Nvidia informs major clients of AI server price hikes exceeding 15%

Nvidia has alerted its biggest customers that the cost of AI-focused servers will climb by over 15% in many cases, with the most notable rises affecting systems that incorporate the Vera Rubin and Grace Blackwell processors. The adjustments, slated for early-next-year shipments, depend on the specific Nvidia chip generation and the amount of DRAM installed, as memory prices have surged because manufacturers such as Samsung Electronics, SK Hynix and Micron have struggled to keep pace with demand.

Companies that assemble servers for large cloud operators—including Microsoft, Alphabet’s Google and Oracle—have already passed the notice on to their clients. Nvidia did not respond to inquiries, and analysts say the price pressure underscores the limited ability of even the sector’s dominant player to absorb rising component costs. The situation also raises questions about whether rivals can gain traction as major tech firms like Amazon, Microsoft, Google and Meta continue to rely heavily on Nvidia’s accelerators while pursuing their own chip initiatives. The upcoming price changes come ahead of Nvidia’s fiscal second-quarter earnings report, a key indicator for investors in AI infrastructure.