Nvidia AI servers are set to cost more than 15 percent more in many cases, according to a report from Bloomberg, as an ongoing memory shortage drives up prices. The hikes will hit systems built around Vera Rubin and Grace Blackwell chips, and they apply to shipments early next year. Contract manufacturers building servers for Microsoft, Google, and Oracle have already informed those customers of the increases. Nvidia has not commented on the report.
The price jump stems from rising DRAM costs at Samsung, SK Hynix, and Micron, the three main memory makers. Their higher prices are flowing straight into the server market, where memory is a major cost component. The report, published by The Decoder on Aug 23, 2026, and written by Matthias Bastian, cites Bloomberg as the original source.
Who Pays the Bill
The added cost lands squarely on the biggest names in tech. Cloud giants Amazon, Microsoft, Google, and Meta will face higher prices, as will AI labs OpenAI and Anthropic. These companies are pouring billions into AI infrastructure, and Nvidia's chips sit at the center of that spending. The price increase sharpens a core tension: the same companies bankrolling Nvidia's market power are also its largest customers.
All of these firms are developing their own chips, yet they still depend on Nvidia for their most demanding AI workloads. That dependence gives Nvidia pricing leverage, and the memory shortage only amplifies it. For customers, the timing is awkward. They are trying to reduce reliance on a single supplier, but the market is moving in the opposite direction.
A Growing Strategic Bind
The price increases sharpen a core tension between Nvidia and its biggest customers. Those customers are effectively bankrolling the market power of the supplier they're trying to break free from. Every dollar they spend on Nvidia servers strengthens the company's position, even as they invest in alternatives.
Stay ahead of the AI curve
The most important updates, news, and content — delivered weekly.
No spam. Unsubscribe anytime.
The dynamic is not lost on industry observers. The AI industry still needs very high revenue growth to justify these investments, especially given Nvidia's significant outstanding liabilities. If growth slows, the math becomes harder for everyone involved.
What Comes Next
For now, the memory shortage shows no sign of easing. DRAM prices have been climbing for months, and the major manufacturers show little appetite for expanding supply quickly. That means the price hikes on Nvidia servers are unlikely to be a one-time event.
Customers have few options in the short term. They can absorb the costs, push back on delivery timelines, or accelerate their own chip development. None of those paths changes the immediate reality: early next year, shipments will cost more.
The report does not specify how long the increases will last or whether they will deepen. Nvidia's silence on the matter leaves room for speculation, but the company has not disputed the numbers. For the cloud giants and AI labs waiting on new hardware, the message is clear. Their next round of servers will carry a heftier price tag.