Nvidia Tells Major Customers AI Chip Server Prices Are Rising Over 15%

Nvidia has told some of its biggest customers — the companies that build AI data centers for hyperscalers including Oracle and Microsoft — that server prices tied to its AI chips will rise more than 15%, Bloomberg reported. The increase follows GPU price hikes Nvidia already made earlier this year, and lands days before the company reports second-quarter earnings on Wednesday.

The hikes point to a widening gap between AI compute demand and available supply, with component costs across servers climbing as the AI data-center buildout accelerates. Nvidia's pricing power here extends well beyond its own margins: its GPUs remain the default substrate for training and inference across the industry, so a cost increase at this scale flows directly into the capital plans of every cloud provider and lab building on Nvidia hardware. Per the AI Market Watch index, our pipeline has logged 238 Nvidia-related items in the last 90 days versus 196 in the prior 90 — a coverage measure, not a census, but consistent with a company sitting at the center of nearly every infrastructure story right now.

For hyperscalers and neoclouds already committing hundreds of billions to data-center buildouts — including the financing backstops Nvidia itself has recently extended toward OpenAI's Ohio campus — a 15%-plus jump in server costs compresses the margin on those bets and raises the bar for the compute-financing arrangements now stacked on top of GPU purchases. Investors watching Wednesday's earnings call should listen for gross-margin guidance and whether Nvidia frames the increase as demand-driven or supply-constrained, since that distinction shapes how durable the pricing power looks going into next year.