GPU Prices in November 2026: AMD's Second Price Hike, the Vanishing RTX 5080, and Prices 77% Above MSRP — The Complete Buyer's Guide

GPU prices in November 2026 have gone from bad to worse. In the last week of October, AMD was reported to be preparing a second price hike in two months for the Radeon kits it sells to board partners, while Nvidia quietly pulled the GeForce RTX 5080 from its own US store as the card's street price hit 2.2 times its launch price. Fresh November tracking data shows the typical graphics card now sells for 77% above its launch MSRP on Amazon — and the flagship RTX 5090 is listed at nearly 285% above MSRP. If you are planning to buy a graphics card before Black Friday on November 27, this guide gives you the verified facts, the rumours (clearly marked), and a practical buy-or-wait plan for every budget.

The Short Version: November 2026 GPU Market at a Glance

If you only have sixty seconds, here is the state of the market as of November 2, 2026:

What's happening	The key fact	Status
AMD Radeon November price hike	Second kit-price increase in two months reported; GDDR6 up 8–10% in Q3	Reported, not confirmed by AMD
RTX 5080 pulled from Nvidia US store	Listings and filters gone; cheapest in stock at $2,199.99, 2.2x the $999 MSRP	Verified by Tom's Hardware (Oct 30)
Street prices vs MSRP	Median 77% above launch price across 8 chips (Basketprice, Nov 2026)	Measured data
RTX 5090	Cheapest listing $7,699.99 — 285% above its $1,999 MSRP	Measured data
24GB RTX 5080	Rumoured for December 2026–January 2027 (CES 2027 window)	Rumour, unconfirmed
New GPU generations in 2026	None from Nvidia, AMD, or Intel — RTX 60 expected late 2027 at the earliest	Confirmed via industry reporting
Black Friday	November 27 — the first real test of whether partners absorb AMD's hike	Date confirmed

The bottom line for buyers: prices are still rising, not falling. Anyone with a concrete need who finds a card near its launch price has little to gain by waiting — but nobody should pay panic prices either. Read on for the data behind each claim and the tier-by-tier buying strategy.

The Numbers: GPU Prices in November 2026, Chip by Chip

The most reliable street-price snapshot comes from Basketprice's GPU price tracker, which records the cheapest in-stock new offer for each graphics chip on Amazon. Its November 2026 reading makes grim reading: of the 8 chips with a launch price on record, all 8 cost more today than at launch, with a median gap of 77% above MSRP.

Chip	Cheapest new card today	Launch price	Today vs launch
RTX 5090	$7,699.99	$1,999	285% above
RTX 5080	$2,881.52	$999	~188% above
RTX 5070 Ti	$1,389.99	$749	86% above
RTX 5060 Ti	$616.93	$379 (8GB) / $429 (16GB)	63% / 93% above
RTX 5060	$517.00	$299	73% above
RX 9070 XT	$919.99	$599	54% above
RX 9070	$749.99	$549	37% above — the least-bad deal
RX 9060 XT	$739.99	$299 (8GB) / $349 (16GB)	16GB model 112% above
RTX 4090	$5,199.99	—	Still selling above its old $1,599 MSRP by 3x
RTX 3060	$459.99	—	A 2021 card selling near its old launch price
RTX 3050	$249.99	—	The cheapest new card tracked

Two things stand out. First, the pain is worst at the top: the RTX 5090's cheapest listing is $7,699.99, up from a $6,685 low on September 24 — the flagship's price nearly quadrupled its $1,999 launch price. Second, the Radeon RX 9070 is the only card remotely near its launch price, at 37% above its $549 MSRP — which tells you why AMD's reported price hike matters so much: it threatens the one remaining value corner of the market. Hardware Unboxed, the YouTube channel that tracks the lowest in-stock prices across ten countries every month, found October prices 10% higher on average month over month, with the RTX 5090 alone up 32% from September. The direction is the same in every region — taxes and exchange rates change the numbers, but memory makers sell to a global market.

AMD's Reported November Price Hike: What the Report Actually Says

On October 30, a widely circulated report claimed AMD is preparing another price increase for November 2026 on the "kits" it sells to board partners — after an increase that already took effect on October 1. The source is ChannelGate, a Chinese WeChat channel account that has previously leaked AMD's supply costs, relayed by the industry sites VideoCardz and Board Channels and picked up by TweakTown. Here is the critical context you need to interpret it correctly:

AMD does not sell finished graphics cards. It sells bundles — the GPU chip plus GDDR6 memory — which partners like Sapphire, ASUS, XFX and Vastarmor mount on boards they design themselves, with their own PCBs and cooling. A hike "on the kits" therefore hits partner margins first, and only later — with a delay, and to a different degree — reaches the price tag in the store. An announced kit increase does not mean every Radeon instantly costs 10% more.

According to the report, GDDR6 memory got 8–10% more expensive in the third quarter, which would have pushed AMD to revise kit prices. The report names neither the percentage of AMD's increase nor the models it covers — and that gap matters, because an entry-level 8GB Radeon uses half the memory of a 16GB card and would see a very different increase. The kits had already climbed about 10% over the summer (July per TrendForce, August per TweakTown), and AMD stayed silent then too. In China, the report says AMD raised fourth-quarter suggested prices by 700 yuan for the RX 9070 and 600 yuan for the 16GB RX 9060 XT, and cites a second factor: the cost of TSMC wafers, the foundry that makes AMD's chips, which is weighing on the fourth quarter for GPUs and AM5 chipsets as well. Again: no figures, no confirmation.

What to believe: treat the November hike as a credible but unconfirmed report from a single channel source. AMD has confirmed nothing, and it also said nothing about the summer and October 1 increases. TechPowerUp, cited by TrendForce, notes the impact on final shelf prices may stay under 10% depending on how much partners absorb. The honest reading is that AMD's kit prices are very likely rising — the memory market forces are real and documented — but nobody can say today whether your store price moves 3% or 10%.

Why Memory Is the Villain: GDDR6 Costs Three Times What It Did a Year Ago

Every thread in this story leads back to one component: memory. GDDR6 is the video memory on the Radeon RX 9000 series and a large share of GeForce cards. According to TrendForce, its spot price — immediate sales outside contracts — went from about $2.5 to about $7.5 per gigabyte in a year. Triple. The reason is structural, not a blip: memory makers shifted production capacity to HBM and server DRAM, which feed AI data centers and pay better margins than consumer graphics memory. Through all of 2025, AMD kept its price lists nearly flat thanks to fixed-price supply contracts. Those contracts expired in 2026 — and since then, every quarter has brought an adjustment. The same dynamic is behind the RAM price surge we covered in our November 2026 RAM prices guide: it is the same DRAM, the same shortage, the same AI data centers bidding against you. SK hynix has described a memory shortage lasting until 2030, and anyone hoping to run AI models at home pays twice for the same conflict — memory costs more, and cards with lots of VRAM are the first to vanish from shelves.

The key insight for buyers: GPU prices will not normalize until memory supply does. That is not a 2026 event. Any buying plan that assumes prices "come back down soon" is a bet against the entire memory industry's capacity allocation — a bet no price list can guarantee.

The RTX 5080 Vanishes From Nvidia's Own US Store

The Nvidia chapter of this story opened in early October, when the RTX 5090 was reported out of production and disappeared from US stores. Now it is the younger sibling's turn. According to Tom's Hardware, reporting on October 30, both the listings and the search filter for the RTX 5080 are gone from Nvidia's US marketplace — while the RTX 5070 Ti and RTX 5070 remain listed, marked as sold out. That detail matters: removing the card entirely breaks Nvidia's own pattern of keeping out-of-stock cards visible, and it is what fuels suspicion about the card's status. WindowsForum's roundup of the coverage from Tom's Hardware, VideoCardz and PCMag confirms the timeline: Reddit users first spotted that shoppers could no longer filter for either the 5090 or the 5080 at the end of October 2026.

On the open market, the 16GB RTX 5080 — launched in January 2025 at $999 — now costs far more. Tom's Hardware lists the cheapest model still in stock, a PNY RTX 5080 ARGB Epic-X RGB OC, at $2,199.99: 2.2 times the starting price. Stock is running low and, per the outlet, the card will hardly make it to Black Friday. For now the removal covers only the US store — the British marketplace still lists the RTX 5080 and RTX 5090, per TweakTown — and Nvidia has said nothing at all about the 5080. A storefront removal is not an official discontinuation: as TechStory notes, listings come and go with stock, and the 5090 went through the same thing earlier in October. But when your two fastest consumer cards both vanish from your own store within weeks, the market reads it as a shift in production priorities — quite possibly toward more profitable AI and professional hardware.

The 24GB RTX 5080 Rumour: What Is Actually Claimed

Circulating alongside the store removal is a supply-chain rumour: Nvidia is preparing a revised GeForce RTX 5080 with 24GB of GDDR7 memory, possibly arriving between December 2026 and January 2027 — the CES 2027 window in Las Vegas. The claims come from the same ChannelGate source, via VideoCardz, with coverage from Wccftech, Club386 and PCTestBench. Specifically, the report claims Nvidia has emptied its warehouse of 16GB 5080s, that distribution holds about two months of stock — enough to last roughly to the end of the year — and that in China both the 16GB 5080 and the RTX 5090D V2 (the 5090 variant built for the Chinese market) are being withdrawn, with the 24GB card aimed at China first and a global launch following closely.

Keep every part of that in the rumour column. Nvidia has not announced the card, confirmed its specifications, or named a price or date. Wccftech's Hassan Mujtaba considers January 2027 the more likely window, since Nvidia could show the card at CES. And Tom's Hardware adds a sobering technical note: a 24GB 5080 would still have half the CUDA cores of the RTX 5090 — eight extra gigabytes do not close that performance gap. It would mainly serve two audiences: gamers who play in 4K and hit the 16GB ceiling, and people who run AI models locally and need more VRAM. If it exists, expect it to debut at a premium over the original $999 — memory costs and the prices the 16GB card has already reached make that nearly certain. Anyone waiting for the new card to buy is making a double bet: that it exists, and that it arrives cheaper than today's inflated 16GB prices. Both legs of that bet are unproven.

The RTX 5090 Situation: $7,700 for a $1,999 Card

The flagship tells the same story in starker numbers. The RTX 5090 launched at $1,999; Basketprice's November tracking puts the cheapest new card at $7,699.99 — 285% above MSRP, up from a $6,685 low on September 24. Hardware Unboxed measured a 32% average price jump for the 5090 from September to October across ten countries. The card disappeared from Nvidia's US store in September, and early-October reports said it was out of production — which Nvidia denied. Senior Vice President of Personal Computing Jeff Fisher, speaking at an event in Taipei (reported via 4Gamers and TechSpot), said the RTX 5090 had not been discontinued and would continue shipping, though he acknowledged supply-chain capacity was somewhat constrained. Lowyat's report notes the tension plainly: Nvidia says the 5090 is alive, while both it and the 5080 keep disappearing from the company's own store page.

The practical takeaway: the RTX 5090 is effectively a $5,000–$7,700 product in November 2026, whatever the $1,999 MSRP says. At those prices it competes with used RTX 4090s (cheapest $5,199.99 new, far less used) and professional cards — and for gaming alone, the price-to-performance math is indefensible. Nobody should buy a 5090 at these prices for gaming; the only buyers who can justify it are professionals billing GPU time or local-AI users who genuinely need 32GB of VRAM and cannot wait.

No New Generations in 2026: Why Waiting Won't Save You Soon

Here is the structural reason the market cannot self-correct quickly: 2026 is the first calendar year since the original GeForce 256 shipped in October 1999 in which Nvidia has not launched a single new gaming GPU — no refresh, no Super variant, no budget card. The RTX 50 series, launched across 2025, remains the current lineup with no successor in sight until late 2027 at the earliest (the RTX 60 "Rubin" window cited by leaker Kopite7kimi and reported by Notebookcheck). Nvidia's engineering resources went somewhere else entirely: the RTX Spark, an AI PC superchip announced with Microsoft that has nothing to do with gaming frame rates. "RTX 60 series release date" became one of the most-searched GPU queries of August 2026 — roughly 880 monthly US searches — precisely because buyers are starved for a next generation.

AMD's side is no better for gamers. Its next consumer architecture, RDNA 5 (sometimes discussed under the UDNA branding as AMD unifies its gaming and compute lines), is tracking to mid-2027 or later. Computex 2026 hardware showcases — including an ASRock prototype Phantom Gaming card — were still built around the existing RDNA 4 flagship, the RX 9070 XT, dressed up with new cooling and a 12V-2x6 power connector rather than new silicon. AMD's one real consumer move in 2026 was the global launch of the Radeon RX 9070 GRE, a trimmed-VRAM variant. And Lisa Su's October 12 keynote at the OCP Global Summit made AMD's priorities explicit: an open full-stack AI portfolio of CPUs, GPUs, networking and software, with Helion AI-rack configurations planned for November 2026. That is data-center hardware, not gaming cards. Intel, meanwhile, has announced no new desktop Arc generation; its loudest October GPU news concerned data-center parts. The conclusion from Tech Insider's analysis is blunt: waiting for a new generation means waiting on a rumour — nothing is announced with a date or a price.

Bolt Graphics Powers On Its First Zeus Chip: Hope, Hype, and a 2028 Timeline

The one genuinely new GPU story of the past week comes from a startup. On October 31, Bolt Graphics powered on the first silicon for its Zeus GPU project — a 12nm test chip fabricated by TSMC that, the company says, switched on without issues. The details are worth reading carefully, because they separate what is real from what is marketing:

What is confirmed: the test chip measures just 6mm² and draws 1.4 watts, compared with the 35 watts used by the FPGA prototype Bolt previously ran its design on. Engineers completed bring-up within 24 hours with no functional issues, and the chip was tested across temperatures from -40°C to 85°C. Its featured capability is dedicated hardware for electromagnetic simulation — hardware acceleration of Maxwell's-equation calculations used in designing radar sensors, camera lenses and similar products. CEO and founder Darwesh Singh told HPCwire the chip shows Bolt can run "a highly complex physics algorithm, accelerated in hardware, which no one's done before." Bolt also opened an office in Taipei to stay close to manufacturing partners.

What is not confirmed: everything about performance. Bolt says the chip beat its internal estimates but has published no independently verifiable figures — and a 6mm² validation chip says nothing about whether a full-sized Zeus GPU can deliver the company's headline claims from its March 2025 announcement: 10 times the rendering performance of existing GPUs, 300 times the speed in electromagnetic simulation, and up to 10 times the path-tracing throughput of an RTX 5090. Those claims remain entirely untested. The commercial Zeus chip is a separate design on TSMC's N5P process (a 2020-era node, deliberately chosen to avoid competing for limited leading-edge wafer supply), built around RISC-V processor cores paired with Bolt's own path-tracing hardware, with PCIe cards combining soldered LPDDR5X and DDR5 SO-DIMM slots expandable to 384GB of memory — a genuinely unusual design aimed first at studios, engineers and scientists, with gaming and AI to follow later.

What the timeline really is: Notebookcheck's coverage notes the schedule has already slipped — developer access in Q1 2027, early-access hardware toward the end of 2027, and volume production in 2028 (previously Q4 2027). Two days after the power-on, Reuters reported Bolt was seeking roughly $100 million at a valuation of about $700 million. In April the company said 14,000 businesses, developers and individuals had joined its early-access program with a pipeline worth over $500 million — naming no customers. The honest verdict: Zeus is a real engineering milestone, not a product you can buy. It will not relieve GPU prices in 2026 or 2027. File it under "worth watching," not "wait for it before buying."

Intel Arc in November 2026: Quiet, Cheap-ish, and Still Here

Intel's Arc B-Series remains the budget corner of the market, and October's buying-guide data is still the best picture we have. The two current desktop cards launched at $250 and $219 (per Tom's Hardware) — but street prices run above launch, and the B580 was reportedly out of stock on October 22. For local-AI work, Intel's Arc Pro B70 offers 32GB of ECC GDDR6 at a $949 starting price for Intel's own card — a genuinely interesting VRAM-per-dollar option that the big two do not match at that price. The downsides: street prices above launch, patchy availability, and no new desktop Arc generation announced — Intel's October GPU headlines were all data-center. Arc remains a "check the price today" option: when a B580 or B570 sits near its launch price, it is one of the few honest deals left; when it does not, there is no reason to prefer it.

The Three-Way Market, Honestly Assessed

Strip away the noise and late-2026 graphics come from three families, none with a new generation on the horizon:

NVIDIA's GeForce RTX 50 Series is still the performance and feature leader — DLSS, the best ray tracing, the strongest creator software stack — but it is also the most overpriced family in the data: 73% to 285% above MSRP across the stack, with the top two cards vanishing from Nvidia's own store. You pay the Nvidia tax, and in November 2026 that tax is enormous.

AMD's Radeon RX 9000 Series is the value play under siege. The RX 9070 at 37% above its $549 MSRP is the least-inflated current-generation card on the market, and the 9070 XT at 54% above $599 is the next-best. But the reported November kit hike lands squarely on this family's margins — Radeon's affordability buffer, which TechTimes reported in October had already vanished, is what the hike threatens to erase entirely. AMD's public energy is going to AI infrastructure (the OCP keynote, Helion racks), not to defending gaming prices.

Intel's Arc B-Series is the wild card that never quite became a contender: decent hardware at launch prices that retailers rarely honor, no roadmap news, and a company whose GPU attention is elsewhere. Worth a look at the right price; not worth waiting for.

And behind them, the used market — RTX 4090s at $5,199.99 new but far less used, RTX 3060s at $459.99 new — where the usual "buy used, save money" logic now collides with cards that have been mined on, baked, and sold at prices their sellers saw on the same inflated trackers you are reading.

What to Buy in November 2026: The Tier-by-Tier Guide

General advice is useless in this market; the right move depends on your resolution, your workload, and your patience. Here is the honest breakdown, using the November street prices above:

1080p gaming on a tight budget: the RTX 5060 at $517 (73% above its $299 MSRP) and the RX 9060 XT 8GB are the entry points, but neither is a good deal at these prices — you are paying mid-range money for entry-level cards. If you can find an Intel Arc B580 near its $250 launch price, it is the only honest value in this tier. Otherwise, the used market (a well-cared-for RTX 3060-class card) beats buying new at 73% over MSRP. Do not pay $517 for a $299 card if you can wait.

1440p high-refresh — the sweet spot: this is where the RX 9070 at $749.99 earns its "least-bad deal" title. At 37% above its $549 MSRP it is the closest any current card gets to sanity, and it trades blows with the RTX 5070 ($969.99, 77% above MSRP) while costing $220 less. The RX 9070 XT at $919.99 is the step-up if you want more headroom. If you must buy a current-generation card in November 2026, the RX 9070 is the answer — with one caveat: buy it before AMD's reported hike filters through to shelves, because it is exactly the card the hike will hit.

4K gaming: there is no good answer. The RTX 5080 at $2,199–$2,881 is a $999 card selling at more than double; the RTX 5090 at $7,699.99 is a $1,999 card selling at nearly quadruple. The RTX 5070 Ti at $1,389.99 (86% above its $749 MSRP) is the least-insane 4K-capable Nvidia option, and it is still listed in Nvidia's store. The rumoured 24GB RTX 5080 would be the card to wait for if you need 4K plus VRAM headroom — but it is a rumour with no date, and waiting means gambling that it launches before your need becomes urgent. For 4K in November 2026, the rational move is to wait unless your current card is dead.

Creators (video, 3D, streaming): Nvidia's software stack (NVENC, CUDA, OptiX, the mature creator drivers) still justifies a premium — but not a 77–285% premium. The RTX 5070 at $969.99 is the lowest rung of the Nvidia creator ladder that makes sense; below it, AMD's cards are fine for gaming but weaker in the applications that pay your bills. If your income depends on renders, buy the cheapest Nvidia card that meets your VRAM floor and get back to work — downtime costs more than the premium.

Local AI (running models at home): VRAM is everything, and the market knows it — which is why high-VRAM cards vanished first. The Arc Pro B70 (32GB ECC, $949 starting) is the standout value for local LLMs if you can find it at list. The RTX 5090's 32GB is the dream config at a nightmare price. The rumoured 24GB 5080 would slot in here perfectly — which is exactly why AI users should watch the CES 2027 window. Do not buy a 16GB card for serious local AI in late 2026; you will hit the memory ceiling on day one, as our RAM prices guide explains in the context of the same memory shortage.

The RX 9070: Why One Card Still Makes Sense

It deserves its own section, because it is the exception that proves the rule. At $749.99 — 37% above its $549 launch price — the Radeon RX 9070 is not a good deal by historical standards. It is a good deal relative to everything else on this page: the RTX 5070 costs $220 more for comparable 1440p performance, the 9070 XT costs $170 more, and every Nvidia card above it is 73–285% over MSRP. It is also the card with the most to lose from AMD's reported November hike — the same report cites a 700-yuan Q4 suggested-price increase for the RX 9070 in China. If you have been waiting for a rational mid-range buy, this is the window, and it may be closing. Check prices weekly through November; if the hike lands, the 9070's value lead evaporates and the whole stack becomes uniformly bad.

Black Friday 2026: Will November 27 Bring Relief?

Black Friday falls on Friday, November 27 this year, and it is the single most important date left in 2026 for GPU buyers. Here is what to realistically expect:

What Black Friday can do: retailers discount to move volume, and partners may absorb part of AMD's kit hike to hit holiday price points — the November 27 shelf prices will show whether the hike was passed through or eaten. Previous-generation and overstock cards (the RTX 3060/3050 tier, older Radeons) are the likeliest candidates for genuine discounts, because retailers want them gone.

What it cannot do: fix a supply shortage. Tom's Hardware expects the RTX 5080 to barely make it to Black Friday at all, and no discount can conjure stock of a card Nvidia has pulled from its own store. Flagship-tier cards will not be "deals" — at best, slightly less extreme prices. And with GDDR6 at triple its year-ago price, retailers have no margin to give on current-generation cards; a 10% "Black Friday discount" on a card selling at 77% above MSRP is still 59% above MSRP.

The strategy: set price alerts now on the two or three cards that fit your tier, know their launch MSRPs by heart, and buy on November 27 only if a price lands within striking distance of rational — not because a red "SALE" tag appeared on a card that is still 60% over MSRP. For the full October baseline to compare against, see our October 2026 GPU prices guide.

7 Mistakes GPU Buyers Are Making Right Now

1. Treating the MSRP as the price. The $999 RTX 5080 and $1,999 RTX 5090 do not exist at those prices in November 2026. Every decision must use street prices, checked the same day you buy — yesterday's price is already stale.

2. Waiting for a crash that the data does not support. GDDR6 tripled in a year, contracts reset quarterly, and SK hynix sees shortage conditions into 2030. Hoping for 2021-style normalisation is not a strategy; it is a wish.

3. Buying the 16GB RTX 5080 at 2.2x MSRP "before it disappears." Panic-buying an inflated card because it might vanish is how you lock in the worst price. If you need that performance class, the 5070 Ti at lower premium or patience for the 24GB rumour are both more rational.

4. Ignoring VRAM for local AI. A 16GB card that benchmarks well in games will choke on a 70B-parameter model. Match the VRAM to the workload first, the frame rates second — memory is the binding constraint of 2026.

5. Assuming "used" means "cheap." Used prices track the same inflated market, and carry the added risks of mining wear, no warranty, and baked-out thermal pads. A used card should cost at least 30–40% less than the new street price to be worth the risk — verify seller history and test before you pay.

6. Buying more card than the monitor needs. A 4K-class card for a 1080p 60Hz monitor is money burned — and in this market, burned money is expensive. Buy for the display you own, not the display you imagine.

7. Forgetting the power supply and the 12V-2x6 connector. Modern flagships want serious, clean power — and the newer 12V-2x6 connector standard (seen on Computex 2026's refreshed boards) matters for cable compatibility. Budget the PSU into the total cost before you fall in love with a card.

The November 2026 Decision Framework: Buy, Wait, or Buy Used

BUY NOW if: your current GPU is dead or dying and you need a working PC; you earn money with the GPU and downtime costs more than the premium; or you find a card near its launch MSRP (the RX 9070 at $749.99 is the prime example) before the reported AMD hike reaches shelves.

WAIT if: your current card still runs the games and apps you use; you are targeting 4K (nothing at the top of the stack is rationally priced); or you specifically want the rumoured 24GB RTX 5080 — set a CES 2027 (January) checkpoint and reassess, but do not hold your breath.

BUY USED if: you find a card at 30–40% below new street price from a verifiable seller, you can test it first, and it has not been run 24/7 in a mining farm. Ask for proof of purchase, check temperatures under load, and walk away from any seller who rushes you.

One more honest note: even the GPU makers' own stocks are under pressure — Intel fell nearly 26% and AMD nearly 13% in the brutal tech selloff week ending November 1, on weak guidance and AI-spending jitters. The market is volatile at every level. Do not make a $2,000 decision on a rumour you read in a headline; make it on a price you verified this morning.

Frequently Asked Questions

1. Has AMD confirmed the November Radeon price hike?
No. The report comes from the Chinese channel account ChannelGate via VideoCardz, Board Channels and TweakTown. It names no percentage and no models, and AMD has said nothing — just as it said nothing about the summer and October 1 increases. Treat it as credible but unconfirmed.

2. Why do graphics card prices keep rising?
Because memory costs three times what it did a year ago. TrendForce puts GDDR6 spot prices at roughly $2.5 to $7.5 per gigabyte as DRAM makers shifted capacity to HBM and server memory for AI data centers. AMD's fixed-price contracts expired in 2026, so every quarter now brings an adjustment.

3. Has Nvidia discontinued the RTX 5080?
There is no official confirmation. The card vanished from Nvidia's US store listings and filters (per Tom's Hardware, Oct 30) but remains in the UK store, and Nvidia has not commented. A model withdrawal is one hypothesis; a stock changeover ahead of a rumoured 24GB version is another.

4. Is the 24GB RTX 5080 real?
It is a rumour. ChannelGate via VideoCardz claims a December 2026–January 2027 launch with 24GB of GDDR7, aimed at China first. Nvidia has announced nothing. Do not plan a purchase around it.

5. What is the best-value graphics card in November 2026?
The Radeon RX 9070 at $749.99 — 37% above its $549 MSRP, the smallest premium of any current-generation card tracked, and $220 cheaper than the comparable RTX 5070. Buy before AMD's reported hike reaches shelves.

6. Should I buy a GPU before Black Friday?
It depends on the card. Mid-range Radeons may get slightly pricier if the hike lands, so an early buy at a fair price is defensible. Flagship Nvidia cards have no reason to get cheaper before November 27 — but they also have no stock to discount. Set alerts and buy on verified price, not on hype.

7. Is Bolt Graphics' Zeus going to fix GPU prices?
No — not in 2026 or 2027. The October power-on was a 6mm² validation chip; volume production is scheduled for 2028, and the 10x-performance claims are unverified. It is an engineering milestone to watch, not a buying option.

8. Will there be new GPUs in 2026 at all?
No. 2026 is the first year since 1999 without a new Nvidia gaming GPU launch. The RTX 60 series is expected late 2027 at the earliest, AMD's RDNA 5 no earlier than mid-2027, and Intel has announced no new desktop Arc generation.

9. Is a used RTX 4090 a good alternative to the RTX 5090?
Often yes, with caution. New 4090s list around $5,199.99 — still far below the 5090's $7,699.99 street price — and used ones go lower. But verify the seller, test thermals, and demand at least a 30–40% discount versus new street price to justify the risk.

10. How much VRAM do I need in late 2026?
12GB is the realistic minimum for 1440p gaming with headroom; 16GB is the comfort zone; 24GB+ is for 4K, heavy creation, or local AI. In a memory-shortage market, VRAM capacity holds its value better than raw speed — buy the memory first.

Conclusion: The Market Is Not Your Friend — So Be a Smarter Buyer Than It

November 2026 is arguably the worst month in years to need a graphics card: AMD is reportedly raising prices again, Nvidia's two fastest cards have vanished from its own store, street prices sit 77% above MSRP at the median, and no new generation is coming to reset anything until late 2027. But "worst month" does not mean "impossible month." The buyers who navigate it well will do three things: decide from verified street prices, not launch MSRPs or headlines; target the least-inflated tier for their actual workload (today, that is overwhelmingly the RX 9070 for 1440p); and use Black Friday on November 27 as a checkpoint, not a rescue. The memory shortage behind all of this will not resolve in a quarter — plan your purchase like prices stay high, and any relief becomes a bonus rather than a bet. If this guide saved you from a panic buy, it did its job. Check back after Black Friday: we will update these numbers and tell you whether AMD's hike landed on the shelf or got absorbed.

Sources

All facts in this article were verified against retrieved sources on November 1–2, 2026. Rumours are marked as rumours in the text.

• AMD reportedly raises Radeon prices again as Nvidia pulls RTX 5080 — Pasquale Pillitteri (Nov 1, 2026): ChannelGate report, GDDR6 +8–10% in Q3, China 700/600-yuan increases, RTX 5080 store removal, 24GB rumour, Hardware Unboxed October data.
• GPU prices, November 2026: 77% above MSRP on Amazon — Basketprice: cheapest-in-stock prices per chip vs launch MSRP.
• Best Graphics Cards 2026: RTX 50 vs RX 9000 vs Arc — Tech Insider (Oct 26, 2026): three families, no new generation, Arc launch prices, Arc Pro B70, Radeon affordability buffer.
• Nvidia Skips New Gaming GPUs in 2026: RTX 60 Delay — Tech Insider: no 2026 launch, RTX 60 late 2027, RDNA 5 mid-2027+, RX 9070 GRE, RTX Spark.
• Bolt Graphics Powers On Its First Zeus GPU — Engadget (Oct 31, 2026): 12nm test chip, 6mm², 1.4W, Taipei office, $100M raise report.
• Bolt Graphics' unconventional Zeus GPU clears first hardware tests, but commercial launch slips to 2028 — Notebookcheck: bring-up details, N5P node, 2028 volume timeline.
• Nvidia RTX 5080 Disappears From US Store as Prices Double and 24GB Refresh Rumors Grow — WindowsForum (roundup of Tom's Hardware, VideoCardz, PCMag): store removal timeline, Jeff Fisher 5090 denial.
• NVIDIA Quietly Pulls RTX 5080 From Its Online Store After RTX 5090 — Lowyat (Oct 31, 2026): 2.2x MSRP pricing, 24GB GDDR7 rumour.
• Nvidia GeForce RTX 5080 Disappears From Marketplace as Prices Surge Past $2,200 — TechStory: $2,199.99 cheapest listing, 120% over MSRP.
• RTX 5080 24GB Rumored for December as 16GB Stock Thins — PCTestBench: December 2026–January 2027 window, China-first reporting.

Related reading on AI Tech Hub: our October 2026 GPU prices guide for the baseline these numbers moved from, and our November 2026 RAM prices guide — the same memory shortage, from the system-memory side.