Nvidia is telling customers AI server prices rise more than 15% on memory costs.

Bloomberg reported August 22 that Nvidia notified its largest customers that servers with its AI chips will cost more than 15% more in many cases, driven by memory prices, effective on systems shipped early next year and covering flagship Vera Rubin and Grace Blackwell configurations. Memory prices are up 500% in twelve months. This lands on the same buyers who just got a 20% inference price cut from OpenAI, which means the squeeze falls on anyone building capacity rather than renting it. If you were modeling a break-even on self-hosting, redo it.