MUMBAI: AI may be getting smarter, but the machines powering it are getting pricier. Nvidia’s major customers could face more than 15 per cent increases in AI server costs as soaring memory chip prices push up the cost of systems built around the company’s processors, Bloomberg News reported.
The price increases are expected to affect systems shipped early next year, including servers based on Nvidia’s Vera Rubin and Grace Blackwell platforms. The exact increase will vary depending on the generation of Nvidia chips and the memory configuration used in each server.
According to people familiar with the matter cited by Bloomberg, server manufacturers supplying major data centre operators such as Microsoft, Google and Oracle have already warned customers about the expected increases. Nvidia representatives did not immediately respond to requests for comment.
At the heart of the increase is a familiar AI bottleneck: memory. Demand for memory used alongside Nvidia’s AI accelerators has continued to outstrip supply, with Samsung Electronics, SK Hynix and Micron accounting for most global DRAM production.
Nvidia’s AI accelerators depend on high-performance memory to train and run AI models. As memory becomes more expensive, the additional cost is filtering through to the servers that house Nvidia’s processors, potentially making large-scale AI infrastructure even more expensive.
The development is notable because Nvidia already commands significant pricing power in the AI chip market, with its accelerators selling for tens of thousands of dollars per chip. The latest pressure shows that even the AI hardware leader is not insulated from rising component costs.
The impact could extend beyond server manufacturers. Amazon, Microsoft, Google and Meta are developing their own AI chips, but all continue to rely on Nvidia for part of their data centre requirements. Their ability to reduce that dependence could also be influenced by the availability and cost of memory supplied by Samsung, SK Hynix and Micron.
The price pressure arrives as technology companies continue pouring money into AI data centres while grappling with project delays, labour shortages, tighter capital markets and community opposition to new developments.
Nvidia is due to report its fiscal second-quarter results on August 26, making the company’s outlook particularly significant for an AI infrastructure market where every extra chip, server and memory module now comes with a bigger price tag.
