NVIDIA Corporation's (NVDA.US) AI servers "chip swapping" is igniting the memory market! Counterpoint warns: server memory prices may double by the end of 2026.

According to the report released on Wednesday by Counterpoint Research, a market research firm focused on the technology sector, NVIDIA Corporation (NVDA.US) adopting low-power memory chips commonly used in smartphones (such as LPDDR) in artificial intelligence servers may lead to a doubling of server memory prices by the end of 2026. In the past two months, as manufacturers have shifted their focus to high-end memory chips suitable for AI chips, the global electronics supply chain has faced a shortage of traditional memory chips. NVIDIA Corporation recently decided to reduce the power consumption cost of AI servers by replacing the memory chips from the DDR5 used in servers with the low-power LPDDR chips commonly found in smartphones and tablets. Counterpoint stated that as the number of memory chips required per AI server far exceeds that of a smartphone, this change is expected to trigger a sudden demand that the industry cannot cope with. Samsung Electronics, SK Hynix, Micron (MU.US), and other memory suppliers have faced a shortage of traditional DRAM products after adjusting their production capacity structure - reducing the production of traditional Dynamic Random Access Memory (DRAM) products and focusing on manufacturing high-bandwidth memory (HBM, a core component of advanced accelerators) that support the global AI boom. Counterpoint indicates that as chip manufacturers consider whether to shift more factory capacity to LPDDR to meet NVIDIA Corporation's demand, the tension in the low-end market may spread upwards. "The greater future risk lies in the advanced memory field, as NVIDIA Corporation's shift to LPDDR means it has become a customer equivalent in scale to large smartphone manufacturers - a huge shift in demand that the supply chain may find difficult to digest easily," Counterpoint said. The organization predicts that server memory chip prices will double by the end of 2026, and overall memory chip prices may rise by 50% from the current level in the second quarter of 2026. Higher server memory prices will increase the operating costs of cloud service providers and AI developers, potentially exacerbating budget pressures on data centers - which are already under strain due to record spending on graphics processor upgrades and power infrastructure.