Eight months into the memory crisis, the fear was that graphics card prices would spiral the way DRAM has. They haven't — at least not yet, and not evenly. Tracking the cheapest in-stock model of each GPU across ten countries paints a more complicated picture than "everything costs more." Some cards have drifted down since spring. Others have quietly become unbuyable. And the gap between the two is almost entirely about how much VRAM is soldered to the board.

How These Prices Were Measured

Every figure here represents the lowest price for a model that is actually in stock and readily available, pulled from regional price comparison sites covering a broad set of retailers in each market. That gives a realistic floor for what a buyer would pay, rather than a theoretical launch price nobody can find.

The second metric worth understanding is "true MSRP." Taking the heavily advertised launch price and converting it into local currency, then layering on regional taxes, produces a fair local benchmark. Comparing street prices against that number shows exactly how badly buyers in a given country are being squeezed relative to what the card was supposed to cost.

Three snapshots are used throughout: November 2025, when prices bottomed out; April 2026; and July 2026.

Nvidia's Low-VRAM Cards Have Barely Budged

The RTX 5050 has been remarkably static since April. Canada and the Philippines both saw increases north of 10%, but most regions are within a percentage point or two of where they were three months ago. Averaged across all ten countries, the card sits about 6% over MSRP. North American buyers get the worst of it — in the US and Canada the 5050 now occupies pricing territory that used to belong to the RTX 5060.

The RTX 5060 is one of the more surprising results. It has actually gotten cheaper in most regions, falling roughly 4% on average, with drops of 5–9% in Brazil, India, the Philippines, and Poland. Small increases hit the US and UK, but overall this 8GB card is available for something close to its MSRP worldwide. The exception is the US, where $340 works out to about 13% above where it should be. Cheap is relative here, but the memory crunch clearly hasn't reached this tier yet.

The RTX 5060 Ti 8GB Has Become Weirdly Compelling

The 8GB 5060 Ti dropped roughly 5% on average and now sells below MSRP in several markets, including the US. In Germany, the UK, and the Netherlands it costs about what it did back in November — a strong hint that demand is soft.

That creates an odd inversion. On paper the 5060 Ti 8GB should cost 27% more than the RTX 5060. In the US, Canada, and Australia that premium has collapsed to under 12%, and across all ten regions the gap averages 16% — for a card that delivers about 17% more performance at 1080p. It is still an overpriced 8GB product in absolute terms. It just happens to be the least bad rung on Nvidia's ladder right now, because the next step up is brutal.

The 16GB Tier Is Where the Damage Is

Every Nvidia card with 16GB or more has been hit, and the RTX 5060 Ti 16GB is the clearest casualty. In the US it now runs $570 — 33% over its $430 launch price and a staggering 58% more than the 8GB version of the same card. Elsewhere the damage was front-loaded: Poland and Brazil are flat versus April, and Germany and India actually ticked down slightly. Global average since April: up just 2%. But flat at a bad number is still a bad number, and buyers are paying roughly 22% over true MSRP.

The 5070 Ti and 5080 tell the same story. Neither moved much between April and July, and both remain stuck around 18–20% above true MSRP. That figure looks like the going rate for a 16GB GeForce card in mid-2026.

The RTX 5070 is the outlier that saves Nvidia's lineup. It's down slightly on average, sits near true MSRP in Australia, Germany, and Poland, and carries only about 6% inflation overall — a direct consequence of having less VRAM than its Ti sibling. With adequate memory and solid performance, it's the most defensible GeForce purchase available.

 

Nvidia GPU

 

 

Avg. inflation vs. true MSRP

 

 

RTX 5050

 

 

6%

 

 

RTX 5060

 

 

~0%

 

 

RTX 5060 Ti 8GB

 

 

Below MSRP in several regions

 

 

RTX 5060 Ti 16GB

 

 

22%

 

 

RTX 5070

 

 

6%

 

 

RTX 5070 Ti

 

 

~20%

 

 

RTX 5080

 

 

18%

 

 

RTX 5090

 

 

94%

 

The RTX 5090 Has Effectively Left the Gaming Market

No card rose faster between April and July. The 5090 climbed 12% on average and as much as 20% in some regions, landing at 94% above true MSRP — and it was never widely available at MSRP to begin with. At roughly $4,000 it's difficult to categorize as a gaming product at all. It's a workstation card for AI workloads carrying RTX Pro-tier pricing. Notably, it isn't even the cheapest route to 32GB of Blackwell VRAM: the RTX Pro 4500 typically goes for around $3,100, though its GPU core is closer to a 5080.

AMD's Mainstream Cards Are Sliding the Wrong Way

Both RX 9060 XT variants rose about 5% since April. The 8GB model took heavy hits in the UK and Canada — 16% in each — and now sits 8% over true MSRP. That increase, combined with the falling 5060 Ti 8GB, has gutted AMD's budget value argument. In the US, UK, and Canada the two cards now cost the same, despite the Radeon being positioned 21% cheaper at MSRP. Performance between them is close enough that the GeForce feature set becomes the tiebreaker.

Asus Tuf Radeon Graphics Card

The 16GB version fares better. At 18% over true MSRP it matches Nvidia's inflation rate but starts from a lower base, making it the cheapest 16GB GPU on the market — about 22% less than the RTX 5060 Ti 16GB. AMD has held the proportional gap between its 8GB and 16GB models steady since November, though the absolute cash cost of stepping up has grown because both ends moved.

The 9070 Series Is the Bright Spot

Radeon's higher-end cards held remarkably firm. The RX 9070 is only 4% above true MSRP after corrections in India and Brazil; the 9070 XT sits at 8%. In Australia, India, and Brazil the XT is as good a deal as it was in November, or better. Across all ten regions the RTX 5070 Ti costs 42% more than the 9070 XT — an enormous spread for competing products. The RTX 5070 and RX 9070, meanwhile, cost roughly the same, which is exactly right given identical MSRPs.

The RX 9070 GRE is the exception to skip. It launched at a puzzling $550 — identical to the RX 9070 — and averages only 6% cheaper while being 17% slower. Only the Philippines offered a discount large enough to justify it.

Intel Arc Is Vanishing From Shelves

Battlemage availability has thinned noticeably since April, with fewer models listed across most regions and no in-stock data at all for India or the Philippines in July. Both the B570 and B580 rose roughly 5% on average. The B580 is now 17% above true MSRP and costs about the same as an RTX 5060 in many markets — despite being priced 17% below it at launch and positioned to compete with the RTX 5050.

Which Countries Are Paying the Most

By price inflation index, Canada is currently worst off, with the US and UK close behind. (The 5090 is excluded from this comparison to keep the focus on cards actually bought for gaming.) Continental Europe fares better — Germany, the Netherlands, and Poland all show milder inflation than North America. Australia is the genuine anomaly: average prices sit slightly below true MSRP there. The RX 9070 XT, for instance, carries a true MSRP of AU$944 but was found for as little as AU$900.

 

Card (US, July 2026)

 

 

Price

 

 

RTX 5050

 

 

$290

 

 

RTX 5060

 

 

$340

 

 

RTX 5060 Ti 8GB

 

 

$360

 

 

RTX 5060 Ti 16GB

 

 

$570

 

 

RTX 5070

 

 

$600

 

 

RTX 5070 Ti

 

 

$960

 

 

RTX 5080

 

 

$1,250

 

 

RX 9060 XT 8GB

 

 

$360

 

 

RX 9060 XT 16GB

 

 

$450

 

 

RX 9070

 

 

$620

 

 

RX 9070 XT

 

 

$690

 

 

Arc B570

 

 

$230

 

 

Arc B580

 

 

$310

 

Another Round of Increases Is Already in Motion

Perspective matters: DDR5 has more than quadrupled in eight months, while GPUs sit just 16% above true MSRP on average — and only 10% once the 5090 is removed. Between April and July, prices rose 2% on average, or essentially nothing excluding the 5090. Given the scale of the memory crunch, that's far better than expected.

It probably won't last. Nvidia has reportedly raised the cost of the GPU-and-memory bundles it sells to AIB partners, driven by climbing GDDR6 and GDDR7 module costs. Retailers report that acquiring a 16GB GeForce GPU costs roughly 10–15% more than it did a few weeks ago, with smaller bumps on lower-VRAM parts. Some AIBs are warning of increases up to 40% further out. Radeon's trajectory is less clear, but since both vendors face the same GDDR costs, flat pricing is unlikely.

None of that is visible at retail yet. Current shelf prices reflect inventory retailers bought months ago, and slowing sales have left cards sitting longer, which has helped hold prices steady. Once that stock clears, incoming inventory arrives at higher cost.

Should You Buy Now?

If you genuinely need a card — as in, your current one can't do the job — buying sooner beats waiting, because the next wave of pricing looks worse. For everyone else, the answer is no. Keep what you have until a new generation arrives. Even if memory stays expensive, you'd rather pay these prices for something faster than pay them for today's hardware.