AMD graphics cards are going to become more expensive in 2026, according to information published and shared by the supply chain a few hours ago, as translated from Chinese. Memory costs, not memory itself, are the linchpin driving warnings; they filter directly into board prices demanded by add-in-board (AIB) partners as well as to us, the wallet-carriers.
Why AMD GPU prices are expected to increase in 2026
And now, it seems that PC enthusiasts and gamers are getting more for less—well, at least when you consider the cost benefits resulting from dropping memory prices.
Taiwanese publication UDN (via channel chatter picked up by VideoCardz) says AMD informed partners that a second round of price adjustments was on its way across its graphics card lineup on account of memory. The problem isn’t theoretical: DRAM pricing and graphics memory have both increased significantly in recent months as chipmakers shifted capacity to high-bandwidth memory (HBM) and low-power DDR for AI servers and mobile, squeezing supply for mainstream DDR5 and GDDR6 used in gaming GPUs.
Industry researchers such as TrendForce are still pointing to ongoing tightness in DRAM supply, and a shift of capacity towards HBM—where demand remains white hot for AI accelerators. Less supply of both wafers and back-end slots continues to support the reallocation toward higher-margin HBM, leaving fewer fabs or packaging lines available for GDDR—sending contract and spot prices up. That dynamic is already evident in retail, where a number of DDR5 kits and GDDR-based parts made some hard summer-to-fall jumps—doubling or worse—before scaling back somewhat to arrive at a new plateau.
In the bill of materials, memory is one of the highest expenses on a graphics card after the GPU and power delivery. Some models list 16GB, and even a small bump in GDDR6 price can be magnified into the sticker price because of it. Freight, currencies, and component rebates are all more challenging on the downside than earlier in the cycle, leaving AIBs with much less room for error.
What a 10% price rise at retail could look like for GPUs
Translate the headline number into actual cards and the math could hardly be simpler. A midrange model that’s now for sale at about $329 could climb toward approximately $360. For example, a performance-tier board that you’d been eyeing at around $599 would land in the vicinity of $659. Those are back-of-the-envelope numbers: the final totals will depend on each partner’s cooler design, VRM spec, and memory configuration, but the trend is obvious.
Partners can pull a second lever: promotions. Even if MSRPs nominally remain the same, anticipate fewer aggressive rebates, briefer discount windows, and more anemic game bundles as AIBs and retailers defend their margins. This “silent” inflation is frequently how the channel deals with increases in costs without rewriting product pages overnight.
It’s particularly worrying for value-focused consumers, as the past few AMD generations have given strong competition in terms of price-performance. If base prices rise and promos dry up, you could see models that simply landed near their target MSRPs—e.g., in-demand XTs at the upper median—lose part of their differential value.
Competition and supply chain wild cards to watch
Would Nvidia hold or cut pricing to take advantage? It’s possible but not guaranteed. Nvidia also depends on the same DRAM ecosystem and is working through tight availability of advanced packaging for its data center line. In the past, neither company gets into a groove of extended undercutting amid inflated component bills; instead, they adjust contract line items and steer demand toward SKUs with better yields or memory footprints.
Changes on the manufacturing side, like wafer pricing and substrate availability, and back-end capacity additions or curbs, can play favorably (or not) with memory-led pressure. DRAM suppliers (Samsung, SK hynix, and Micron) have communicated disciplined production and healthy HBM backlogs that bode well for a higher-for-longer memory pricing environment. If there is to be a GDDR respite, it may not spread very far if AI server buildouts continue at their current rate.
Currency swings also matter. A softer yen or euro relative to the dollar can create friction for international buyers even if U.S. pricing remains flat, while freight costs and regulatory fees can nudge regional prices in opposite directions.
Should you upgrade now, or should you wait?
If you’re already buying in the midrange, there’s a strong case to be made for upgrading sooner rather than later. Look for end-of-year and early-quarter deals, which may be the last broad discounts before channel repricing kicks in. Just make sure you’re comfortable with clock and memory configurations—8GB cards are more at risk to longevity concerns—so any extra you spend buys some actual headroom.
Those gunning for higher tiers, though, would need to compare total platform costs. When GPU prices do go up, there are frequently CPU, SSD, and monitor deals that offset the sticker shock to your build budget. The used market is another pressure release valve: As new lines cycle in, last-gen boards tend to see rapid price erosion—just consider warranty coverage and history of previous mining or heavy use.
If you can afford to be patient, keep tabs on signals like TrendForce’s DRAM contract pricing, guidance from memory makers, and partner notes from board vendors like PowerColor and Sapphire. A sustained pullback in DRAM or a big increase in supply would be your sign that the heat is off.
Outlook for AMD GPU pricing as memory costs rise
Across the channel, the consensus is that memory continues to be the pivot for GPU pricing as we move into 2026. With HBM soaking up production capacity and GDDR remaining taut, a potential 10 percent hike across the board for many of AMD’s graphics cards sounds feasible. So, budget accordingly and shop smart around promotional windows, but keep an eye on memory markets for the next round of GPUs; it may be that the price of bits dictates the price of frames.
