Nvidia has told some large customers that prices for many AI servers due for shipment early next year will rise by more than 15% as memory costs climb.

Contract manufacturers have started informing buyers, including Microsoft, Google, and Oracle.

Nvidia has described Vera Rubin as a next-generation AI supercomputing platform, suggesting that some affected systems belong to a future server platform rather than a single chip.

The increases are expected to affect systems using Nvidia’s Vera Rubin and Grace Blackwell products, though pricing will vary by chip generation and memory configuration.

The move also reflects a broader memory squeeze.

S&P Global Market Intelligence said an AI memory boom is squeezing dynamic random access memory supply and pushing prices higher.

🔗 Source: Bloomberg

Recent Nvidia developments