Nvidia AI Server Prices Rise More Than 15% as Memory Costs Surge

Nvidia is reportedly raising prices on some AI server systems by more than 15% as soaring memory costs spread through the data-center supply chain.

The increases affect systems scheduled to ship early next year and vary depending on chip generation and memory configuration, Bloomberg reported. Hardware built around Nvidia’s Vera Rubin and Grace Blackwell platforms is among the products affected.

Server manufacturers supplying Microsoft, Google, Oracle and other large data-center customers have been informed of the changes, highlighting how memory shortages are raising the cost of the broader AI infrastructure boom.

Memory Becomes the AI Bottleneck

Advanced Nvidia accelerators require large amounts of DRAM and high-bandwidth memory, making server prices increasingly sensitive to memory costs.

Samsung, SK Hynix and Micron dominate global DRAM production, and AI data centers are consuming growing amounts of their output. Micron has said industry supply is likely to remain substantially below demand through and beyond 2026 as AI workloads require more memory per server.

The shortage has already become a major investment theme across AI chip stocks, particularly for Nvidia and memory supplier Micron.