrict DRAM supply and inflate prices that the complaint says have risen roughly 700% over four years. The case, filed as Garciaguirre v. Samsung Electronics and assigned to Judge Noel Wise, invokes Section 1 of the Sherman Act. The complaint argues the companies used their coordinated shift toward HBM, the stacked DRAM that feeds AI accelerators, as cover to curtail production of older DDR3 and DDR4 modules, and cites Apple's recent iPad and Mac price increases as evidence of the squeeze; plaintiffs are seeking class status, an injunction, and treble damages. Samsung and SK hynix have pleaded guilty to criminal DRAM price fixing once before, with SK hynix paying a $185 million fine in April 2005.
Background
This is not the first time these three companies have faced this exact claim. A 2018 class action brought by law firm Hagens Berman made similar allegations of coordinated production cuts; a district court dismissed it in 2020, and the Ninth Circuit upheld that dismissal in 2022, ruling the companies' conduct was "more likely explained by lawful, unchoreographed free-market behavior" than an illegal agreement, since the plaintiffs' evidence didn't clear Section 1's threshold for showing an actual agreement rather than ordinary parallel behavior in a three-supplier market. The new complaint leans on the AI-driven pivot to HBM as the additional evidence the earlier case lacked, but it will need to clear the same evidentiary bar to survive a motion to dismiss.
Market context
The lawsuit lands amid a broader, well-documented memory crunch. Apple and Microsoft both announced consumer price increases the same week, and CNBC reports the pain is far worse for companies without their cash cushion and supply-chain leverage: IDC analyst Nabila Popal said smaller buyers "won't be able to get the memory because memory suppliers are only answering calls of the big players." CNBC cites Mono Technologies, a small networking company whose cost for 8 gigabytes of Micron DRAM rose from $35 to $300, and GoPro, which warned this month it could go out of business after memory costs rose 80-115% in a single quarter. The Verge's Terrence O'Brien, citing NYU Stern professor Srikanth Jagabathula, frames the underlying allocation logic simply: "the same chip earns far more inside an AI server than inside a consumer device," which is why vendors prioritize data-center customers over everyone else.
For practitioners
Teams building or maintaining on-prem workstations, edge inference devices, or any hardware that depends on commodity DDR memory should expect continued price and availability pressure independent of how the lawsuit resolves, since the underlying demand shift toward HBM-fed AI accelerators is real and well-documented regardless of whether it is found to be an illegal coordination. Budgeting for workstation refreshes and edge deployments should assume elevated RAM costs persist, and teams with flexibility should weigh locking in hardware purchases or contracts earlier rather than waiting for prices to normalize.
What to watch
Whether Samsung, SK hynix, and Micron file a motion to dismiss citing the 2022 Ninth Circuit ruling, and whether the HBM-pivot theory is enough to distinguish this case from the one that failed; any capacity-expansion announcements from memory makers that would ease DDR3/DDR4 supply; and further consumer-price moves from device makers as the underlying memory crunch continues.
Key Points
- 1A June 25 class action accuses Samsung, SK hynix, and Micron of using an AI-driven HBM pivot to curtail DDR supply and inflate prices 700%.
- 2A near-identical 2018 suit against the same firms was dismissed and upheld on appeal in 2022 for lacking evidence of an actual agreement.
- 3Independent of the lawsuit's outcome, AI-driven memory demand is already raising real hardware costs for on-prem and edge deployments.
Scoring Rationale
Well-corroborated and genuinely useful for practitioners: a real antitrust complaint with a specific, AI-driven theory (HBM pivot as cover for DDR curtailment), balanced by the important context that a nearly identical claim already failed in court. Modest downward adjustment from 7.0 since this is fundamentally a market/legal story, not a technical development, and the core price-pressure facts were already well known before this specific lawsuit.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems


