GPU prices keep surging in 2026 — Lack of sale discounts and rising VRAM costs keep current-gen prices sky-high

We all know the current state of pricing for PC components is through the roof. Looking back, we have the hindsight to see what the prices were a year or two ago, and be thankful that we bought something when it was cheaper or on a really good sale. But for those who didn’t upgrade back then, what does the future look like if you want to upgrade your graphics card?

Currently, we have access to the 50-series range of graphics cards from Nvidia, and the 90-series from AMD, with a couple of entries from Intel. These cards were primarily released in January 2025, with different graphics card SKUs released throughout the year. We’re now more than halfway through 2026, with the rumored release of the next generation of graphics cards not arriving until late 2027, or even 2028, so we’re looking at at least a year before we see any hint of a new GPU release.

Why the delay or extended release cadence? Well, the simple answer is the costs. Just one aspect of the increased costs incurred with any new graphics card release will be the high-speed memory used, and, in the GPU’s case, the amount of VRAM. Prices have become exponentially more expensive due to the AI crunch, so adding more memory to a GPU is going to not only increase the production costs of any new graphics card direct from Nvidia or AMD, but also boost the expenses that AIB board partners have to incur to offer the partner GPUs. These manufacturing cost rises will, of course, be passed on to us poor consumers. There might even be reductions in the amount of onboard VRAM for different card SKUs; we will have to wait to find out.

There are no stock issues with picking up a new GPU; the main issue is just the cost. I keep track of all the current graphics card prices twice weekly in our GPU Price Index, and have seen the prices creep up steadily since release, especially with the higher-end and more popular GPUs, or those deemed the best performance per dollar.

Here are two tables that show the initial MSRP launch prices of Nvidia’s 50-series and AMD’s 90-series graphics cards, along with the current price and the lowest prices that we’ve seen for each card. This generation of cards was released with a higher MSRP, and many GPUs released were well above the announced MSRP prices set by Nvidia and AMD, garnering a lot of criticism from both the press and consumers. What’s also been highly noticeable this generation is the lack of notable below-MSRP discounts in major sales events such as Black Friday or Prime Day.

We can see that, in general, the lowest-ever prices seen are barely below the launch MSRP prices, if at all. With cards such as Nvidia’s flagship GeForce RTX 5090 never dropping below its initial release price. In the case of the RTX 5090, the price of the top-tier GPU that features 32GB of precious VRAM has only ever increased, rising from $1,999 to $4,288, a staggering $2,289 (114.5%) difference.

As I see the cheapest models sell out while I’m checking on the weekly prices, the cost of the GPU just keeps climbing a little higher every time. Although the RTX 5090 is the most extreme example of GPU inflation, it is unfortunately not alone.

AMD’s Radeon RX 9070 XT, which many feel to be the value champion for high-end GPUs, competes against Nvidia RTX 5070 Ti, but for several hundred dollars less. Up until very recently, the 9070 XT held a steady $699 retail price, but as of writing, the price has jumped to $719, a $120 hike from the $599 launch MSRP price, a 20% increase.

To help in your decision of what card you want to buy, consider what features the card offers and what you will make use of. Is ray-tracing important to you? Does frame gen technology interest you? Are the video encoders and codecs for streaming or video capture better for your associated software? Do your research; don’t just look at graphs with FPS numbers. Think about what is going to give you the most overall benefit.

Other things to also consider are the differences in the supporting software, such as FSR and DLSS, with Nvidia slightly ahead for the most part for the latest technology, but with the caveat that they are also more expensive. AMD cards this generation have offered more VRAM on lower-end GPU SKUs, which can be an important feature for gamers playing on higher-resolution monitors with game settings dialed up.

If you want a graphics card now, you don’t have much choice but to grab what you can at the lowest price that’s available to you at the moment. Holding out for the next generation of GPUs often isn’t going to save you any money; as with all technology, obsolescence is a fact of life.

We have to wait for the definitive specifications to be announced for the new GPU lineups, and until then, everything is educated guesswork and speculation. Whether reductions in onboard VRAM allocations take place, GPU die size shrinks, or core counts multiply, all wait to be seen. Whatever happens, though, one thing is guaranteed, and that is that the new lineup will be more expensive than this generation; we just don’t know yet by how much.

Finding a massive discount on a graphics card in the current economic climate is going to be limited to stores making pricing mistakes or some serious combo bundle deals from retailers. Seeing how even the Nvidia 40-series and AMD 70-series GPUs have held value and not greatly depreciated is an indication that the current generation of cards is not going to see massive discounts closer to the launch of the next generation. We hold out hope that we may see some discounts in the next Black Friday sales, but I wouldn’t count on seeing anything truly jaw-dropping, although I do hope to be proven wrong.