NVIDIA has told some of its biggest customers that prices for servers containing its AI chips will rise by more than 15%, according to a Bloomberg News report published on Saturday. The price increases, driven by surging memory chip costs, will take effect on systems shipped from early next year.

The hikes will impact systems built around NVIDIA's flagship Vera Rubin and Grace Blackwell chips, which underpin much of the world's AI infrastructure. Server manufacturers building under contract for hyperscale data center operators such as Microsoft, Alphabet's Google, and Oracle have already begun notifying their customers of the upcoming increases.

The increases vary depending on NVIDIA's chip generation and memory configurations, according to people familiar with the process. Server DRAM contract pricing has roughly doubled across the first quarter of 2026, squeezing margins for server builders and pushing costs downstream.

NVIDIA is set to report second-quarter results on August 26, and the pricing pressure is expected to be a key topic on the earnings call. The company has become a proxy for the broader AI ecosystem, and the price hikes could have cascading effects on AI training and inference costs across the industry.

The news comes as hyperscale operators are investing hundreds of billions of dollars in AI infrastructure. Amazon recently lifted its 2026 AI capital expenditure forecast to $220 billion, and the broader AI-optimized IaaS market is projected to reach $42 billion this year. Higher server costs could slow the pace of expansion or force customers to absorb billions in additional expenses.

Sources