Story
August 26, 2026

Nvidia’s AI Boom Is Forcing Its Biggest Buyers to Pay Even More

Nvidia server customers are reportedly facing price rises of more than 15% as soaring memory costs squeeze the AI infrastructure boom. The increases expose how even the market leader remains dependent on a constrained chip supply chain.

Nvidia’s AI gold rush is becoming more expensive for the companies building the data centers behind it. Even the industry’s dominant chip supplier is reportedly passing a fresh supply-chain shock on to its biggest customers.

Server makers working for Microsoft, Google and Oracle have told clients that systems carrying Nvidia AI chips will cost more than 15% in many cases, with the increases set to apply to shipments early next year. The affected machines include systems built around Nvidia’s flagship Vera Rubin and Grace Blackwell products.

The immediate pressure is memory. Nvidia’s accelerators depend on large amounts of DRAM, a market dominated by Samsung, SK Hynix and Micron. Those suppliers have expanded output but have not caught up with demand from the worldwide AI build-out, shifting unusual leverage toward memory makers and raising the cost of the servers that hyperscalers need.

That is a notable reversal of expectations around Nvidia’s pricing power. The company can still command tens of thousands of dollars for an AI chip and reported a 75% gross margin, yet it apparently cannot fully absorb the component squeeze. Its major customers are developing in-house chips, but their plans for greater independence will also hinge on access to the same constrained memory supply.

The warning signs had already spread beyond data centers. Nvidia had raised GPU prices this year, while a separate report said some of its biggest buyers had been notified of server increases above 15%; it added that the company could address pricing changes in its upcoming quarterly results. The added hardware bill lands on top of project delays, labor shortages, tighter financing and local resistance that have complicated AI data-center plans.

Amid that investment frenzy, Elon Musk amplified Epoch AI Research’s claim that official US GDP figures miss much of Nvidia’s contribution to the economy, saying growth had been understated by roughly 0.3 percentage points over the past year. The contrast is sharp: Nvidia’s boom may be helping power the economy, but its customers are being asked to fund an ever-costlier expansion.