Nvidia has told several of its largest customers that server systems built around its AI chips will cost more than 15% more in many cases, driven by rising memory chip prices, Bloomberg reported.
The price increases are expected to apply to systems shipped from early next year. They will affect some of Nvidia’s biggest AI platforms, including the Vera Rubin and Grace Blackwell families.
However, according to Bloomberg, the increase will not be the same for every customer. The final price will depend on which Nvidia chip is being used and how much memory is included in the system.
Nvidia’s higher costs are already moving through the supply chain
The expected price increase is not coming only from Nvidia. Companies that build AI servers for major data centre operators have also started informing customers about higher prices.
These server makers supply systems to companies such as Microsoft, Google and Oracle. This means higher costs are already moving through the supply chain even before Nvidia’s new prices officially take effect.
Nvidia has not publicly confirmed the reported price increases. For now, the information comes from people familiar with the matter rather than an official statement from Nvidia.
Memory chips are becoming a bigger problem for AI companies
For much of the AI boom, Nvidia’s powerful GPUs have been the main thing companies have struggled to get enough of. Now, memory chips are becoming another major pressure point.
Nvidia’s AI chips need large amounts of high-bandwidth memory, or HBM, to handle demanding AI training and computing tasks. Without enough memory, even powerful AI chips cannot work at their full potential.
Samsung Electronics, SK Hynix and Micron control most of the world’s DRAM production. Even though these companies have been increasing production, demand from AI data centres has grown so quickly that supply is still struggling to keep up.
And this is not something that can be fixed overnight.
Building new memory production capacity takes a lot of money and time. Even if chipmakers decide to increase production today, the extra supply could take several quarters to reach the market.
That means high memory costs could remain a problem for some time.
Nvidia has enjoyed strong profits because demand for its AI chips has far outpaced supply. But rising memory costs now raise the question of who will bear the extra expense. So far, reports suggest Nvidia is passing much of the increase on to customers rather than taking the full hit itself, though the exact split is still unclear.
Higher costs could reach customers at every level
The increase in Nvidia’s costs does not necessarily stop with Nvidia.
First, Nvidia raises prices for its chips and server systems. The companies that build and assemble those servers then have to decide whether to absorb the extra cost or pass it on.
Cloud giants such as Microsoft, Google and Oracle buy these systems for their data centres. They, in turn, provide computing power to businesses, AI companies and developers.
So, by the time the higher cost reaches the final customer, it could be larger than the original increase from Nvidia.
The actual impact will depend on how much each company in the chain chooses to absorb and how much it passes on to the next customer.
It could put pressure on the huge AI spending plans
The world’s biggest technology companies have committed hundreds of billions of dollars to building AI data centres and buying computing equipment. A big part of those spending plans is based on the idea that newer technology will provide more computing power for every dollar spent.
Higher memory costs could make AI infrastructure more expensive and force tech companies to rethink some of their massive spending plans. While this is unlikely to stop the AI boom, it could slow the expected decline in the cost of building and running AI systems.
What remains unconfirmed
There are still several unanswered questions around the reported increase. It is not yet clear how much of the higher cost will eventually reach businesses, AI developers and other end users.
It is also too early to say whether higher server prices will force companies such as Microsoft, Google and Oracle to change their massive AI infrastructure spending plans for 2027 and beyond.
Most importantly, Nvidia has not officially confirmed the reported price changes.
For now, the clearest takeaway is that the cost of building AI infrastructure is facing pressure from more than just expensive GPUs. Memory chips have become a major part of the equation, and if supply remains tight, those higher costs could continue to spread across the AI industry.
Disclaimer: This article provides factual analysis only and is not, and should not be construed as, an offer, solicitation, or recommendation to buy or sell securities. Investors must conduct their own independent due diligence and seek advice from a registered financial advisor in the respective jurisdiction.