What the RTX 50 Super refresh is supposed to be
Nvidia has never officially announced a "Super" refresh of the RTX 50 (Blackwell) desktop lineup. Everything circulating comes from leakers and supply-chain reports, and you should treat it as rumor until Nvidia says otherwise.
That said, the leaks have been remarkably consistent. The core idea is a mid-cycle refresh that keeps the same GPU dies and memory bus widths but swaps in new 3GB GDDR7 memory chips. Because each module holds 50% more data, Nvidia could bump VRAM capacity by roughly 50% without redesigning the boards.
The rumored lineup looks like this:
| Model (rumored) | Rumored VRAM | Current card | Current VRAM |
|---|---|---|---|
| RTX 5060 Super | 12GB | RTX 5060 | 8GB |
| RTX 5070 Super | 18GB | RTX 5070 | 12GB |
| RTX 5070 Ti Super | 24GB | RTX 5070 Ti | 16GB |
| RTX 5080 Super | 24GB | RTX 5080 | 16GB |
The pitch is obvious: more VRAM is exactly what the current RTX 50 stack is criticized for lacking, especially the 12GB RTX 5070 and 16GB RTX 5080. An 18GB RTX 5070 Super and a 24GB RTX 5080 Super would age far better for 4K and heavily modded games.
Why the RTX 50 Super may be delayed to 2027
Here is the twist that changed the whole conversation. Reports through mid-2026 indicate the Super series has slipped from a rumored 2026 launch to a potential CES 2027 reveal — and some supply-chain chatter goes further, describing the refresh as "delayed indefinitely" after Nvidia reportedly briefed its board partners. In other words, even the 2027 timing is a rumor stacked on a rumor.
The culprit is memory. The exact same high-density 3GB GDDR7 chips that would give these Super cards their marquee VRAM bump are in severe short supply. Memory makers are pouring wafer capacity into HBM and other chips for AI data centers, where margins dwarf what they earn selling GDDR7 to gaming-GPU vendors.
Industry analysts have described AI as consuming close to 20% of global DRAM wafer capacity in 2026, and GDDR7 reportedly needs around 1.7x the manufacturing resources of standard DRAM. Multiple outlets, citing memory suppliers like Samsung and SK hynix, expect the crunch to persist into 2027 and possibly 2028, when new fabrication capacity finally ramps.
In that environment, launching a whole new lineup that depends on scarce, expensive memory makes little business sense for Nvidia. Reports even suggest Nvidia cut RTX 50-series supply to partners by roughly 20% in the first half of 2026 to conserve GDDR7. If anything, the delay is the rational move.
Should I wait for RTX 50 Super? The honest answer
For most US buyers, the honest answer is: do not put your life on hold for it. Here is the reasoning.
First, the timeline is a moving target. Best case, you are looking at a CES 2027 announcement, which historically means retail availability weeks or months after the reveal. Refresh launches also tend to sell out instantly and command inflated prices at first — we saw exactly that with the original RTX 5080 and 5090 Founders Editions, which sold out in minutes at MSRP.
Second, nothing is confirmed. Betting a year of your gaming life on an unannounced product is a gamble, not a plan. Nvidia could change the VRAM configs, the pricing, or the timing entirely.
Third — and this is the part people miss — waiting does not obviously save you money. The memory shortage inflating today's prices is the same shortage delaying the Super cards. When those refreshed GPUs do arrive, they will be built from the most expensive component in the current market. There is little reason to expect aggressive MSRPs.
If you have a working GPU that runs your games acceptably, waiting is reasonable — not because the Super will be cheap, but because you lose nothing by holding. If your card is dead or genuinely holding you back, waiting 12-plus months for a rumor is a poor trade. Our upgrade advisor tool can help you gauge whether your current setup is actually the bottleneck or if a CPU or RAM upgrade would do more.
What today's RTX 50 prices actually look like
To decide, you need a clear picture of what you would pay right now. Street prices in July 2026 sit well above the original MSRPs because of the memory crunch. Approximate US pricing:
| GPU | Launch MSRP | Approx. July 2026 street price | VRAM |
|---|---|---|---|
| RTX 5060 (8GB) | $299 | Near MSRP | 8GB |
| RTX 5060 Ti (16GB) | $429 | Near-to-modestly above | 16GB |
| RTX 5070 | $549 | ~$599 | 12GB |
| RTX 5070 Ti | $749 | ~$900-$1,250 | 16GB |
| RTX 5080 | $999 | ~$1,249+ | 16GB |
| RTX 5090 | $1,999 | ~$2,999+ | 32GB |
Note the pattern: the higher up the stack, the worse the markup. The RTX 5090 has been spotted at roughly 50% or more over MSRP — around $2,999 at the friendlier end and far higher for premium AIB models — and the 5080 has run 25-45% high, commonly landing near $1,249. The RTX 5070 Ti has been the poster child for dysfunction — a $749 MSRP card that has spent much of 2026 selling for $900 to $1,250, with even Prime Day deals only pulling it back to roughly $900. Meanwhile the entry-level RTX 5060 and 5060 Ti have stayed closest to sanity because their smaller memory configs are less exposed to the GDDR7 squeeze.
Always cross-check live prices at Newegg, Best Buy, Amazon US, and Micro Center before buying — trackers move daily, and Micro Center in particular occasionally has in-store bundles that beat online listings.
The buy-now case
Buy now if any of these describe you:
- Your GPU is dead, dying, or so old that you are missing modern features (DLSS 4, decent ray tracing, current driver support).
- You game at 1080p or 1440p and can get an RTX 5060 Ti 16GB, 5070, or a fairly priced 5070 Ti without paying scalper-adjacent markups.
- You need the machine for work — content creation, 3D, local AI — where the card pays for itself.
The wait case
Waiting makes sense if:
- You already own a capable card (RTX 30-series, RTX 40-series, RX 6000/7000, or better) that handles your games today.
- You specifically want the extra VRAM for 4K, VR, heavy mods, or local AI workloads, where 18-24GB genuinely matters.
- You are targeting the top of the stack, where current pricing is at its most abusive and the Super refresh promises the biggest capacity jump.
Don't forget AMD and Intel
Nvidia is not the only game in town, and in a shortage the alternatives matter more. AMD's RX 9070 and 9070 XT have been competitive at 1440p, and crucially they use 16GB of GDDR6 rather than the scarce 3GB GDDR7 — which is a big reason both cards clawed back to their $549 and $599 MSRPs by mid-2026 while equivalent GeForce cards stayed inflated. AMD's FSR upscaling has also closed much of the gap with DLSS. Intel's Arc B-series remains a budget-tier option worth a look for 1080p builds.
Cross-shopping is smart here because the vendors are exposed to the memory crunch unevenly, and Nvidia's reliance on high-density GDDR7 has hit it hardest. Comparing raw positioning across brands on a GPU tier list is a faster way to find value than fixating on a single unreleased Nvidia refresh.
What could change this outlook
A few things would shift the calculus. If Nvidia officially confirms the Super lineup with firm dates and MSRPs, waiting becomes far less speculative. If GDDR7 supply loosens faster than expected, current prices could ease and the Super cards could arrive at friendlier pricing. Conversely, if the shortage deepens, both today's prices and the refresh could get worse — which, counterintuitively, strengthens the case for buying a reasonable card now before things tighten further. Watch CES 2027 in early January as the key checkpoint.
FAQ
Is the RTX 50 Super officially delayed to 2027?
Nvidia has not officially confirmed the RTX 50 Super lineup at all, so technically it cannot be "officially delayed." Multiple credible reports and leakers indicate the refresh has slipped from a rumored 2026 window to a possible CES 2027 reveal, and some go as far as calling it "delayed indefinitely," all driven by the GDDR7 memory shortage. Treat the 2027 timing as a well-sourced rumor, not a guarantee.
Should I wait for RTX 50 Super or buy a GPU now?
If you already have a capable card, waiting costs you nothing and could net you 50% more VRAM. If your GPU is dead or badly outdated, waiting 12-plus months for an unconfirmed product is a weak plan — buy a fairly priced card now, ideally an RTX 5060 Ti 16GB or RTX 5070. Do not assume waiting will save money; the same shortage delaying the Super cards is inflating today's prices.
How much more VRAM will the RTX 50 Super cards have?
Leaks point to roughly 50% more VRAM via new 3GB GDDR7 modules: an 18GB RTX 5070 Super (up from 12GB), a 24GB RTX 5070 Ti Super (up from 16GB), a 24GB RTX 5080 Super (up from 16GB), and a 12GB RTX 5060 Super (up from 8GB). These figures are unconfirmed by Nvidia.
Why are GPU prices so high in July 2026?
An industry-wide memory shortage is the main driver. AI data-center demand has soaked up DRAM and GDDR7 supply, pushing memory costs up sharply, and Nvidia reportedly trimmed RTX 50 supply to partners by around 20% to conserve chips. Because VRAM is a major slice of a GPU's bill of materials, cards like the RTX 5070 Ti, 5080, and 5090 have sold well above their launch MSRPs throughout 2026.
Will waiting for the RTX 50 Super actually save me money?
Probably not on price alone. The Super cards will be built from the same scarce, expensive GDDR7 that is inflating current prices, so aggressive MSRPs are unlikely at launch, and refreshes typically sell out and carry premiums early on. The stronger reason to wait is capacity — more VRAM per dollar — rather than a lower sticker price.
The bottom line
So, should you wait for RTX 50 Super? If your current GPU still gets the job done and you crave more VRAM for 4K or AI work, sit tight through CES 2027 and see if the rumors firm up — you risk nothing by waiting. But if you need a card today, do not chase an unannounced product through a memory shortage that may keep prices elevated well into 2027. Buy the best-value card you can find near MSRP, lean on AMD and Intel alternatives to dodge the worst Nvidia markups, and let the Super refresh prove it is real before you count on it.
