0
Server, GPU, and CPU racks are installed at the "Nebius AI UK Data Center," housing equipment from Nvidia and other tech companies, at the ARK Data Centre in Chertsey, UK, on November 6 of last year (local time). / Reuters-Yonhap

Nvidia, the world's largest AI chip developer, is reportedly planning to raise the prices of its AI servers due to rising memory chip costs.

Nvidia has recently notified some of its major customers that prices could increase by more than 15% from current levels, Bloomberg reported on the 22nd (local time), citing sources.

Major server original equipment manufacturers (OEMs) have recently informed customers such as Microsoft, Google parent Alphabet, and Oracle of the planned price hikes, though the change has not yet been officially announced.

The higher prices are expected to apply to servers shipped starting early next year, including Nvidia's next-generation flagship chips, Vera Rubin and Grace Blackwell. The size of the increase will vary depending on the chip generation and memory configuration.

The price hike appears to be a response to the ongoing global shortage of memory chips. The efficiency of Nvidia's AI accelerator processors depends heavily on how much DRAM is packed into them.

Bloomberg reported that because memory chipmakers Samsung Electronics, SK Hynix, and Micron account for the vast majority of global DRAM production, they are exerting unprecedented influence by raising chip prices amid surging AI infrastructure demand.

Apple and Qualcomm have both recently raised prices on their products due to chip shortages. Nvidia's major customers, including Amazon, Microsoft, Google, and Meta, are all pursuing their own in-house chip development programs but remain dependent on Nvidia's products to build out their data centers.

Bloomberg also reported, citing industry news site Tom's Hardware, that Nvidia recently raised prices on gaming PC graphics cards as well.

Industry observers expect the server price hikes to add further pressure to plans for building large-scale AI data centers. Companies are already grappling with challenges including project delays, labor shortages, tight capital markets, and local community pushback against data center construction.

                                                                                                             Kim Hyun-min