GPU Pricing in 2026: MSRP and Street Price Statistics (2026)

Graphics-card pricing in 2026 has decoupled almost entirely from manufacturer MSRPs, as an AI-driven memory shortage, repeated wholesale price hikes and constrained supply pushed street prices to record highs. This reference page compiles the latest verified 2026 figures on GPU MSRPs, street prices, market share and the memory costs driving the surge, for journalists and researchers covering the story. Free to use with a link back to ShaneTheGamer as the source.

Key Takeaways for Journalists

• RTX 5090 street prices topped $4,600 against a $1,999 MSRP — more than double launch pricing, with some retailers and reviewers citing figures near or above $5,000 by September 2026.
• Nvidia ran three price hikes in 2026 — a 10–15% increase in January, a narrower May increase on the 5090, and a reported 20–30% wholesale bundle hike in mid-2026.
• Memory is the root cause, not the GPU die — DRAM/GDDR accounts for up to 80% of a card’s cost, and contract prices rose sharply through 2026.
• Nvidia held ~94% AIB market share in Q4 2025, per Jon Peddie Research, with AMD at 5% and Intel at 1%.
• RTX 5080 street prices ran roughly $1,300–$1,600 versus a $999 MSRP, with premium AIB models crossing $2,000.
• AMD raised Radeon RX 9000 MSRPs by 5–20% in Q3 2026, pushing some RX 9070 XT listings past $1,000 while typical street prices sat near $719–$749.
• The five-year-old RTX 3060 remains Steam’s most-used GPU at ~3.76% in August 2026 — a sign pricing is freezing gamers on old hardware.
• Discrete GPU shipments still rose 14.1% year-over-year in Q2 2026, per JPR, despite soaring prices — demand proved resilient.
• Analysts expect no broad price relief before 2027–2028, with the memory shortage forecast to persist for years.
• Entry-level cards face “even more severe shortages” in H2 2026, per PC Partner Group, likely pushing budget-tier prices higher.

The Headline Number: RTX 5090 More Than Doubled Its MSRP

The RTX 5090 is the clearest symbol of the 2026 GPU market. The RTX 5090 launched at a $1,999 starting MSRP in the United States, according to Nvidia’s January 2025 announcement, and Tom’s Hardware reported a tracked best price of $4,381 on August 7, 2026. That figure was a dated market observation rather than a universal price, but it captured a card selling for more than twice its launch cost.

Tracking sites confirm the scale. Nvidia set the RTX 5090’s MSRP at $1,999 in January 2025, and as of August 2026, the cheapest one available in stock at a major retailer was $4,599.99. Some outlets went further still: Tom’s Hardware’s September 2026 pricing archive reported that Nvidia’s top-end gaming GPU “now costs at least $5,000” and described Blackwell cards as continuing to endure drastic price hikes.

The reason the 5090 moved further than the rest of the stack is memory volume. The RTX 5090 has 32GB of GDDR7 memory on a 512-bit bus — a card carries roughly twice the memory of an RTX 5080, so every increase in GDDR7 pricing hits it about twice as hard. A cost-per-gigabyte lens makes this concrete: at its $1,999 launch MSRP, the RTX 5090’s 32GB of GDDR7 worked out to about $62.47 per gigabyte, almost identical to the RTX 5080’s launch-day rate of $62.44 per gigabyte, and the two cards started life priced in near-lockstep on a per-gigabyte basis.

They have not stayed aligned. At August 2026’s roughly $4,700 Newegg-based median, the RTX 5090’s 32GB cost about $146.87 per gigabyte, and a South Korean street price of about $5,112 pushed that rate to nearly $159.75 per gigabyte.

Three Price Hikes in One Year: The 2026 Timeline

What separates 2026 from prior GPU-crunch years is the cadence of wholesale increases. Nvidia implemented its third GPU price hike of 2026, raising GeForce RTX costs by 20% to 30% across models according to Taiwan’s Economic Daily News, following a 10–15% hike in January and a narrower May increase targeting only the RTX 5090. These are wholesale bundle increases, not just retail markups.

The mid-year round was the broadest. The latest increase applied to complete GPU bundles — the chip plus GDDR memory — rippling across virtually all consumer models, and unlike the narrower May increase that targeted only the 5090, this round spared almost no segment, with board partners paying nearly a third more for core components than months earlier. Retail listings moved in step: an Asus RTX 5080 jumped from $1,199 to $1,595 on Amazon.

The 5090 also received its own targeted wholesale bump. Nvidia notified board partners around May 13, 2026 of a further estimated $300 increase on the 5090 and 5090D V2 SKUs, citing rising GDDR7 memory costs. Even that trailed reality: as of March 2026, U.S. RTX 50-series prices averaged 19% above official MSRP, with premium RTX 5090 models routinely listed above $3,600.

Notably, the year did not start uniformly upward. GPU prices were actually trending slightly downward during the first half of 2026 for some models. The RTX 5060 Ti illustrates the whiplash: during the first half of 2026 the 8GB version fell in price and frequently sold below MSRP, but recent increases pushed it back toward or above normal pricing, leaving the 8GB card 24% more expensive than the prior month and 13% more expensive than in February.

Why Prices Are Rising: The AI Memory Shortage

The single biggest driver is memory. The sharp increase is due to DRAM memory (GDDR6 and GDDR7), which accounts for up to 80% of the cost of a video card. With AI data-center demand consuming fabrication capacity, gaming memory is being squeezed out. GDDR6 and GDDR7 memory for PC gaming graphics cards are only being allocated “limited capacity,” which means higher prices and scarcity.

TrendForce data shows how steep the increases were. The firm’s memory pricing survey projected conventional DRAM contract prices to rise 13% to 18% quarter-over-quarter in Q3 2026, with NAND Flash up 10% to 15% — substantial gains, but a marked slowdown from the roughly 60% jumps recorded in the second quarter. The magnitude at the consumer level has been dramatic: a 64GB DDR5-5600 kit that averaged $191 in August 2025 was going for $1,118 by August 17, 2026, according to Tom’s Hardware — a 485% increase in a single year.

TrendForce attributes the trend directly to AI. RTX 50 models equipped with 16GB or more of VRAM recorded price hikes of roughly 15% to 20%, while broader supply-chain increases across AMD and Nvidia ran 10% to 15% depending on memory configuration. The cooldown in the pace of increases is being driven by exhausted buyers rather than easing supply: the cooldown is driven by consumer electronics manufacturers’ unwillingness and inability to absorb higher memory costs after months of relentless increases, rather than by improving supply.

Relief is not imminent. Intel has pointed to 2028 before conditions normalize, and Silicon Motion has warned the shortage across DRAM, NAND and HBM will persist into 2028. For GPU buyers specifically, modest corrections of 3–7% are possible for RTX 5060/5070 in Q3 2026 if GDDR7 yields improve, but no broad deflation is projected before 2027.

AMD Radeon: Cheaper Rises, but Still Rising

AMD’s RDNA 4 cards weathered the storm slightly better, largely because of their memory choice. AMD GPUs haven’t been affected nearly as badly, although its 16GB Radeon models are still around $100 to $150 more expensive than MSRP — partly because AMD uses GDDR6 instead of GDDR7 and its margins differ from Nvidia’s. “Better” is relative, though: AMD officially raised the MSRP/SEP prices of its Radeon RX 9000 GPUs in Q3 2026, with the RX 9070 XT now costing more than $1,000 on average.

The formal MSRP increase covered the whole lineup. Per AMD’s notification, all Radeon RX 9000 cards based on RDNA 4, plus a few older popular models, saw price bumps ranging from 5% to 20%. Even so, real street prices for the flagship stayed well below the alarming four-figure headlines: the RX 9070 XT was $749 on Amazon currently, against a US launch MSRP of $599.

Down the stack, AMD arguably offers the most rational value in 2026. The Radeon RX 9070 is the best AMD card for most 1440p gamers with 16GB of GDDR6 and a tracked US price of approximately $620 in late July 2026, while the RX 9060 XT 16GB is the best value option with a tracked price around $440 to $459.99. The mid-range Radeon parts have held closer to launch pricing than any Nvidia tier.

Even AMD’s relative resilience is expected to erode. Radeon cards are still facing many of the same pricing pressures, and Radeon GPUs will likely become more expensive throughout the remainder of 2026. For deeper context on the wider PC and console market, see our gaming industry statistics page.

Market Share: Nvidia’s Near-Monopoly Meets Resilient Demand

Pricing power flows from concentration. Nvidia still controls roughly 90% of the discrete GPU market, according to Jon Peddie Research, so there is little competitive pressure stopping any of this. In the add-in-board channel the figure runs even higher: in Q4 2025, Nvidia’s graphics card market share surged to 94%, while AMD’s dropped to 5% and Intel remained at 1%.

Despite the prices, shipments have not collapsed. Sales of discrete graphics processors for consumer PCs were up both sequentially and year-over-year in Q2 2026 despite soaring prices, with standalone GPUs for consumer computers up 12.2% sequentially and 14.1% year-over-year — the best market dynamics in some time, per Jon Peddie Research. Total PC GPU volume also grew: the global PC-based GPU market in Q2 2026 grew to 75.5 million GPUs, an increase of 10.4% quarter-to-quarter and 1.1% year-over-year.

But JPR still expects a soft full year for cards. Due to constrained GPU supply, high GDDR prices and geopolitical uncertainty, the desktop graphics card market will decline about 10% year-over-year in 2026, with JPR noting customers who should be replacing PCs are holding off amid unstable conditions. AMD did claw back a fraction of share more recently: AMD’s overall market share increased by 0.6% from the prior quarter in Q2 2026, Intel’s decreased by 1.0%, and Nvidia’s rose 0.46%.

The Install Base: Pricing Is Freezing Gamers on Old Hardware

The clearest downstream effect of 2026 pricing is visible in what gamers actually own. The Nvidia GeForce RTX 3060, a desktop card that launched in early 2021, remained Steam’s most-used GPU with a 3.76% share in August 2026, up 0.05 points from the previous month. A five-year-old mid-range card topping the chart is a direct symptom of unaffordable upgrades.

Analysts have been blunt about the cause. PCGamesN’s framing puts it plainly: pricing, not preference, is keeping millions of gamers on more than five-year-old silicon, as new GPU generations carry higher launch prices and DDR5 memory prices climb sharply through 2026. The current-gen presence is growing but slowly: within Nvidia’s RTX 50 lineup, the mid-tier RTX 5060 is the most common choice, found in 2.98% of rigs recorded in the survey.

Even mobile silicon briefly challenged the desktop leaders. In July a laptop GPU topped the chart for the first time in survey history, with the RTX 4060 Laptop GPU at 3.81% edging the desktop RTX 3060 at 3.73%; that reversed in August, with the RTX 3060 back in first at 3.76% and the RTX 4060 Laptop GPU second at 3.63%. There is a plausible unlock scenario: the RTX 3060’s reign could finally end if pricing eases, since its long run at the top has been sustained largely by high replacement costs. For related audience data, see our Twitch statistics page.

Entry-Level and Regional Outlook for H2 2026

Budget buyers may see the worst of the next wave. PC Partner — the company behind ZOTAC, Inno3D and Manli — warned of a “substantial increase in VGA Card costs” in H2 2026, with graphics card shipments already down significantly and entry-level cards specifically facing “even more severe shortages” that will likely push their average prices higher. The volume damage is already measurable: PC Partner’s graphics card business suffered an 18.4% drop in unit sales during the first half of 2026 due to component shortages, while the average selling price of those cards increased 10.7%.

Regional variation is significant. India and Poland have been the most volatile, with the RTX 5090 jumping 50%+ since November 2025 due to parallel-import restrictions and distributor markups, and buyers outside the US should factor in 15–25% overhead beyond base price even before taxes. In Europe, the RTX 5080 averaged around €1,429 — about 11% above the US equivalent after conversion.

Tariffs add a further wildcard for US buyers, layering import duties on top of the memory-driven increases. The combined picture, according to multiple trackers, is a market where sale discounts have effectively disappeared and last-generation and used cards have become the pragmatic value plays. For anyone reporting on affordability, the takeaway is that both the MSRP and the street price have become moving targets in 2026.

Frequently Asked Questions

How much does an RTX 5090 actually cost in 2026?

Its MSRP is $1,999, but street prices reached about $4,600 or more by August 2026, with some reports citing figures at or above $5,000.

Why are GPU prices so high in 2026?

The main driver is an AI-fueled memory shortage; DRAM/GDDR accounts for up to 80% of a card’s cost, and DRAM contract prices rose by roughly 60% quarter-over-quarter at their Q2 2026 peak.

Did Nvidia raise GPU prices multiple times in 2026?

Yes. Nvidia ran a 10–15% hike in January, a narrower May increase on the RTX 5090, and a reported 20–30% wholesale bundle increase in mid-2026.

What is the most popular graphics card in 2026?

The five-year-old Nvidia RTX 3060 remained Steam’s most-used GPU at about 3.76% in August 2026, as high prices keep gamers from upgrading.

When will GPU prices come back down?

Broad relief is not expected before 2027, and Intel and Silicon Motion have warned the underlying memory shortage could persist into 2028.

Methodology & Sourcing

This page compiles the most recent verified 2026 figures from primary research firms (Jon Peddie Research, TrendForce), reputable trade press (Tom’s Hardware, TechSpot, TechPowerUp, Wccftech, PCGamesN), price-tracking services and Valve’s Steam Hardware Survey. Street prices are inherently volatile and represent dated snapshots that vary by retailer, region, board partner and stock at the time of observation; where a single tracked figure is cited (for example, a specific RTX 5090 price on a given date), it is labelled as such and should not be read as a universal or fixed price. MSRPs are manufacturer figures at launch. Projections — such as Q3 2026 memory increases or the timing of price relief — are labelled as estimates or forecasts and reflect the cited firms’ models, not certainties. Market-share and shipment data measure units leaving manufacturers into the channel, not point-of-sale purchases, and may reflect channel build-ahead. Figures were current as of the publication date and will be updated as newer data is released.

Sources

• Jon Peddie Research — Q4 2025 and Q2 2026 Market Watch / AIB reports (2025–2026)
• TrendForce — DRAM and NAND contract pricing surveys (2026)
• Tom’s Hardware — GPU and memory pricing coverage (2026)
• TechSpot — Q3 2026 GPU pricing analysis (2026)
• TechPowerUp — PC Partner Group / entry-level pricing report (2026)
• Wccftech — AMD Radeon RX 9000 price increase report (2026)
• PCGamesN — Steam Hardware Survey coverage (2026)
• Valve — Steam Hardware & Software Survey, August 2026
• Nvidia and AMD — official MSRP announcements (2025–2026)
• ThePCEnthusiast, iTechGuides, TrackaLacker, videocardprices.com, Tech Insider — price tracking (2026)

Spotted a newer figure or an error? Contact the ShaneTheGamer desk and we will update this page.