Nvidia Corp. has notified its largest customers that the price of servers equipped with its AI accelerators will rise by more than 15% in many cases, with the changes taking effect on units shipped early next year. The hikes apply to systems featuring the flagship Vera Rubin and Grace Blackwell chips and will vary according to the chip generation and memory configuration.
Price increase details
Contract manufacturers that build servers for data-centre operators such as Microsoft Corp., Alphabet Inc.'s Google and Oracle Corp. have already passed the notice on to their own customers. Nvidia did not comment on the matter.
Memory market pressure
The surge reflects the growing influence of memory-chip makers, Samsung Electronics Co., SK Hynix Inc. and Micron Technology Inc., as demand for AI infrastructure outstrips supply. While the three firms have expanded output, they have not kept pace with the rapid rise in demand for dynamic random-access memory (DRAM), a key component that determines the performance of Nvidia's accelerators.
Implications for data-centre builders
Higher server costs add another layer of complexity to the massive AI data-centre build-out plans already challenged by project delays, labour shortages and tighter capital markets. Major cloud providers such as Amazon, Microsoft, Google and Meta are developing in-house chips, but they remain dependent on Nvidia for many AI workloads and on the same memory suppliers for DRAM.
Looking ahead
The price adjustments come as Nvidia prepares to release its fiscal second-quarter earnings next week, a report that investors watch closely given the company's 75% gross-margin and its role as the world's most valuable semiconductor firm. How customers absorb the higher costs could shape the competitive dynamics of the AI hardware market and may open opportunities for rivals if alternative memory supplies become available.
For further information, visit Nvidia's official site.