Nvidia Raises AI Server Prices Amid Memory Chip Shortages

Elon Musk and Space Exploration Technologies Corp. plan to deploy 15 to 20 gigawatts of compute capacity for AI data centers by the end of 2027. During a SpaceX conference call, Musk identified memory computer chips as the primary bottleneck for this growth, stating that customer demand is growing at 200% while unit volume for memory chips is only increasing by 20% year-over-year.

In response to these supply constraints and soaring costs for dynamic random access memory (DRAM), Nvidia Corp. is raising prices for servers containing its artificial intelligence chips by more than 15% in many cases. These hikes will affect systems featuring Grace Blackwell and Vera Rubin chips and will take effect on shipments starting early next year.

Contract server builders have already notified major data center operators, including Microsoft Corp., Alphabet Inc., and Oracle Corp., of the price changes. While companies like Amazon and Meta are developing in-house chip programs, they remain dependent on Nvidia for their data center build-outs. The pricing shift provides significant leverage to memory chip manufacturers Samsung Electronics Co., SK Hynix Inc., and Micron Technology Inc.