Nvidia Chip Shift Sparks Looming Surge in Server Memory Prices

Nvidia’s latest change in its AI server design has set off fresh concerns across the global electronics supply chain. The chipmaker’s decision to adopt LPDDR — the low-power memory typically found in smartphones and tablets — is expected to push server-memory prices sharply higher by the end of 2026, according to new research released on Wednesday by Counterpoint.

Over the past two months, manufacturers worldwide have struggled with a growing shortage of legacy memory chips. Many chipmakers shifted their focus toward high-end memory suited for advanced AI semiconductors, leaving older product lines undersupplied. Now, Nvidia’s pivot to LPDDR is adding an unexpected new layer of strain.

The company aims to reduce power consumption in its AI servers by replacing DDR5 server memory with LPDDR. But each server requires far more chips than a handset, creating a sudden demand surge that current production levels are not prepared to meet. Counterpoint noted that the scale of Nvidia’s demand mirrors that of a major smartphone brand — a shift that could reshape how memory suppliers allocate their factory capacity.

Samsung Electronics, SK Hynix and Micron are already facing tight supply conditions after cutting output of older dynamic random-access memory to ramp up high-bandwidth memory production, which is crucial for AI accelerators. The added pull toward LPDDR threatens to push this tightness further up the market.

If chipmakers redirect capacity to meet Nvidia’s needs, shortages could widen and prices may rise sharply. Counterpoint forecasts that server-memory prices could potentially double by late 2026, a development that would add fresh pressure on cloud providers and AI developers already grappling with soaring GPU and power-infrastructure costs.

Nvidia is expected to release its earnings report later on Wednesday, with investors closely watching how this supply-chain shift may influence its future product strategy.