Nvidia’s AI Servers Are About to Get Pricier — What a 15%+ Hike Means for the Whole Stack

NVIDIA Corporation's (NASDAQ:NVDA) largest customers have been informed that prices of servers containing its AI chips will increase by over 15%, according to a Bloomberg News report on August 22. The hikes will take effect on systems shipped early next year, including those with NVIDIA's flagship Vera Rubin and Grace Blackwell chips.

While NVIDIA has not yet confirmed the news, the price increases are tied to soaring memory costs and will depend on the chip generation and memory configurations. Server manufacturers supplying Microsoft, Oracle, and Google have already begun notifying customers. The news coincides with the company's earnings release on August 26.

The price hikes are largely due to a broader industry-wide memory shortage, with memory manufacturers such as Samsung Electronics and SK Hynix gaining pricing power through the AI infrastructure boom. This pressure has now reached the top of the NVIDIA stack, with server prices rising by hundreds of thousands of dollars per rack.

NVIDIA's ability to pass on these costs to customers while accelerator supply remains tight may demonstrate pricing power, indicating a strong demand for AI infrastructure. However, as customers absorb most of the memory inflation, NVIDIA is likely to have pricing power in the stack. Companies such as Amazon, Microsoft, Meta, and Google are also developing in-house accelerators, but they remain dependent on NVIDIA for their data center build-out.