Every week, I search through all of the major online retailers for the best graphics card deals, hunting down the lowest prices for the latest Arc, GeForce, and Radeon GPUs. Although a handful of models are wildly over their MSRP, many are still within an acceptable margin above it. However, according to a breakdown of the total cost of making a graphics card, certain GPUs are about to get a lot more expensive.
That’s according to the figures presented by Igor’s Lab, and while the report focuses on the current price and production cost of the Radeon RX 9070 XT for the European market, the same circumstances apply across all cards and sectors.
The primary reason why we’ve seen significant price hikes for certain graphics cards is, of course, the heavily constricted supply of DRAM modules, better known as the RAMpocalypse. AMD and Intel use GDDR6 across all their graphics cards, while Nvidia uses GDDR7 for almost all RTX 50-series models (the RTX 5050 being the exception).
These are manufactured by the same companies that produce High Bandwidth Memory (HBM) for AI data centers, and since the growth of those has exponentially increased in recent times, demand for memory chips, of any kind, is significantly outstripping the supply. However, that’s only one element of the total production cost for a graphics card.
You need to add in the cost of the GPU itself, and since all of them in the latest cards are made by TSMC—which has increased wafer prices—vendors are having to pay AMD, Intel, and Nvidia more for them. Then you’ve got all the other components and materials used in the entire graphics card, before you come to the cost of putting it all together and then packaging it.
And there’s more: Every graphics card is manufactured in Asia, predominantly China and Taiwan. Therefore, they need to be shipped and distributed for sale in other countries, so freight costs, import duties and other taxes and insurance have to be included in the final price tag.
Graphics card vendors also need to cover the cost of research & development, marketing and advertising, warranty claims, credit, currency conversion, and so on. Lastly, and most importantly to them, they need to make a sufficient profit from it all. Inflation naturally causes all these aspects to rise in price over time, but the volatile nature of geopolitics and conflicts throws in a variety of spanners into the works.
The type of graphics card most affected by all of this is the high-end model, due to its increased number of costly materials and components, especially VRAM. Hence, the likes of the RTX 5070 Ti, RTX 5080, and RTX 5090 are currently sporting minimum price tags that are 20%, 29%, and 90% over their respective launch MSRPs.
However, the RX 9070 and 9070 XT are currently only 9% and 17% overpriced (in the US, at least), and Igor believes that this won’t remain so, once inventory levels of old stock eventually get replaced.
One GPU plus eight VRAM chips equals a whole lot of money (Image credit: Future)
“The RX 9070 XT is currently less of a typical product in price comparisons and more of an indicator of a delayed price surge. This is precisely why the card still seems relatively attractive in May 2026, even though it’s objectively not cheap. The real news, therefore, isn’t that graphics cards have become more expensive. The real news is that part of the price increase may not have fully reached store shelves yet.”
The global supply of memory chips isn’t going to improve any time soon (if anything, it could actually get worse, as the push for new AI servers gets ever more frantic), and there is no sign of any positive changes in import policies or fuel prices on the horizon. Paying $700 for an RX 9070 XT could well become a thing of the past very soon.
