NVIDIA has reportedly informed several of its largest customers that server systems equipped with its artificial intelligence chips will see price increases of more than 15%, driven by a sharp rise in memory chip costs.

According to sources familiar with the matter, the price adjustment will apply to systems scheduled for shipment early next year, including those built around the flagship Vera Rubin and Grace Blackwell processors. The sources, who requested anonymity because the discussions have not been made public, indicated that the exact scale of the increase will vary depending on the generation of NVIDIA chips and the memory configuration involved.

Companies that manufacture servers on behalf of major data center operators such as Microsoft, Google's parent company Alphabet, and Oracle have recently alerted their clients to the impending price rise. NVIDIA representatives have not yet responded to requests for comment.

Even NVIDIA, the most dominant player in the industry, cannot maintain its original pricing or absorb the rising costs on its own. This situation underscores the significant bargaining power held by memory chip makers, including Samsung Electronics, SK Hynix, and Micron, especially against the backdrop of surging demand for AI infrastructure.

Other major technology companies, such as Apple and Qualcomm, have also recently acknowledged that chip shortages have forced them to raise the prices of their products.