Nvidia Hikes Server Prices Amid AI Chip Supply Chain Strains
Nvidia has informed its top customers that servers containing its AI chips will see price increases of up to 15% starting early next year. The hikes apply to systems using Nvidia's flagship Vera Rubin and Grace Blackwell chips, with prices varying based on the chip generation and memory configuration.
The decision to pass costs on to customers rather than absorb them reflects the tight supply chain in the industry. This is evident from the fact that Samsung, SK Hynix, and Micron produce most of the world's dynamic random access memory (DRAM), a commodity-like component whose prices have surged due to high demand and limited output.
Nvidia's move comes as its CEO reports second-quarter results this week. The company operates with a 75% gross margin and charges tens of thousands of dollars per chip, making it sensitive to cost pressures in the supply chain. Other companies like Apple and Qualcomm have also raised product prices due to chip shortages.
The AI industry's reliance on memory has been highlighted by Micron CEO Sanjay Mehrotra, who described memory as the 'strategic infrastructure of the AI era.' He emphasized that AI systems require more memory with higher performance and lower power consumption.