Nvidia warned about AI server price increase
Servers with Nvidia AI accelerators could rise in price by more than 15% at the beginning of 2027. The largest customers have already been warned about the upcoming price increase, the main reason for which is the rapid rise in memory chip costs.
According to Bloomberg, the increase will affect, in particular, systems with flagship Vera Rubin and Grace Blackwell accelerators. The specific amount of the price increase will depend on the generation of Nvidia chips and memory configuration.
The new prices will apply to systems whose delivery is planned for the beginning of 2027. Server manufacturers working under contracts for large data center operators are already notifying customers about the upcoming cost increase.
Among the companies for which suppliers assemble such servers are Microsoft, Google, and Oracle. Large tech corporations remain dependent on Nvidia hardware, even despite developing their own AI chips.
The main factor behind the price increase is memory. AI accelerators require large volumes of high-speed DRAM and HBM, and demand for them has surged along with the massive construction of data centers.
Global production of such chips is largely controlled by Samsung Electronics, SK Hynix, and Micron. Despite expanding capacity, manufacturers still cannot keep up with demand, which allows them to significantly raise prices.
The increase in server prices puts additional pressure on large-scale AI infrastructure development programs. Large tech companies are already spending tens of billions of dollars on new data centers, and higher equipment costs could further increase the budgets of such projects.
Nvidia has not officially commented on the information about the upcoming price increase.
