


Nvidia has reportedly shifted towards memory chips used in smartphones for artificial intelligence servers, which could lead to a doubling of server memory prices by the end of 2026. This report was published by Counterpoint Research.
In the past two months, electronic supply chains have experienced difficulties with older generation memory chips due to manufacturers focusing on high-quality memory chips suitable for AI applications. However, Counterpoint has indicated that a new problem is on the horizon. Nvidia has decided to change the type of memory chip from DDR5 to LPDDR, which are typically low-power memory chips found in phones and tablets, in order to reduce the costs of AI servers.
Nvidia is expected to release its results report later this evening. Due to the fact that each AI server requires more memory chips than a smartphone, this change is anticipated to create a sudden demand that the industry is not equipped to meet.
Memory suppliers such as Samsung Electronics, SK Hynix, and Micron are facing shortages in older generation dynamic random-access memory products after reducing production to focus on high-bandwidth memory production. Counterpoint warns that the low-level supply difficulties carry a risk of upward spillover. Chip manufacturers are evaluating whether to shift more factories to increase LPDDR production capacity to meet Nvidia's needs.
“The risk of being a comprehensive customer, as indicated by Nvidia's shift to LPDDR, has made it a customer similar in stature to a large smartphone manufacturer. The supply chain cannot easily absorb such a large demand,” states Counterpoint.
The firm forecasts that the prices of server memory chips could double by the end of 2026. High server memory prices could increase costs for cloud providers and AI developers, adding additional pressure to data center budgets that are already strained by record spending on graphics processing units and power upgrades.
```.png)
Sizlere kesintisiz haber ve analizi en hızlı şekilde ulaştırmak için. Yakında tüm platformlarda...