A year ago, a mainstream PC memory kit cost about $75. Today, the same kit can sell for as much as $460. The easy explanation would be another chip shortage. But this time, the culprit isn’t a lack of factories or broken supply chains. It’s a business decision.
As artificial intelligence spending accelerates, Samsung Electronics Co., Ltd. (OTC:SSNLF), SK hynix Inc. (NASDAQ:SKHY) and Micron Technology, Inc. (NASDAQ:MU) are increasingly allocating manufacturing capacity to high-bandwidth memory (HBM)—the premium memory chips powering Nvidia Corp.‘s (NASDAQ:NVDA) AI accelerators and hyperscale data centers. Conventional DRAM for laptops, desktops, smartphones and other consumer devices is taking a back seat.
The result? AI customers get priority, while everyone else pays more.
AI Is Paying More—So It Gets The Wafers
Samsung, SK Hynix and Micron control the vast majority of the global DRAM market, giving the three companies enormous influence over where memory production goes.
Unlike conventional DRAM, HBM commands significantly higher prices while consuming much more manufacturing capacity. Every wafer redirected toward AI memory means less supply for PCs, smartphones and automotive chips.
As semiconductor commentator Shanaka Fernando recently argued in a post on X, no coordinated action is needed to create today’s tight memory market. The economics are doing the work. AI memory generates higher returns, and manufacturers are simply following the margins.
The numbers show just how dramatic that shift has become.
According to TrendForce data, conventional DRAM contract prices surged 93% to 98% in the first quarter before climbing another 58% to 63% in the second quarter. NAND flash prices also rose 70% to 75% as suppliers continued prioritizing AI-related products over mainstream memory.
Even the Biggest Customers Are Feeling the Pressure
The squeeze is now rippling across the technology industry.
Tesla Inc. (NASDAQ:TSLA) CEO Elon Musk warned earlier this year that the company was running into a “chip wall” as AI infrastructure strained semiconductor supply. Dell Technologies Inc. (NYSE:DELL) CEO Michael Dell has similarly pointed to an explosion in memory requirements driven by AI deployments.
Meanwhile, HBM capacity is effectively sold out through 2026, with much of 2027 production already committed. That has allowed memory makers to lock in premium pricing while demand continues to outstrip supply.
For Samsung, SK Hynix and Micron, the strategy has translated into expanding margins. By selling more high-value AI memory and less conventional DRAM, the industry’s biggest players are earning more from fewer consumer-focused chips.
Today’s Shortage Could Become Tomorrow’s Glut
The current pricing boom is unlikely to last forever.
Micron is building new fabs in Idaho and New York, while Samsung and SK Hynix continue expanding production capacity. Those investments are expected to come online over the next two years, increasing supply just as China’s CXMT rapidly expands its presence in the commodity DRAM market.
For now, however, AI remains first in line.
The bigger story isn’t simply that PC memory has become dramatically more expensive. It’s that AI has fundamentally changed how the world’s three largest memory makers allocate capital. As long as AI data centers continue delivering the highest returns, consumer electronics will keep competing for whatever capacity is left behind.
Photo: Pete Hansen / Shutterstock
Login to comment