Nearly every laptop, phone and PC runs on a type of memory chip called DRAM. Samsung, SK hynix and Micron make close to 90 percent of the world’s supply. A class action lawsuit accuses all three of illegally coordinating to restrict supply and push prices higher, using the AI boom as cover.
The suit was filed on June 25, 2026, in U.S. federal court. It claims the three shifted production toward a stacked memory chip built for AI data centers, called high-bandwidth memory, which earns far richer margins than ordinary memory. A chip factory has only so much capacity, so whatever goes to AI memory comes out of DDR3 and DDR4, the standard RAM chips in most PCs.
According to the complaint, prices for ordinary DRAM have climbed roughly 700 percent over four years. When many firms make a product, holding back supply invites a rival to take the business. In DRAM that check is missing, because no other maker has the scale to undercut the three. If all three hold back at once, no competitor is left to break ranks and sell more. None of them did.
Samsung and SK hynix have faced this accusation before. In the mid-2000s, the U.S. Justice Department prosecuted a DRAM price-fixing conspiracy that ran from 1999 to 2002. Samsung pleaded guilty and paid a $300 million criminal fine, SK hynix, then known as Hynix, paid $185 million, and several industry executives went to prison. Micron cooperated with that investigation and avoided a fine, yet it is a defendant in the new case.
This time, Samsung has called the lawsuit unfounded, and SK hynix says it is reviewing the complaint. Both point to their own spending as evidence of real demand. They have committed roughly $1.3 trillion over ten years to new factories, chip clusters and AI data centers. That scale of spending only makes sense if memory demand keeps rising.
That spending doubles as their best defense. To win, the plaintiffs need evidence, such as internal records or messages, that the companies moved together on purpose. Three firms each reaching the same conclusion about where the money is would not count. From the outside, chasing AI money looks the same as a planned shortage.
The new allegations are unproven. The question for a federal court is whether the shortage is history repeating itself or the AI boom working as advertised.
