Nvidia servers face more than 15% price hikes as memory costs surge

Rising DRAM prices are pushing server manufacturers to revise costs for systems using Nvidia’s latest AI platforms, adding pressure to data centre expansion plans and increasing dependence on a handful of memory suppliers.

Nvidia’s AI-powered server systems could become significantly more expensive from early next year as rising memory chip costs force manufacturers to raise prices for some of the company’s largest customers.

According to reports citing people familiar with the matter, server makers that supply major data centre operators have informed customers that prices for certain Nvidia-based systems are expected to increase by more than 15% in several cases. The revisions are expected to apply to systems shipped from the beginning of next year and will vary depending on the Nvidia chip generation and memory configuration.

The affected platforms include systems built around Nvidia’s next-generation Vera Rubin architecture as well as Grace Blackwell products, the people said. The information has not yet been made public. Nvidia did not respond to requests for comment.

Memory costs put pressure on AI infrastructure

The planned increases highlight the growing influence of memory manufacturers as the global technology industry races to expand AI computing capacity.

Samsung Electronics, SK Hynix and Micron Technology dominate global DRAM production, while demand for memory used alongside AI accelerators has continued to climb rapidly. Although the manufacturers are expanding production, supply has struggled to keep pace with requirements from data centres and other technology applications, pushing memory prices higher.

Nvidia’s AI accelerators form the computing foundation for systems used to train and operate increasingly sophisticated AI models. Their performance also relies heavily on high-capacity, high-speed memory, making memory availability an important factor in the overall cost of AI infrastructure.

The latest developments also illustrate the cost pressures emerging across the broader semiconductor industry. Technology companies including Apple and Qualcomm have indicated that supply constraints and higher component costs are affecting product pricing.

Nvidia itself remains among the semiconductor industry’s most profitable companies, supported by exceptionally strong demand for its AI processors. Its accelerators can command prices of tens of thousands of dollars, while the company has maintained a gross margin of around 75%. Limited alternatives to Nvidia’s products and constrained chip manufacturing capacity have helped sustain pricing power.

The company has also reportedly increased prices for some gaming-focused graphics cards in recent months.

Data centre expansion faces fresh cost challenge

For major cloud and technology companies, the higher server prices could add another layer of complexity to already expensive AI infrastructure programmes. Amazon, Microsoft, Google and Meta are developing their own AI processors to reduce dependence on Nvidia over time, but continue to rely heavily on Nvidia hardware for data centre deployments.

Their ability to diversify will partly depend on securing sufficient memory supplies from Samsung, SK Hynix and Micron. If memory remains constrained, switching to alternative processors may not immediately shield operators from rising infrastructure costs.

The pricing changes could also complicate large-scale data centre projects already facing delays caused by labour shortages, tighter financing conditions, construction challenges and opposition from local communities.

Nvidia’s upcoming fiscal second-quarter results are therefore expected to receive close attention from investors and the wider technology industry. As the company remains a central supplier to the AI infrastructure boom, its outlook could provide further indications of whether demand, supply constraints and rising component costs are beginning to alter the economics of AI expansion.