In short
Graphics card prices are far above launch levels. On 26 October 2026 Tom's Hardware's US tracker listed the RTX 5090 at $8,477 versus a $1,999 launch MSRP, and the Radeon RX 9070 XT at $869 versus $599. Reporting links the rise to AI-driven memory demand [1][2].
Key takeaways
- On 26 October 2026 Tom's Hardware's US tracker listed the RTX 5090 at a best price of $8,477 against a $1,999 launch MSRP, and the Radeon RX 9070 XT at $869 against $599 [1].
- The RTX 5090's best US price was $4,288 on 3 August 2026, so it nearly doubled in under three months by Pouyam's calculation [2][1].
- NVIDIA's Data Center unit generated $89.0 billion of $96.2 billion in Q2 fiscal 2027 revenue, about 92.5% by Pouyam's calculation, which in Pouyam's reading points to where its supply priorities lie [3].
- Jon Peddie Research says discrete GPU shipments rose 12.2% quarter on quarter in Q2 2026 even as a memory crisis pushed component prices up [4].
- Tom's Hardware found no notable below-MSRP discounts at big sales events, and expects the next generation, rumoured for late 2027 or 2028, to cost more [2].
- How much of the price rise is memory cost and how much is extra margin is unproven in the sources reviewed, so every price should be quoted with its date and method.
On this page 12 sections
- How much do graphics cards cost in 2026?
- Why are GPUs so expensive right now?
- NVIDIA RTX 50-series vs AMD Radeon RX 9000: which is better value?
- Have AMD and NVIDIA raised GPU prices?
- Is it a good time to buy a graphics card?
- What is the best GPU to buy in 2026?
- How are graphics card sales holding up despite higher prices?
- Will GPU prices go down in 2027?
- Common mistakes
- Frequently asked questions
- Glossary
- Sources
How much do graphics cards cost in 2026?
Graphics card prices in late October 2026 are well above launch prices for almost every current NVIDIA GeForce RTX 50-series and AMD Radeon RX 9000-series model. Tom's Hardware's GPU price tracker, last updated on 26 October 2026, lists the best US price of the RTX 5090 at $8,477 against a launch MSRP of $1,999 [1]. The Radeon RX 9070 XT sits at $869 against $599 [1].
What do MSRP, street price and best listed price mean?
MSRP is the manufacturer's suggested retail price, the launch price a maker names for a card. Street price is what shops actually charge, which can sit above or below MSRP. Best listed price is the lowest current US listing a tracker finds, so it is a floor, not an average of what buyers pay. Every figure in this article is a best US price from Tom's Hardware, shown with the date of the snapshot.
How have prices moved since August 2026?
Tom's Hardware's 3 August 2026 article gave an RTX 5090 best US price of $4,288, up $2,289 or 114.5% on the $1,999 launch MSRP, and said the card's price has only ever risen since launch [2]. By 26 October 2026 the tracker showed $8,477 [1]. For the RX 9070 XT the figure went from $719 on 3 August to $869 on 26 October [2][1].
The table below sets both snapshots beside launch MSRP. The premium column is Pouyam's calculation from the 26 October figures. Two points stand out: every listed card is above launch price, and the premium varies widely. The RX 9070 GRE carries the smallest premium ($599 against $549), while the RTX 5090 carries the largest [1].
Are prices the same in every country?
No. The figures here are US listings. TechSpot's Mid-2026 pricing update compares graphics card prices across 10 countries between April and July 2026. Its tables show that high-end card prices rose in several markets while mid-range prices were mostly flat or slightly lower [5]. Check local retailers before treating any US number as your price.
Why are GPUs so expensive right now?
The most-cited explanation is memory. Tom's Hardware's tracker says the memory industry is swamped by AI data-centre orders, and that NVIDIA is purportedly allocating memory based on the money it can make per GB of VRAM, which it says has made the situation dire for some models [1].
What is VRAM and why does it affect the price?
VRAM is the video memory soldered onto a graphics card. Consumer cards use GDDR memory such as GDDR6 and GDDR7, while AI accelerators use HBM, high-bandwidth memory stacked next to the chip. All of it competes for memory-making capacity with ordinary DRAM, the main memory in computers. Citing TrendForce, Astute Group reports memory prices rising by several hundred per cent and VRAM accounting for more than 80% of the bill of materials of some high-end GPUs. These are reported figures from industry sources, not company-confirmed data [6].
Where is NVIDIA's money coming from?
NVIDIA reported Q2 fiscal 2027 revenue of $96.2 billion on 26 August 2026, up 106% from a year earlier. Data Center revenue was $89.0 billion, up 117% [3]. By Pouyam's calculation that is about 92.5% of revenue ($89.0 billion divided by $96.2 billion). In Pouyam's reading, that points to where the company's supply priorities lie; it is an inference, not an NVIDIA statement. The release does not break out GeForce gaming revenue. Its separate Edge Computing segment ($7.2 billion, up 13% on the previous quarter and 27% on a year earlier) should not be read as a gaming figure [3].
Is it only memory, or are makers adding margin?
Sources emphasise memory. Jon Peddie Research (JPR) says a global memory crisis pushed component prices up [4]. A headline linked from Tom's Hardware's August 2026 article, however, reports an analyst suspecting that makers are raising prices beyond memory costs [2]. How much of the rise comes from memory and how much from margin is unproven, and no source reviewed here settles it.
Is this a shortage like the crypto-mining or pandemic years?
The evidence points to a cost squeeze more than a collapse in output. JPR counted 75.5 million GPUs shipped worldwide in Q2 2026 [4], and TechSpot reports JPR's figures putting discrete GPUs up 14.1% on a year earlier [7]. Stock is mixed. Tom's Hardware's August article said there are no stock issues and the problem is cost [2], while its October tracker says higher-tier cards such as the RTX 5070 Ti and RTX 5080 are in poor supply at reasonable prices [1].
NVIDIA RTX 50-series vs AMD Radeon RX 9000: which is better value?
Both lines are priced above launch, but by different margins. The table below pairs the closest-named cards by tier using the 26 October 2026 tracker. The pairings are Pouyam's editorial grouping, not benchmark equivalents; this article contains no performance tests, so it does not say which card is faster. The price gap column is Pouyam's calculation.
At every tier in that table the Radeon card was cheaper, by $70 at the 8GB entry level and by $360 between the RTX 5070 Ti and the RX 9070 XT [1].
What do NVIDIA and AMD each offer beyond raw speed?
Tom's Hardware says NVIDIA is slightly ahead on DLSS versus AMD's FSR, but that its cards cost more. It also says AMD's RX 9000 cards have offered more VRAM on lower-end models [2]. Three terms matter here:
- DLSS and FSR are upscaling technologies: the game renders at a lower resolution and software reconstructs a sharper image, raising frame rates.
- Ray tracing is a lighting method that simulates how light bounces in a scene. It looks more realistic but is demanding on the GPU.
- Frame generation creates extra frames between the ones the GPU renders, raising displayed frame rates at the cost of some added delay.
How big is the 8GB versus 16GB gap?
The same card name can hide a large price difference. On 26 October 2026 the RX 9060 XT 16GB was $599 against $459 for the 8GB version, a $140 gap. The RTX 5060 Ti 16GB was $765 against $529 for the 8GB version, a $236 gap (both gaps are Pouyam's calculation) [1]. Whether 8GB is enough depends on the games and resolution you play, so check reviews of those specific titles before paying for the larger version.
What stays uncertain?
Supply varies by tier. Tom's Hardware reports that higher-tier cards such as the RTX 5070 Ti and RTX 5080 are in poor supply at reasonable prices, so the card you want may not be available at the listed best price [1]. A cheaper card with a software disadvantage can still be the better value, and the reverse can also hold. Pick by your own criteria, not by brand.
Have AMD and NVIDIA raised GPU prices?
Yes, according to several reports, though they differ in how well confirmed they are. The retail rise is documented by Tom's Hardware's trackers. The partner-level increases come from industry sources rather than company announcements. The table below separates the two kinds of evidence.
What was reported for December 2025?
In December 2025 Tom's Hardware reported, citing an anonymous distribution source, that AMD had raised prices on Radeon RX 9000 cards and memory kits sent to board partners by $10 per 8GB of VRAM, with a further, undefined increase said to be scheduled for January 2026. This was not an AMD announcement. Tom's added that MSRPs would likely rise if distributors pass those costs on, and its implied new MSRPs were $619 for the RX 9070 XT (from $599), $569 for the RX 9070 (from $549), $369 for the RX 9060 XT 16GB (from $349) and $309 for the RX 9060 XT 8GB (from $299). Those figures are Tom's estimates, not announced prices [8]. TweakTown's article, headlined as a confirmation, repeats Tom's report and cites no independent source, so it is not a second confirmation [9].
What was reported for mid-2026?
TweakTown, citing Gazlog and Chinese board-channel sources, reported in June 2026 that AMD was planning a further 10-15% increase to Radeon RX 9000 cards in Q3 2026 because of rising GDDR6 costs. It said NVIDIA had not yet told partners of a new increase but had reportedly already raised RTX 5090 partner pricing in May. Gazlog flagged the report as not definitive [10]. AMD did not confirm the 10-15% figure in any source reviewed here.
How did retail prices compare with the forecasts?
Citing TrendForce, Astute Group reported that Nvidia and AMD were preparing phased price rises from Q1 2026 [6]. Earlier TrendForce-cited reporting guessed at an RTX 5090 price of up to about $5,000 [6]. Tom's Hardware's tracker later showed $8,477 on 26 October 2026 [1]. Treat this as a rough comparison only: forecasts and best-listed prices use different methods, and a tracker's lowest listing is not an average street price. The RX 9070 XT's $599 launch MSRP became $719 on 3 August and $869 on 26 October 2026 [2][1].
Is it a good time to buy a graphics card?
That depends on how badly you need a card, and the sources lean toward buying sooner if you do. This is general information, not financial advice, and prices change daily.
What is the case for buying now?
- Tom's Hardware says lowest-ever US prices for this generation are barely below launch MSRP, if at all, and that major sales such as Black Friday and Prime Day have lacked notable below-MSRP discounts [2]. TweakTown, however, reported that Black Friday 2025 brought many cards below MSRP [9], so sale results vary.
- It says rumours place the next GPU generation in late 2027 or even 2028, and that new generations are expected to cost more because of memory costs [2].
- Its advice is that if you need a card now, you should buy at the lowest price available, because waiting for the next generation often does not save money [2].
- TweakTown said in June 2026 that waiting for a Radeon price drop looks like a losing strategy [10].
What is the case for waiting?
- The premium over MSRP is large: the RTX 5090 was 324% above its launch price on 26 October 2026 (Pouyam's calculation) [1].
- Tom's Hardware's case for buying now rests on memory and VRAM costs staying high [2]. The case for waiting is Pouyam's own reading: a memory-price reversal or a Black Friday 2026 discount cannot be ruled out, although no source reviewed here forecasts either.
- If your current card still runs the games you play, the cost of waiting is small.
What practical steps help?
- Set a budget and a minimum VRAM size before you look at listings.
- Check the card's price history, not just today's price, since the figures above moved sharply in under three months.
- Consider a lower tier. The RTX 5060 and RX 9060 XT 8GB sat at $449 and $459 on 26 October 2026 [1].
- Look at previous-generation cards, but compare prices first: Tom's Hardware says RTX 40-series and Radeon 7000-series cards have held their value, and its tracker listed the RTX 4090 at $4,450 against a $1,599 launch price, the RTX 4070 Super at $969 against $599 and the RTX 4060 at $530 against $299 [2][1].
- Check the retailer's return policy before you buy.
What is the best GPU to buy in 2026?
There is no single best card, because the answer depends on resolution, refresh rate, VRAM needs, ray-tracing use and budget. Tom's Hardware itself frames the choice as a trade-off: NVIDIA is slightly ahead on DLSS but costs more, while AMD has offered more VRAM on lower-end models [2]. The worked examples below use 26 October 2026 best US prices and are budget illustrations, not recommendations [1].
Worked example: a budget of about $500
The RTX 5050 ($419), RTX 5060 ($449) and RX 9060 XT 8GB ($459) fit under $500. The RTX 5060 Ti 8GB ($529) does not [1]. At this level the question is whether 8GB of VRAM suits the games you play. The RX 9060 XT 16GB at $599 is the cheapest 16GB card in this list from either maker [1].
Worked example: a budget of about $900
Three cards fit: the RX 9070 at $699, and the RTX 5070 and RX 9070 XT at $869 each. The RTX 5070 Ti, at $1,229, is over budget [1]. Between the two $869 cards, the trade-off is NVIDIA's DLSS lead against AMD's lower price per tier, so the deciding factor is the features you use.
What do players actually own?
The Steam Hardware & Software Survey for October 2026 shows the GeForce RTX 3060 back at the top with a 3.78% share, tied with the RTX 5070 at 3.78%, with RTX 5060-series cards at roughly 2.7-3.15% [11][12]. The September 2026 survey had shown the RTX 5070 on top at 6.15%, so month-to-month swings are large [12]. The October table lists the RTX 5090 at 0.43% and the Radeon RX 9060 XT at under 1% [11]. Table scope (all users or Linux only) varies, so confirm it against the live page before quoting. The survey covers Steam users only, and one month is a weak basis for ranking cards.
What about used and previous-generation cards?
They can make sense, but they are not automatically cheaper. Tom's Hardware says the RTX 40-series and Radeon 7000-series have held their value [2], and the October 2026 tracker prices above show why [1].
How are graphics card sales holding up despite higher prices?
Overall shipments are steady. JPR reports that the global PC GPU market, integrated plus discrete, shipped 75.5 million GPUs in Q2 2026, up 10.4% on the previous quarter and 1.1% on a year earlier [4].
What do the segments show?
Discrete GPU (dGPU) shipments rose 12.2% quarter on quarter in Q2 2026, even as the memory crisis pushed component prices up [4]. TechSpot adds that JPR's figures put discrete GPUs up 14.1% on a year earlier [7]. JPR's release gives no unit total for discrete GPUs, and it attributes the rise to a mix of supply-side positioning, artificial demand shocks and localised market dynamics [4]. Desktop graphics shipments fell 4% on the quarter while notebook GPU shipments rose 16.8% [4]. So the discrete increase does not mean desktop graphics card buyers are paying more happily.
How does that compare with 2025?
JPR reported 76.6 million PC GPUs shipped in Q3 2025, with a GPU attach rate of 120%. Attach rate is the number of GPUs, integrated and discrete combined, shipped per PC. Jon Peddie cited rising memory prices and tariff uncertainty as disruptions to normal seasonality [13]. By Pouyam's calculation, Q2 2026 shipments of 75.5 million were about 1.1 million below that Q3 2025 figure, although the two quarters are not directly comparable because of seasonality [4][13].
Are buyers rushing to beat price rises?
That is a theory, not a finding. JPR's explanation for the rise in discrete shipments does not mention buyers stocking up [4], and no source here shows that buyers are stocking up before further increases.
Will GPU prices go down in 2027?
No source reviewed here forecasts falling graphics card prices. The sources that look ahead expect costs to stay high, and the picture is uncertain.
What do the sources expect?
Tom's Hardware says the RTX 50-series and Radeon RX 9000-series were primarily released from January 2025, that rumours place the next GPU generation in late 2027 or even 2028, and that new generations are expected to cost more because of memory costs [2]. TweakTown reported in June 2026 that AMD planned a further 10-15% Radeon increase in Q3 2026, though this was a board-channel report and not confirmed by AMD [10].
What should buyers watch?
- Memory prices: DRAM and GDDR contract prices, since memory is the main cost driver cited [6].
- Company earnings: NVIDIA reported Q2 fiscal 2027 results on 26 August 2026, and the Data Center share of revenue is one clue, in Pouyam's reading, to where supply priorities lie [3].
- Launch timing: any confirmation of the next-generation window beyond the current rumours [2].
- Shipment data: JPR's quarterly figures on discrete GPU shipment growth [4].
What is still uncertain?
Pouyam's caution is that a memory-price reversal or a Black Friday 2026 discount cannot be ruled out, although none of the sources reviewed forecasts either. The split between memory cost and margin is unproven, and a tracker's best price can move sharply in weeks, as the RTX 5090's $4,288 on 3 August and $8,477 on 26 October show [2][1]. Treat any single forecast, including those in this article, as provisional.
| Card | Launch MSRP | Best US price, 3 Aug 2026 | Best US price, 26 Oct 2026 | Premium over MSRP (26 Oct) |
|---|---|---|---|---|
| GeForce RTX 5090 | $1,999 | $4,288 [2] | $8,477 [1] | +324% |
| GeForce RTX 5080 | $999 | $1,256 [2] | $1,699 [1] | +70% |
| GeForce RTX 5070 Ti | $749 | $949 [2] | $1,229 [1] | +64% |
| GeForce RTX 5070 | $549 | $629 [2] | $869 [1] | +58% |
| GeForce RTX 5060 Ti 16GB | $429 | $569 [2] | $765 [1] | +78% |
| GeForce RTX 5060 Ti 8GB | $379 | $429 [2] | $529 [1] | +40% |
| GeForce RTX 5060 | $299 | $339 [2] | $449 [1] | +50% |
| GeForce RTX 5050 | $249 | $289 [2] | $419 [1] | +68% |
| Radeon RX 9070 XT | $599 | $719 [2] | $869 [1] | +45% |
| Radeon RX 9070 | $549 | $639 [2] | $699 [1] | +27% |
| Radeon RX 9070 GRE | $549 | Not listed in source | $599 [1] | +9% |
| Radeon RX 9060 XT 16GB | $349 | $449 [2] | $599 [1] | +72% |
| Radeon RX 9060 XT 8GB | $299 | $359 [2] | $459 [1] | +54% |
| When | Reported change | Evidence |
|---|---|---|
| December 2025 | AMD reportedly raised prices on cards and memory kits sent to board partners by $10 per 8GB of VRAM; Tom's implied MSRPs such as RX 9070 XT $599 to $619 [8] | Anonymous distribution source relayed by Tom's Hardware, not an AMD announcement; the implied MSRPs are Tom's inference [8]. TweakTown repeats Tom's report [9] |
| January 2026 | A further, undefined AMD increase was said to be scheduled [8][9] | Same anonymous source relayed by Tom's Hardware and repeated by TweakTown; no confirmation in the sources reviewed |
| May 2026 | NVIDIA reportedly raised RTX 5090 partner pricing [10] | TweakTown citing board-channel sources; NVIDIA had not told partners of a new increase [10] |
| Q3 2026 | AMD planned a further 10-15% RX 9000 increase because of GDDR6 costs [10] | Gazlog and Chinese board-channel sources, flagged as not definitive and not confirmed by AMD [10] |
Common mistakes
Treating one headline price as the market price
Fix: Quote the date, the method and the country with every price. The RTX 5090 was $4,288 on 3 August 2026 and $8,477 on 26 October 2026, and both are best US listings, not averages.
Comparing cards by name and ignoring the VRAM variant
Fix: Check the memory size in the exact model name. In the 26 October 2026 index the RX 9060 XT 16GB was $140 above the 8GB version, and the RTX 5060 Ti 16GB was $236 above its 8GB sibling.
Waiting for a below-MSRP sale
Fix: Do not plan around it. Tom's Hardware found no notable below-MSRP discounts at Black Friday or Prime Day for this generation, although TweakTown reported many cards below MSRP on Black Friday 2025. Set a budget and buy when a card fits it.
Treating a channel rumour as a confirmed price rise
Fix: Separate confirmed facts from reports. AMD's reported 10-15% Q3 2026 rise came from board-channel sources and was not confirmed by AMD, so treat it as a signal, not a fact.
Ranking cards from a single month of the Steam survey
Fix: Look at several months and remember the sample. The RTX 5070 led Steam at 6.15% in September 2026 and tied at 3.78% in October.
Assuming previous-generation cards are cheap
Fix: Check current listings. On 26 October 2026 the RTX 4090 was $4,450 against a $1,599 launch price, and Tom's Hardware says RTX 40-series and Radeon 7000-series cards have held their value.
Frequently asked questions
Why are graphics card prices so high in 2026?
Sources point to AI data-centre demand that has swamped the memory industry, raising VRAM costs, which reportedly make up more than 80% of the bill of materials of some high-end GPUs [1][6]. NVIDIA's Data Center unit generated $89.0 billion of its $96.2 billion Q2 fiscal 2027 revenue [3]. How much of the rise is memory cost and how much is margin is unproven in the sources reviewed.
How much does an RTX 5090 cost now?
Tom's Hardware's tracker, updated 26 October 2026, lists a best US price of $8,477 against a $1,999 launch MSRP [1]. The same kind of figure was $4,288 on 3 August 2026 [2]. These are lowest listed US prices, not averages, and prices differ by country [5].
Is it a good time to buy a graphics card?
Tom's Hardware found no notable below-MSRP discounts at sales such as Black Friday and Prime Day, and advises buying the lowest price available if you need a card now [2]. TweakTown said in June 2026 that waiting for a Radeon price drop looks like a losing strategy [10]. Pouyam's caution: a memory-price reversal cannot be ruled out. This is general information, not financial advice.
Will GPU prices go down in 2027?
No source reviewed here forecasts falling prices. Tom's Hardware says rumours place the next GPU generation in late 2027 or even 2028, and expects new generations to cost more because of memory costs [2]. TweakTown reported further Radeon increases planned for Q3 2026 [10]. A memory-price reversal remains possible but unforecast.
NVIDIA RTX or AMD Radeon: which is better value?
On 26 October 2026 the RX 9070 XT was $869 and the RTX 5070 Ti $1,229 [1]. Tom's Hardware says NVIDIA is slightly ahead on DLSS versus AMD's FSR but its cards cost more, while AMD's RX 9000 cards have offered more VRAM on lower-end models [2]. Value depends on your games, resolution and budget.
Are used or previous-generation graphics cards a bargain?
Not necessarily. Tom's Hardware says the RTX 40-series and Radeon 7000-series have also held their value [2]. On 26 October 2026 its tracker listed the RTX 4090 at $4,450 against a $1,599 launch MSRP, the RTX 4070 Super at $969 against $599, and the RTX 4060 at $530 against $299 [1].
Are graphics card sales falling because of high prices?
Not in total. Jon Peddie Research (JPR) reported 75.5 million PC GPUs shipped in Q2 2026, up 10.4% on the previous quarter and 1.1% on a year earlier [4]. Discrete GPU shipments rose 12.2% on the quarter, though desktop graphics shipments fell 4% [4]. JPR attributes the discrete rise to a mix of supply-side positioning, artificial demand shocks and localised market dynamics [4].
Glossary
- VRAM
- VRAM is the video memory on a graphics card that holds the textures and frame data the GPU is working on.
- GDDR6 and GDDR7
- GDDR6 and GDDR7 are generations of graphics memory used on consumer graphics cards.
- HBM
- HBM, or high-bandwidth memory, is stacked memory placed beside the chip, used mainly on AI accelerators.
- MSRP
- MSRP is the manufacturer's suggested retail price, the launch price a maker names for a card.
- AIB (add-in-board) partner
- An AIB partner is a company that builds and sells graphics cards around a GPU from NVIDIA or AMD.
- DLSS and FSR
- DLSS is NVIDIA's AI-based upscaling technology and FSR is AMD's; both render at lower resolution and reconstruct a sharper image to raise frame rates.
- Discrete GPU (dGPU)
- A discrete GPU is a separate graphics card or chip, as opposed to an integrated GPU built into the processor.
- Attach rate
- Attach rate is the number of GPUs, integrated and discrete combined, shipped per PC.
Sources
- GPU price tracking 2026 — Lowest price on every graphics card from Nvidia, AMD, and Intel today (page last updated 26 October 2026) Tom's Hardware · 2026-10-26
- GPU prices keep surging in 2026 — Lack of sale discounts and rising VRAM costs keep current-gen prices sky-high Tom's Hardware · 2026-08-03
- NVIDIA Announces Financial Results for Second Quarter Fiscal 2027 NVIDIA · 2026-08-26
- PC GPU shipments increased 10.4% Q-Q, up 1.1% Y-Y (Q2 2026 Market Watch) Jon Peddie Research · 2026-09-03
- Absurd GPU Pricing Update: Mid-2026 Edition TechSpot
- GPU pricing set for reset as AI-driven memory shortages push costs sharply higher Astute Group (citing TrendForce)
- GPU shipments surged 12% despite weaker PC demand and higher memory costs TechSpot
- AMD raises Radeon RX 9000 GPU prices, increasing by $10 for every 8GB of VRAM — another price hike is also scheduled for January 2026 Tom's Hardware
- Radeon RX 9000 Series price increases confirmed, second price increase coming January 2026 TweakTown
- AMD plans to increase the price of its graphics cards by 10-15% in the second half of 2026 TweakTown
- Steam Hardware & Software Survey: October 2026 Valve (Steam)
- NVIDIA RTX 3060 Takes Back Number One Spot From RTX 5070 While 32 GB System RAM Took A Massive Hit According To The Latest Steam Hardware Survey Wccftech · 2026-11-02
- 3Q25 GPU Shipments Up 2.5% as Data Center GPUs Surge 145% I-Connect007 (reproducing Jon Peddie Research) · 2025-11-25
