Gamers pricing out a new graphics card in the second half of 2026 got hit with another jolt. Taiwan-based outlet BenchLife first reported Nvidia’s move on GDDR6 and GDDR7 kit pricing, a story picked up fast by Wccftech; TechTimes followed on July 25, 2026 with confirmation that Nvidia had notified add-in-board (AIB) partners of increases spanning both the GDDR7-based RTX 50-series and older GDDR6 GeForce products. AMD had already flagged the same direction: TrendForce reported on July 3, 2026 that AMD raised its own GPU-and-memory kit prices by roughly 10% for AIB partners. AMD then made it official. On July 31, 2026, the company sent partners a formal notice, not a rumor, according to TechTimes, confirming that Radeon GPU and GDDR6 memory kit prices would rise by at least 10% starting in August.
Neither company has published an official retail price sheet. Retailers didn’t wait around for one. In China, RTX 5070 Ti listings jumped 21.8% and RTX 5070 listings climbed 16.2% inside a single 24-hour window on July 25 and 26, according to market-tracking reports, while RTX 5060 listings reportedly rose $75 overnight. By early August, the RTX 5090D, Nvidia’s China-market variant of the RTX 5090, was reportedly up **22.5%**, and TrendForce flagged on August 4 that RTX 50-series prices in South Korea could climb by as much as **30%**. This is the story of how a memory shortage born in AI data centers keeps reshaping the price of a graphics card, why this round of hikes looks structurally different from anything PC gamers dealt with before, and where the numbers plausibly go from here.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Just Happened: Nvidia and AMD Both Raise GPU Kit Prices
The mechanism at the center of this GPU price increase is unusual and worth understanding before anything else. Nvidia and AMD don’t sell bare GPU chips to board partners like ASUS, MSI, Gigabyte, or Zotac. They sell “kits,” bundles that already include the graphics processor and its memory chips soldered onto a reference board. That bundling means a board partner can’t shop around for cheaper memory on its own. When Nvidia or AMD raises the kit price, the increase lands on every card built from that kit, no matter which brand puts its sticker on the box.
According to the TrendForce report, AMD notified partners of an increase of roughly 10% on Radeon GPU-and-memory kits on July 3, 2026. Tech outlet MyDrivers, cited in that same report, put it plainly: “The latest price hike is primarily driven by the ongoing global shortage of GDDR memory.” Nvidia followed with a broader move: TechTimes reported on July 25, 2026 that Nvidia had notified AIB partners of increases touching the entire GDDR7-based RTX 50 series alongside older GDDR6 GeForce cards, a scope broad enough to include a recently revived RTX 3060 12GB still built on older memory, and wider than AMD’s Radeon-only kits at that point.
AMD closed the gap six days later. On **July 31, 2026**, the company sent board partners a formal notice, not an industry report, confirming that Radeon GPU and GDDR6 memory kit prices would rise by **at least 10%** starting in **August**, according to TechTimes. That distinction matters: a formal partner notice is a firmer commitment than a third-party report, even one as well-sourced as TrendForce’s July 3 story, and it means AMD’s increase now has a confirmed floor and a confirmed start date where it previously had neither.
Nvidia still hasn’t confirmed an exact figure for its own increase publicly, and reporting on its exact percentage varies. AMD’s number is firmer: the July 31 formal notice put a specific floor under it, at least 10%, even though AMD hasn’t issued a public retail price sheet either. What multiple outlets agree on is the direction: both companies that control the discrete GPU market raised partner costs at least once apiece inside a single month, and Nvidia’s per-kit pricing reportedly remained below AMD’s even after AMD’s August round, according to one industry account. That detail matters for anyone comparing Radeon and GeForce prices later in this piece.
The July–August 2026 Timeline: From Rumor to Formal Notice
Lay the last five weeks end to end and the pattern is hard to miss: each report firmed up the one before it. TrendForce’s July 3 story was framed as a report, sourced to tech outlet MyDrivers, that AMD had raised Radeon kit prices roughly 10%. Three weeks later, TechTimes reported that Nvidia had gone further in scope, if not necessarily in percentage, by notifying AIB partners of increases across both GDDR7-based RTX 50-series cards and older GDDR6 GeForce products still being sold. Six days after that, AMD closed the loop: a formal notice to partners on July 31, 2026, confirmed in TechTimes’ reporting, put a firm floor under the Radeon increase at **at least 10%** and pushed the effective date to **August**. By the following week, the story had moved past kit prices entirely, with TrendForce flagging potential RTX 50-series retail increases of up to **30%** in South Korea.
| Date | Company | Development | Source |
|---|---|---|---|
| July 3, 2026 | AMD | Reportedly raised Radeon GPU-and-memory kit prices roughly 10% for AIB partners | TrendForce (via MyDrivers) |
| July 25, 2026 | Nvidia | Notified AIB partners of increases spanning GDDR7-based RTX 50-series and older GDDR6 GeForce kits | TechTimes (also BenchLife, Wccftech, Videocardz) |
| July 31, 2026 | AMD | Sent a formal partner notice confirming Radeon GPU and GDDR6 memory kit prices would rise at least 10%, effective August | TechTimes |
| August 4, 2026 | TrendForce | Flagged potential RTX 50-series retail price increases of up to 30% in South Korea | TrendForce |
The distinction between “reported” and “formal notice” matters more than it might look. Reported figures, even well-sourced ones, can shift before they reach a partner’s invoice. A formal notice is closer to a committed number. That AMD moved from the former to the latter inside a single month, while Nvidia’s own reported scope broadened at the same time, is the clearest signal yet that board partners should plan around higher kit costs through at least Q3 2026 rather than treat the July reports as a one-time adjustment.
How Much More Gamers Are Paying Right Now
The gpu price increase reported at the wholesale level is only half the picture. Kit prices are a wholesale number, invisible to anyone shopping at retail. Street prices are what actually show up in a shopping cart, and they’ve moved faster than the reported wholesale increases would suggest on their own. A Tom’s Hardware analysis found that $1,000 bought an RTX 5080 in November 2025. By mid-2026, that same budget only reached an RTX 5070 Ti, one tier down. The piece pegged the average global price increase across Nvidia, AMD, and Intel GPUs at 15%, spanning the full desktop graphics card market rather than one brand or one card.
China’s Retail Market Moves First
China’s e-commerce platforms tend to reprice faster than US or European retail chains, which run on fixed contracts and slower restock cycles. That’s why the sharpest numbers so far come out of Chinese listings: RTX 5070 Ti up 21.8%, RTX 5070 up 16.2%, both inside a 24-hour window on July 25 and 26. Distributors there appear to be front-running expected cost increases rather than waiting for an official price sheet, a pattern that showed up during the 2021 mining boom too and tends to overshoot the eventual “official” increase before settling back down.
The trend didn’t stop there. Retail tracking published in early August 2026 put the RTX 5090D, Nvidia’s China-specific variant of the RTX 5090, up **22.5%**, an even sharper move than the RTX 5070 Ti’s 21.8% jump in late July. Two data points inside two weeks is a pattern, not a blip, and it suggests China’s distributors are still front-running expected cost increases rather than settling into a new baseline.
South Korea Joins the Price Surge
China’s marketplaces moved first, but they’re no longer the only retail channel showing it. TrendForce noted on **August 4, 2026** that RTX 50-series prices in South Korea could climb by as much as **30%**, a bigger potential swing than anything confirmed in China so far. Korea’s retail channel runs through a smaller number of major importers rather than a sprawling e-commerce marketplace, which tends to make price moves there slower to start but sharper once they land. The table below lines up every regional retail data point reported so far.
| GPU / Product | Region | Reported Change | Window |
|---|---|---|---|
| RTX 5070 Ti | China | +21.8% | Jul 25-26, 2026 (24-hour window) |
| RTX 5070 | China | +16.2% | Jul 25-26, 2026 (24-hour window) |
| RTX 5060 | China | +$75 (reported overnight) | Jul 25-26, 2026 |
| RTX 5090D | China | +22.5% | Early August 2026 tracking |
| RTX 50-series (general) | South Korea | Up to +30% (flagged, not yet confirmed) | TrendForce note, Aug 4, 2026 |
The Revived RTX 3060 Wildcard
One detail complicates the simple “everything costs more” narrative. Nvidia has reportedly brought the RTX 3060 12GB back into production, a card originally launched years ago on GDDR6 memory that’s now comparatively cheaper to source than GDDR7. A separate Tom’s Hardware report says the revival could take some pressure off entry-level GPU prices, even as it also notes a rumored RTX 5050 9GB variant was abruptly shelved around the same time, seemingly for the same underlying memory math. Reviving an old card to dodge a new memory shortage is an odd look for a company selling “next-generation” silicon, but it says a lot about how tight GDDR7 supply has gotten.
Why GDDR6 and GDDR7 Memory Got So Expensive
None of this is really a GPU story at its root. It’s a memory story that happens to land hardest on GPUs, because a modern graphics card carries more memory, by both volume and cost, than almost any other consumer device. An RTX 5090 ships with 32GB of GDDR7. A high-end gaming laptop might ship with 32GB of system RAM total. The card’s memory footprint alone rivals an entire PC’s.
Demand for that same memory technology, and for DRAM more broadly, is being pulled hard in a different direction: AI data centers. TrendForce noted in its July 3 report that AI server demand continues to support elevated memory prices through the third quarter of 2026. Standard consumer memory has been swept up in the same shortage. A 32GB DDR5 kit that cost roughly €80 in mid-2025 was fetching close to €167 by July 2026, according to market pricing data, more than double in about a year. Memory fabs allocate wafer capacity to whichever buyer pays the most, and AI infrastructure buyers have been paying far more than consumer GPU and PC makers ever have.
That’s the structural piece of this GPU price increase that separates it from an ordinary component shortage. It isn’t a factory fire or a shipping delay. It’s a bidding war for finite fab capacity between AI hyperscalers and everyone else, and everyone else is losing.
This Isn’t 2026’s First GPU Price Hike
The July kit increases are the latest entry in a list, not an isolated event. Board partners started signaling trouble early in the year, and AMD in particular has now raised Radeon pricing more than once. Tom’s Hardware reported that AMD had already added $10 to Radeon RX 9000 pricing for every 8GB of VRAM on a card earlier in the year, with further increases flagged as likely. A separate Tom’s Hardware report tied that broader 2026 pattern directly to the “ongoing AI-related DRAM supply crisis,” putting a floor under expected increases at roughly 10%.
Nvidia’s timeline runs in parallel. Board partners including ASUS confirmed price increases at the start of the year, and Nvidia reportedly pushed board-partner costs on RTX 5090 and 5090D boards up by roughly $300 in the spring, citing GDDR7 costs directly. The July and August kit increases from both companies are best read as the third or fourth round in a slow-motion repricing that started in January, not a sudden, isolated shock.
That pattern is exactly why one hardware outlet, FrameReady, framed its coverage of the latest move with a blunt question in the headline: “GPU prices are rising again: build now or wait?” It’s a fair question for anyone who has watched a build budget slip further out of reach every few months this year.
The Bill-of-Materials Problem, Explained
Every graphics card has a bill of materials, the running total of what it costs to physically build one unit before marketing, shipping, or margin get added. Memory chips are one of the largest line items on that list for any modern GPU, alongside the processor die itself, the cooler, the power delivery components, and the PCB. When memory costs rise, the finished product gets more expensive without a single design change. Nvidia and AMD don’t need a new chip architecture to justify higher prices. Rising input costs do that on their own.
That also explains why board partners have so little room to absorb the hit quietly. A company like MSI or Gigabyte buys a finished kit, chip and memory together, and stacks its own cooler, PCB design, and warranty support on top. If the kit itself costs 10% more, the finished card has to cost close to 10% more too, unless the partner is willing to eat the difference out of an already thin margin. Few are. One Tom’s Hardware report says MSI is planning to raise prices across its gaming product line by as much as 30%, a figure that spans more than graphics cards alone and hints at how far this cost pressure has spread through a single company’s catalog.
2021 Mining Boom vs. 2026 Memory Crisis
Anyone who shopped for a GPU in 2021 remembers what happened when demand outran supply. The RTX 3080 launched at a $699 MSRP. By mid-2021, cards were reportedly changing hands for as much as $2,400, and average eBay pricing that September still sat around $1,673, well above the sticker price. The RTX 3090 told the same story at a higher altitude: a $1,500 launch price against an average resale price near $2,863 that same September. Estimates from that period suggested roughly 700,000 GPUs, close to a quarter of total shipments in the first quarter of 2021 alone, were bought by cryptocurrency miners rather than gamers.
The 2026 situation shares a symptom, higher prices, but not a cause. The 2021 spike was demand-side and speculative. Miners were chasing crypto profits that could evaporate the moment coin prices dropped, which is exactly what happened when prices began normalizing in late 2021. The 2026 memory crunch is supply-side and structural. AI data center buildouts aren’t a speculative trade that unwinds in a bad quarter. They’re multi-year capital commitments from the largest technology companies in the world, and they compete for the same finite memory-fab capacity that gaming GPUs depend on. That’s a harder problem to wait out than a mining bubble, and it’s the reason few analysts are predicting a quick return to 2025-era pricing.
| Factor | 2021 Mining Boom | 2026 Memory Crisis |
|---|---|---|
| Root cause | Cryptocurrency mining demand | AI data center memory demand |
| Demand type | Speculative, tied to coin prices | Structural, multi-year capital spending |
| RTX 3080 pricing | $699 MSRP vs. up to $2,400 street price | Not applicable (discontinued card) |
| RTX 3090 pricing | $1,500 MSRP vs. ~$2,863 average resale (Sep 2021) | Not applicable (discontinued card) |
| Estimated share diverted from gamers | ~25% of Q1 2021 GPU shipments to miners | Undisclosed, but spans GPUs, DRAM, and consoles |
| How it ended / is ending | Prices normalized once coin prices dropped in late 2021 | No confirmed end date, TrendForce projects pressure through Q3 2026 |
Nvidia vs. AMD vs. Intel: Who’s Raising Prices Fastest
All three companies that sell discrete or integrated graphics silicon are exposed to the same memory market, but not evenly. Nvidia’s RTX 50 series leans on GDDR7 across most of the lineup, the newest and tightest-supplied memory standard on the market. AMD’s Radeon RX 9000 series splits between GDDR6 on some models and newer memory on others, giving it a bit more flexibility to source around the worst of the shortage. Intel’s Arc lineup is a smaller piece of the discrete market, but it isn’t insulated either. The 15% average global price increase that Tom’s Hardware reported spanned all three brands, not just the two market leaders.
RTX 50-Series vs. Radeon RX 9000 Pricing
The table below lines up official launch pricing against where each tier has effectively landed as 2026’s repeated hikes stack up. Treat the “2026 pricing trend” column as a directional signal built from the reporting cited throughout this piece rather than a fixed price list. Actual shelf prices vary by region, retailer, and week.
| GPU | Launch MSRP | Memory Type | 2026 Pricing Trend |
|---|---|---|---|
| RTX 5090 | $1,999 | 32GB GDDR7 | Repeated board-partner cost increases reported through the year |
| RTX 5080 | $999 | 16GB GDDR7 | Reportedly no longer reachable at original MSRP by mid-2026 |
| RTX 5070 Ti | $749 | 16GB GDDR7 | China retail listings up 21.8% in a single 24-hour window (Jul 25-26) |
| RTX 5070 | $549 | 12GB GDDR7 | China retail listings up 16.2% in the same window |
| RTX 5060 Ti | $379 (8GB) / $429 (16GB) | GDDR7 | Included in the broader GDDR7 kit price increase Nvidia confirmed on July 25, 2026 |
| Radeon RX 9070 XT | $599 | 16GB GDDR6 | Subject to AMD’s confirmed at-least-10% AIB kit price increase, effective August 2026 |
| Radeon RX 9070 | $549 | 16GB GDDR6 | Subject to the same AIB kit price increase |
Enterprise Isn’t Spared: RTX PRO 6000 Blackwell Hits $13,250
It would be easy to assume this is purely a consumer gaming problem. It isn’t. Nvidia’s RTX PRO 6000 Blackwell, a workstation and small-scale AI card aimed at studios and enterprise buyers rather than gamers, has climbed to roughly $13,250, according to Tom’s Hardware, a 55% increase over its original MSRP inside a single year.
That’s a steeper climb than anything reported on the consumer side, and it points to something worth sitting with. The same memory that’s short for gamers is short for professional buyers who need it for content creation, simulation, and small-scale AI inference. When the highest-margin, most memory-hungry professional cards see the sharpest increases, it signals the shortage is being felt hardest exactly where memory allocation is most valuable. The squeeze on consumer gaming cards may keep tightening as long as enterprise and AI buyers stay willing to absorb even steeper prices for the same underlying chips.
Board Partners and System Builders Feel the Squeeze
Board partners are stuck in the middle of this GPU price increase with the least ability to change it. They don’t set memory prices, they don’t set kit prices, and they compete against each other on the same underlying hardware. Their only real levers are cooler design, warranty terms, and how much of the cost increase they choose to pass straight through to buyers versus absorb.
System builders and prebuilt PC makers face a similar bind, one step further down the chain. A prebuilt gaming desktop bundles a GPU, CPU, RAM, storage, and a power supply, and at least three of those five components (GPU, RAM, and often storage) have been caught up in the same memory shortage simultaneously this year. That compounding effect is a big part of why prebuilt PC pricing has drifted upward across the board in 2026, not just on the graphics card line item.
What Industry Voices Are Saying
The clearest read on why this is happening now, rather than a year ago or a year from now, comes from the outlets tracking the supply chain directly. MyDrivers, reporting on AMD’s July kit price increase and cited in TrendForce’s coverage, said it plainly:
“The latest price hike is primarily driven by the ongoing global shortage of GDDR memory.”
MyDrivers, via TrendForce News, July 3, 2026
Hardware outlet FrameReady, summing up the mood among PC builders watching prices climb again mid-year, put the reader’s dilemma into a single headline:
“GPU prices are rising again: build now or wait?”
FrameReady
Both framings point at the same underlying reality. Nobody covering this closely is describing it as a blip. Tom’s Guide, in its own coverage, went as far as compiling a running list of deals for readers trying to buy ahead of the next round of increases, treating further hikes as close to inevitable rather than a possibility.
Market Impact: Consoles, Handhelds, and Prebuilt PCs
GPUs are the sharpest edge of this story, but they’re far from the only hardware category caught up in the same memory shortage. Console makers have already passed costs on to buyers this year. Sony raised the PS5 price a second time as DRAM costs climbed, and Microsoft pushed the Xbox Series X to $799.99. Handheld gaming PCs have moved too, with the MSI Claw and ROG Ally X20 both approaching or exceeding the $2,000 mark in some configurations, and Valve’s Steam Frame headset saw its own release timeline slip amid the same DRAM crunch.
None of this is a coincidence. Every one of those products depends on the same category of memory chip that’s now going into AI servers instead. A console maker, a handheld manufacturer, and a GPU board partner are all bidding against hyperscale data center buyers for the same wafer capacity, and none of them have the budget to win that auction outright. The result is a slow, steady pass-through of memory costs across nearly every category of gaming hardware sold in 2026, not an isolated GPU pricing story.
Laid out side by side, the pattern is consistent across every category:
| Product | 2026 Price Move |
|---|---|
| PS5 | Sony raised the price a second time this year as DRAM costs climbed |
| Xbox Series X | Microsoft’s price now sits at $799.99 |
| MSI Claw | Approaching or exceeding $2,000 in some configurations |
| ROG Ally X20 | Approaching or exceeding $2,000 in some configurations |
| Valve Steam Frame | Release timeline slipped amid the same DRAM crunch |
Is There Any Relief in Sight?
There are a few small signals worth watching, even if none of them amount to a clear turning point yet. The revived RTX 3060 12GB, built on older and comparatively cheaper GDDR6, suggests Nvidia is at least willing to lean on legacy memory supply chains to keep an entry-level option available. TrendForce’s own July 3 note only commits to memory prices holding up “through the third quarter” of 2026, which leaves open the possibility of a plateau, if not an outright decline, once that window closes.
The bigger question is whether AI data center demand itself levels off. Memory fabs take years to bring new capacity online, so a slowdown in AI infrastructure spending would likely show up as easing memory prices only after a significant lag, not immediately. Nothing in the current reporting points to hyperscalers pulling back on memory orders anytime soon, which is the main reason few outlets are predicting a fast fix. If anything, the trend has broadened rather than eased: TrendForce’s August 4 note on South Korea suggests the pressure is showing up in new markets, not fading from the ones it already hit.
5 Predictions for GPU Pricing Through Early 2027
- More kit price adjustments are already arriving. AMD confirmed the next round itself: a formal partner notice on July 31, 2026 put Radeon GPU and GDDR6 memory kit prices up by at least 10%, effective August. With TrendForce flagging elevated memory demand through Q3 2026, additional rounds from Nvidia or AMD before January remain more likely than not.
- Workstation and enterprise GPU pricing keeps outpacing consumer cards. If the RTX PRO 6000 Blackwell’s 55% year-over-year jump is any guide, professional-tier cards will likely keep climbing faster than gaming SKUs, since enterprise buyers can absorb more cost before demand actually drops.
- Legacy-memory cards become a bigger part of entry-level lineups. Expect more revivals like the RTX 3060 12GB, since GDDR6-based designs offer a cheaper path to a sub-$300 card than anything built on GDDR7.
- Console and handheld pricing stays under pressure into 2027. With Sony, Microsoft, and Valve all having already adjusted pricing or timelines once in 2026, a second round tied to continued DRAM costs is a real possibility rather than a stretch.
- Retail price volatility, especially in China and now South Korea, keeps outrunning official kit increases. Distributors have shown a pattern of pricing in expected hikes before they’re confirmed, which means street prices will likely keep overshooting official numbers before correcting.
What This Means for Buyers Right Now
For anyone actually shopping, the practical takeaway is less about picking a side in the Nvidia-versus-AMD pricing debate and more about timing. If a build is time-sensitive, buying before the next confirmed round of increases, rather than waiting for a dip that multiple analysts aren’t forecasting anytime soon, is the safer bet based on everything reported so far in 2026. If a build can wait, watching for legacy-memory options like the revived RTX 3060 12GB, or for a specific card’s price to stabilize after an initial post-hike spike, has been a reasonably reliable pattern this year based on how China’s retail market has behaved after previous rounds.
The honest answer is that nobody, including the companies setting these prices, has committed to a date when this GPU price increase cycle ends. Until AI memory demand meaningfully cools or new fab capacity comes online at scale, gaming hardware is going to keep competing for leftover supply, and losing more often than not.
Frequently Asked Questions
Why did GPU prices go up again in July 2026?
AMD reportedly raised GPU-and-memory kit prices for its board partners by roughly 10%, according to TrendForce’s July 3 report, and Nvidia followed with a broader increase covering both GDDR7-based RTX 50-series and older GDDR6 GeForce kits, reported by TechTimes on July 25, 2026 and picked up by BenchLife, Wccftech, and Videocardz. AMD then confirmed a further increase in a formal notice to partners on July 31, 2026, putting Radeon GPU and GDDR6 memory kit prices up by at least 10% starting in August. Both rounds trace back to the same shortage of GDDR and DRAM memory driven by AI data center demand.
Which GPUs are affected by the price increase?
Reporting indicates the scope covers the full GDDR7-based RTX 50 series (RTX 5090, 5090D, 5080, 5070 Ti, 5070, 5060 Ti) along with older GDDR6-based GeForce cards, including a revived RTX 3060 12GB. On the AMD side, the Radeon RX 9000 series, including the RX 9070 and RX 9070 XT, falls under the same kit price increase, confirmed **at least 10%** starting in August 2026, per AMD’s July 31 formal notice.
Is this the same as the 2021 GPU shortage?
No. The 2021 shortage was driven by cryptocurrency mining demand, a speculative trade that unwound once coin prices dropped in late 2021. The 2026 shortage is driven by structural AI data center memory demand, which isn’t tied to a speculative asset price and isn’t expected to unwind on a similar timeline.
Will GPU prices come down in 2026?
There’s no confirmed timeline for prices falling. TrendForce’s own analysis only projects elevated memory prices holding through the third quarter of 2026, without committing to a decline after that point. Most outlets covering the story are treating further increases as more likely than a near-term drop.
Is AMD or Nvidia raising prices more?
AMD moved first on paper, with TrendForce reporting a roughly 10% kit price increase on July 3, 2026. But Nvidia’s move was broader: TechTimes reported on July 25, 2026 that Nvidia had notified AIB partners of increases spanning both GDDR7-based RTX 50-series cards and older GDDR6 GeForce products, a wider scope than AMD’s Radeon-only kits at that point. AMD then followed with a formal notice of its own on July 31, 2026, confirming Radeon GPU and GDDR6 memory kit prices would rise by at least 10% starting in August, a firmer commitment than its earlier reported figure. Nvidia’s exact percentage still hasn’t been officially confirmed, but according to one industry report, Nvidia’s board-partner kit pricing remained lower than AMD’s even after AMD’s latest round.
Does this affect console and handheld gaming prices too?
Yes. The PS5, Xbox Series X, Steam Frame, and handhelds like the MSI Claw and ROG Ally X20 have all seen price increases or delays tied to the same DRAM and memory shortage in 2026, since they rely on overlapping memory supply chains with discrete GPUs.
Should I buy a GPU now or wait for prices to drop?
Based on the reporting so far in 2026, waiting for a price drop hasn’t been a reliable strategy this year, since each pause in price increases has generally been followed by another round rather than a decline. Buyers with time-sensitive builds have generally fared better buying ahead of confirmed hikes than waiting for relief that hasn’t materialized.
What is a GPU memory “kit” price?
It’s the wholesale bundle price Nvidia or AMD charges board partners for a GPU chip pre-paired with its memory chips on a reference design. Because partners buy the chip and memory together rather than sourcing memory separately, any kit price increase passes through to every card built on that reference design, regardless of brand.
Did AMD confirm another price increase for August 2026?
Yes. AMD sent partners a formal notice on July 31, 2026, confirming that Radeon GPU and GDDR6 memory kit prices would rise by at least 10% starting in August, according to TechTimes. Unlike the roughly 10% figure TrendForce reported on July 3, this one came through a formal partner communication rather than an industry report.
Are GPU prices rising outside China too?
Yes. Early August 2026 tracking put the RTX 5090D up 22.5% in China, and TrendForce reported on August 4, 2026 that RTX 50-series retail prices in South Korea could rise by as much as 30%. The increases that started in Chinese retail listings in July have since shown up in a second major market.
Related Coverage
- Gaming GPU Prices Surge: RTX 5090 Tops $5,000 [2026]
- RAM Shortage 2026: AI Chips Hit 5 Gaming Platforms’ Prices
- DDR5 RAM Prices Up 110% as AI Devours Memory [2026]
- RX 9070 vs RTX 5070: 8-15% Faster at the Same $549 [2026]
- RTX 5090 vs RTX 5070 Ti for AI: 2x VRAM, 2.7x Price [2026]
- SSD Prices Double as NAND Shortage Hits Gaming PCs [2026]
More hardware coverage is available in the Hardware section.
