Affiliate Disclosure: As an Amazon Associate we earn from qualifying purchases. Some links may earn us a commission at no cost to you.

Wait for the RTX 5080 Super or Buy an RTX 5080 Now? (2026)
Buy the RTX 5080 now. The RTX 5080 Super is real, it is reportedly sitting finished at board partners, and NVIDIA is holding the launch over memory supply that nobody has put a date on.
The one exception: if 24 GB of VRAM is the entire reason you are shopping this tier, the wait is defensible. Everything else about the refresh is smaller than the headlines read. Below is where the window stands today, what the leaked spec says, and the three cards worth buying now.
Quick verdict
The rule is short. Buy now if you game at 4K, your current card is dead or going, and you want to play this year. Step down to the RTX 5070 Ti if the number bothering you is the price rather than the memory. Hold out only if you already know a 16 GB card will not carry your workload, and go in knowing the wait has no announced end.

At a glance
Your situation | Card | Why | Buy |
|---|---|---|---|
4K now, and the old card is dying | Full GB203 today. The refresh adds memory, not cores. | ||
1440p, or 4K with upscaling | Single-digit gap at 1440p for a tier less money. | ||
Set on waiting for 24 GB | Current-gen stopgap that holds 1440p until the refresh lands. |
4K now, and the old card is dying
- Card
- Why
Full GB203 today. The refresh adds memory, not cores.
- Buy
1440p, or 4K with upscaling
- Card
- Why
Single-digit gap at 1440p for a tier less money.
- Buy
Set on waiting for 24 GB
- Card
- Why
Current-gen stopgap that holds 1440p until the refresh lands.
- Buy
Specs at a glance
Spec | RTX 5080 Super (rumored) | |
|---|---|---|
Die | NVIDIA GB203 | NVIDIA GB203, same silicon |
CUDA cores | 10,752 | 10,752, unchanged |
Memory | 16 GB GDDR7 | 24 GB GDDR7 on 3 GB modules |
Memory speed | 30 Gbps | 32 Gbps |
Bandwidth | 960 GB/s | About 1 TB/s |
Board power | 360 W | 400 to 415 W |
Availability | On shelves now | Unannounced. CES 2027 is the leading rumored window. |
Source | Shipping specification | kopite7kimi leak, via Tom's Hardware and Igor's Lab |
Die
- RTX 5080 (buy today)
NVIDIA GB203
- RTX 5080 Super (rumored)
NVIDIA GB203, same silicon
CUDA cores
- RTX 5080 (buy today)
10,752
- RTX 5080 Super (rumored)
10,752, unchanged
Memory
- RTX 5080 (buy today)
16 GB GDDR7
- RTX 5080 Super (rumored)
24 GB GDDR7 on 3 GB modules
Memory speed
- RTX 5080 (buy today)
30 Gbps
- RTX 5080 Super (rumored)
32 Gbps
Bandwidth
- RTX 5080 (buy today)
960 GB/s
- RTX 5080 Super (rumored)
About 1 TB/s
Board power
- RTX 5080 (buy today)
360 W
- RTX 5080 Super (rumored)
400 to 415 W
Availability
- RTX 5080 (buy today)
On shelves now
- RTX 5080 Super (rumored)
Unannounced. CES 2027 is the leading rumored window.
Source
- RTX 5080 (buy today)
Shipping specification
- RTX 5080 Super (rumored)
kopite7kimi leak, via Tom's Hardware and Igor's Lab
How we weigh the Super rumors
Start with what is confirmed, because the honest answer is nothing. NVIDIA has not announced an RTX 5080 Super, has not shown one, and has not given anyone a date. VideoCardz reported this year that the cards had already reached board partners and that the launch was put on hold anyway. TweakTown and TrendForce have since moved the expected window into early 2027, with CES as the leading guess. The October slot the rumor mill circled in August passed without a product. That is the single most useful data point a buyer has, and it is not in the spec sheet.
Then read the leak for what it says rather than what it implies. The spec attributed to kopite7kimi, carried by Tom's Hardware and analysed by Igor's Lab, keeps the same GB203 die and the same 10,752 CUDA cores as the card on shelves. What changes is 24 GB of GDDR7 instead of 16 GB, memory running at 32 Gbps instead of 30, and board power climbing from 360 W toward 415 W. Igor's Lab puts the likely uplift around 7 to 14 percent, and credits bandwidth and clocks rather than cores. A refresh with no extra shaders is a memory product. Treat it as one.
The last piece is pricing the wait honestly. The launch is held for the same reason every card is expensive: Guru3D, KitGuru and OC3D all report that the 3 GB GDDR7 modules the refresh needs cost roughly three times what the 2 GB chips in current cards do, with that supply pulled toward AI accelerators. A card built from the most expensive component on the market does not arrive cheap, and refresh launches sell out before they settle. If you want the lineup-wide version of this call rather than the 5080 tier alone, we cover the whole RTX 50 Super timing question separately.
Buy now for 4K: ASUS TUF RTX 5080 OC

Specs
Chip | NVIDIA GB203 (Blackwell) |
CUDA cores | 10,752 |
Memory | 16 GB GDDR7 |
Memory bus | 256-bit, 960 GB/s |
Board power | 360 W |
Power connector | 16-pin 12V-2x6 |
Cooling | Triple-fan, 3.6-slot |
Outputs | HDMI 2.1b, 3x DisplayPort 2.1b |
Chip
NVIDIA GB203 (Blackwell)
CUDA cores
10,752
Memory
16 GB GDDR7
Memory bus
256-bit, 960 GB/s
Board power
360 W
Power connector
16-pin 12V-2x6
Cooling
Triple-fan, 3.6-slot
Outputs
HDMI 2.1b, 3x DisplayPort 2.1b
What it does well
This is the top of the consumer stack below the 5090, and it already uses the full GB203 die. That matters more than it sounds. There is no better-binned version of this chip waiting in the wings, because the rumored Super is the same silicon with denser memory bolted to it. Whatever the refresh turns out to be, it is not a faster GPU in any sense a frame counter will show you.
In practice it holds 4K with ray tracing in current AAA titles once DLSS 4 is doing its share, and it has the raster headroom to run high-refresh 1440p without thinking about it. If the open question is whether this tier clears 4K at all rather than whether the refresh clears it better, we worked through whether a 5080 is enough for 4K in detail.
What you give up
16 GB is the number buyers argue about, and the Super is the card that answers it. For most 4K gaming today that pool is fine. For heavy texture mods, VR simulators, or local AI work, it is the ceiling you will hit first, and that is the one case where holding out is a real decision rather than a reflex.
There is also the gap to the 5090 above it, which never stopped feeling like a tier that went missing. And watch the variant: ASUS sells a BTF version of this same card whose detachable GC-HPWR connector only works in back-connect motherboards. Buyers have been caught by that listing. The standard OC Edition is the one you want unless your build is already BTF.
Who it's for
The 4K buyer whose card has died or is on the way out, who is not willing to leave a build open against an unannounced product. Give it a CPU that will not hold it back, which we sized up in our CPU pairings for the RTX 5080.
The value step down: ASUS TUF RTX 5070 Ti OC

Specs
Chip | NVIDIA GB203 (Blackwell) |
CUDA cores | 8,960 |
Memory | 16 GB GDDR7 |
Memory bus | 256-bit, 896 GB/s |
Board power | 300 W |
Power connector | 16-pin 12V-2x6 |
Cooling | Triple-fan, 3.125-slot |
Outputs | HDMI 2.1b, 3x DisplayPort 2.1b |
Chip
NVIDIA GB203 (Blackwell)
CUDA cores
8,960
Memory
16 GB GDDR7
Memory bus
256-bit, 896 GB/s
Board power
300 W
Power connector
16-pin 12V-2x6
Cooling
Triple-fan, 3.125-slot
Outputs
HDMI 2.1b, 3x DisplayPort 2.1b
What it does well
I have called the 5070 Ti the quiet value winner of this generation since launch, and the Super delay has not changed that. It carries the same 16 GB GDDR7 pool as the 5080, the same DLSS 4 feature set, and the same GB203 family, on a board that pulls meaningfully less power.
At 1440p the measured gap to the 5080 sits in single digits, which is the sort of difference that shows up on an overlay and nowhere else. For a 1440p 144 Hz or 180 Hz panel this card already delivers the experience the monitor is capable of.
What you give up
The 4K ray-tracing headroom, and it is a real loss rather than a rounding error. Reviewer comparisons put the 5080 roughly 13 percent ahead at 4K and closer to 15 percent once ray tracing is on. That margin is often the difference between a path-traced title sitting above or below a frame rate you would accept.
You also inherit the same 16 GB question the 5080 has, so stepping down does not solve the thing the Super is rumored to fix. For the tier picture above this card, see how the 5080 stacks up against the 5090. One listing note: the White OC Edition is a separate product page at a separate price for identical silicon. Buy on the cooler and the deal, not the colour.
Who it's for
1440p high-refresh buyers, and 4K buyers who lean on upscaling and would rather keep the difference. If you are not sure which side of that line you are on, start with how to size a GPU to the monitor.
The wait-it-out stopgap: Sapphire Pulse RX 9070 XT

Specs
Chip | AMD Navi 48 (RDNA 4) |
Compute units | 64 |
Memory | 16 GB GDDR6 |
Memory bus | 256-bit, 645 GB/s |
Board power | 304 W |
Power connectors | 2x 8-pin |
Upscaling | FSR 4 |
Cooling | Triple-fan, 3.1-slot |
Chip
AMD Navi 48 (RDNA 4)
Compute units
64
Memory
16 GB GDDR6
Memory bus
256-bit, 645 GB/s
Board power
304 W
Power connectors
2x 8-pin
Upscaling
FSR 4
Cooling
Triple-fan, 3.1-slot
What it does well
Deciding to wait still leaves a hole in the case. This is the card that fills it without funding a flagship twice. Tom's Hardware found the 9070 XT trading blows with the 5080 in 4K raster, which is a strange thing to be able to say about a card a full tier below it.
It brings 16 GB, FSR 4, and the clearest value win AMD has landed in several generations. Resale on a current-generation card also holds up better than on anything you would pull out of a parts bin, which is the point of a stopgap you intend to replace.
What you give up
Ray tracing, where the distance to Blackwell is still real, and the DLSS 4 ecosystem that comes with it. FSR 4 closed much of the upscaling gap, but the path-traced showcase titles are still where Radeon gives ground.
Sapphire also lists more than one Pulse RX 9070 XT, so match the SKU rather than the name. Stock at this tier has been thin since launch, and the memory squeeze has not made that better.
Who it's for
The reader who has decided the 24 GB card is worth waiting for and wants something capable in the slot meanwhile, without writing off the money when the refresh finally arrives.
Bottom line
If you game at 4K and need a card this quarter, buy the RTX 5080 and stop watching the rumor feed. If the sticker is what is stopping you rather than the memory, buy the 5070 Ti and spend the difference on the monitor. If 24 GB is genuinely load-bearing for your workload, hold out, and put the 9070 XT in the slot while you do. Nobody should hold a build open for a product with no announced date, built from the one component the whole market is short of. We will revisit this call every month the window stays open.
FAQ
When is the RTX 5080 Super coming out?
There is no announced date. NVIDIA has never confirmed the card exists. VideoCardz reported that finished boards had already reached partners before the launch was put on hold, and TweakTown and TrendForce now point at an early-2027 window with CES as the leading guess. The October 2026 slot the rumor mill circled came and went with no product, which is the strongest signal a buyer currently has.
Is the RTX 5080 Super worth waiting for?
Only if 24 GB of VRAM is the specific thing you are buying. The leaked spec keeps the same GB203 die and the same core count as the card you can buy today, so the gaming gain is modest. Against that you are holding your build open against an unannounced product for an unknown number of months. For most 4K buyers that trade does not pay.
How much faster will the RTX 5080 Super be than the RTX 5080?
Igor's Lab reads the leaked configuration as roughly 7 to 14 percent, and credits the faster memory and higher clocks rather than extra shaders. The rumored card keeps all 10,752 CUDA cores of the standard 5080. Anything promising a generational jump is reading the VRAM figure, not the compute one.
Will the RTX 5080 get cheaper when the Super launches?
Do not plan around it. The shortage delaying the refresh is the same shortage pushing current card prices up, and both NVIDIA and AMD have raised what partners pay this year. A refresh built from 3 GB GDDR7 modules that cost roughly triple the 2 GB parts is not arriving as a discount event, and refresh stock tends to sell out before pricing settles.
Is 16 GB of VRAM enough for 4K gaming in 2026?
For 4K gaming with upscaling, yes, in current titles. Where 16 GB starts to bind is heavy texture packs, VR simulators, large scenes in 3D tools, and local AI work. If your use case is on that list, the Super is aimed squarely at you and waiting is rational. If it is not, buy the card that is on the shelf.
Related Articles

Should You Wait for the RTX 50 Super or Buy Now? (2026)
Should you wait for the RTX 50 Super or buy a GPU now? We break down the rumored VRAM jump, the memory-shortage delay, and which current cards are safe buys.
Jul 23, 2026

Do You Need a 5090, or Is a 5080 Enough for 4K? (2026)
Do you need a 5090, or is a 5080 enough for 4K gaming? The RTX 5080 clears 4K today, and the 5090 is a productivity and headroom buy, not a gaming must.
Aug 12, 2026

Best CPUs for the RTX 5080 (2026): 5 No-Bottleneck Picks
The best CPU for an RTX 5080 depends on the resolution you play at. Five picks by tier, with measured CPU scaling at 1080p, 1440p and 4K, and where it stops.
Aug 14, 2026

Is 16GB VRAM Enough in 2026? 1440p, 4K and 24GB Verdict
Is 16GB VRAM enough in 2026? Yes for 1440p ultra and nearly all 4K. Only 4K path tracing breaks it. See which games exceed 16GB and when 24GB is worth it.
Oct 5, 2026

Best Headset for Helldivers 2 (2026): 5 Picks for PC and PS5
Best headset for Helldivers 2 in 2026: 5 picks that place bug flanks by ear and keep stratagem callouts clear on PC and PS5, from budget wired to planar.
Oct 5, 2026