Some of Nvidia Corp.'s biggest customers have been told that prices for servers containing its artificial intelligence chips will rise by more than 15% in many cases, as memory chip costs soar.

The price hikes will go into effect on systems shipped early next year and will affect configurations including those built on Nvidia's flagship Vera Rubin and Grace Blackwell chips, according to people familiar with the process, who asked not to be identified commenting on communications that have not yet been made public. The size of the increases will depend on the generation of Nvidia chip and the memory configuration, these people said. Grace Blackwell pairs Nvidia's Grace central processor with its Blackwell AI chips, and Vera Rubin is the succeeding generation in the annual platform cadence Nvidia has laid out for its data center products.

Companies that build servers under contract for large data center operators — including Microsoft Corp., Alphabet Inc.'s Google and Oracle Corp. — have recently notified their customers of the forthcoming increases, the people said. Nvidia itself designs the chips but does not assemble the servers; that work is typically carried out by contract manufacturers such as Hon Hai Precision Industry, better known as Foxconn, and Taiwan's Quanta Computer, which build systems around Nvidia's silicon. Nvidia representatives did not respond to requests for comment.

The inability of the industry's most dominant company to hold the line on prices or absorb growing costs shows how much leverage memory chip makers Samsung Electronics Co., SK Hynix Inc. and Micron Technology Inc. now hold amid a surge in demand for AI infrastructure. Other major technology companies, including Apple Inc. and Qualcomm Inc., have recently said they have been forced to charge more for their products because of chip shortages.

Nvidia's accelerator processors are the heart of computers that create and run AI software, and their effectiveness depends on how much dynamic random access memory, or DRAM, they are paired with. The most advanced of those accelerators use high-bandwidth memory, or HBM — a premium, vertically stacked form of DRAM — whose production is likewise dominated by the same three companies. Samsung, SK Hynix and Micron account for most of the world's production of that type of chip. While the three manufacturers have been increasing output, they still have not caught up with surging demand — a gap that has driven the price of the commodity-like components up massively and given their makers unprecedented influence in technology.

Nvidia is one of the most profitable companies in semiconductors. It is able to charge tens of thousands of dollars per chip because supply from its contract manufacturer, Taiwan Semiconductor Manufacturing Co., still cannot meet runaway demand. The company's gross margin — the percentage of sales remaining after deducting the cost of production — is 75%. Originally derived from PC gaming chips that sold for hundreds of dollars, Nvidia's AI accelerators have seen prices driven up by incessant demand and, as yet, a dearth of viable alternatives to its offerings.

Nvidia has also raised prices for its gaming-oriented PC graphics cards, industry news site Tom's Hardware reported earlier this month, with U.S. prices on its GeForce RTX 50-series cards spiking as much as 39% at retail.

How Nvidia's customers react to this latest move — and whether it will create an opening for its competitors — will likely depend on whether they are able to secure enough memory themselves. Major customers such as Amazon, Microsoft, Google and Meta are all pursuing their own in-house chip programs — Google's tensor processing units and Amazon's Trainium chips among them — but remain dependent on purchases from Nvidia for their data center build-outs. Their ability to push toward greater independence will also depend on their access to supply from Samsung, SK Hynix and Micron.

The price increases are also likely to add complexity to the industry's massive AI data center build-out ambitions. Project delays, labor shortages, tightening capital markets and community resistance to developments have already complicated many plans.

Nvidia is scheduled to report fiscal second-quarter earnings next week. The updates from the world's most valuable publicly traded company have become a key event for the technology industry and for investors who have poured money into AI infrastructure on the promise that it will transform the economy.

This story was originally featured on Fortune.