Nvidia’s most-rumored graphics cards of 2026 just got more complicated. Leaked spec sheets for the RTX 50 Super refresh, the RTX 5070 Super, RTX 5070 Ti Super and RTX 5080 Super, have circulated across enthusiast hardware sites for months, promising the VRAM boost that RTX 50-series buyers have wanted since launch. New reporting through August 2026 tells a different story: the cards are reportedly finished and already sitting with add-in board partners, yet Nvidia has quietly held back the launch, according to Tom’s Hardware, Wccftech and German outlet Basic Tutorials.
The holdup has little to do with chip design and everything to do with memory economics. The 3GB GDDR7 modules needed for the bigger VRAM pools now reportedly cost close to three times what the 2GB chips in current RTX 50 cards cost. That single line item is reshaping Nvidia’s launch calendar and pushing the RTX 5080 Super release date to CES 2027 in January, now the leading launch window in industry reporting, instead of a 2026 ship date.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
RTX 5080 Super: Quick Answer
The short version, based on reporting through August 2026:
- Launch window: Delayed out of 2026 and into early 2027, with CES 2027 in January now the leading target, according to TrendForce, VideoCardz and Tom’s Hardware.
- Power draw: Rumored at 415W TGP, up from 360W on the standard RTX 5080.
- Why the delay: Denser 3GB GDDR7 modules reportedly cost $60 to $70 each, roughly 3x the approximately $20 price of the 2GB modules used in current RTX 50 cards.
What Just Leaked About the RTX 50 Super Lineup
The RTX 50 Super rumor cycle has run for more than a year, but the picture sharpened considerably in mid-2026 when unreleased SKUs began turning up inside third-party tools built for PC system builders. Tom’s Hardware reported that unannounced RTX 50 Super cards had surfaced inside Seasonic’s online power supply calculator, a tool builders use to size a PSU before buying parts. The listing showed the unreleased cards carrying 10 to 17 percent higher total graphics power (TGP) than their non-Super counterparts, an unusually specific detail for a product Nvidia has never acknowledged.
That leak followed a wave of spec-sheet roundups from Wccftech and VideoCardz-sourced reporting, all converging on largely the same numbers for three core models: RTX 5070 Super, RTX 5070 Ti Super and RTX 5080 Super. A fourth card, RTX 5060 Super with 12GB of VRAM, appears in some but not all of the leaked roundups, and a cut-down RTX 5050 refresh with roughly 9GB has also surfaced in European reporting. None of this is confirmed by Nvidia. Every figure in this story traces back to leaks, partner-facing tools, and supply-chain sourcing rather than an official spec sheet.
Model-by-Model: RTX 5070 Super, 5070 Ti Super and 5080 Super Specs
Across the leak roundups, the specs converge on the same basic pattern: significantly more VRAM, unchanged or barely changed CUDA core counts, and higher power budgets. Here is how the rumored Super cards reportedly compare to Nvidia’s current, officially shipping RTX 50 lineup, based on spec tables published by Wccftech, Tom’s Hardware and Basic Tutorials.
| GPU | Current VRAM | Rumored Super VRAM | VRAM Change | CUDA Cores (Current → Super) | TGP (Current → Super) |
|---|---|---|---|---|---|
| RTX 5070 | 12GB GDDR7 | 18GB GDDR7 | +50% | 6,144 → 6,400 | 250W → 275W |
| RTX 5070 Ti | 16GB GDDR7 | 24GB GDDR7 | +50% | 8,960 → 8,960 | 300W → 350W |
| RTX 5080 | 16GB GDDR7 | 24GB GDDR7 | +50% | 10,752 → 10,752 | 360W → 415W |
| RTX 5060 (rumored) | 8GB GDDR7 | 12GB GDDR7 | +50% | Not yet detailed | Not yet detailed |
The pattern is consistent: every rumored Super model gets a 50 percent VRAM increase, while shader counts stay completely flat on the two higher-end cards. Only the RTX 5070 Super reportedly gets a small CUDA core bump, from 6,144 to 6,400, alongside its memory upgrade. Nvidia is also expected to raise memory clock speeds on the flagship card, with the RTX 5080 Super reportedly running GDDR7 at up to 32 Gbps versus 30 Gbps on the current card, pushing memory bandwidth close to 1TB/s according to multiple leak roundups.
RTX 5060 Super and RTX 5050 Rumors Are Less Certain
The lower end of the stack is murkier. Some outlets, including Tom’s Hardware, report a RTX 5060 Super with 12GB of VRAM, up from 8GB on the standard card, though detailed CUDA core and clock figures haven’t consistently surfaced. Basic Tutorials also references a cut-down RTX 5050 variant with roughly 9GB of memory among the affected models. Because fewer sources corroborate these two cards compared with the 5070 Super, 5070 Ti Super and 5080 Super tier, they carry more uncertainty and could still be dropped or reworked before any eventual launch.
Why Nvidia Is Reportedly Delaying the Launch
The specs alone don’t explain the holdup, since Nvidia has shipped memory-heavy refreshes before without much drama. What’s different this time, according to reporting from Basic Tutorials and echoed by other European outlets, is the input cost of the memory itself. Nvidia needs higher-density GDDR7 modules, 3GB per chip instead of the 2GB modules used in current RTX 50 cards, to hit the leaked VRAM figures without changing the memory bus width. Those 3GB modules are reportedly running $60 to $70 each on the spot market, compared with roughly $20 for the 2GB modules Nvidia currently buys. That’s close to triple the cost for only 50 percent more capacity per chip, a trade that erodes board margins fast once it’s multiplied across eight to twelve memory chips per card.
The GDDR7 Cost Crisis, By the Numbers
Run the math across a full board and the pressure becomes clearer. A RTX 5080 Super’s 24GB frame buffer, built from higher-density 3GB modules, still needs a full memory bus worth of chips, but at roughly three times the per-chip price, the bill of materials for memory alone reportedly climbs by several hundred euros per card. Multiply that across the hundreds of thousands of boards Nvidia’s add-in card partners typically build for a launch window, and it becomes a cost increase measured in the tens of millions of dollars before a single card reaches a shelf. Nvidia has not commented publicly on Super pricing plans, and no outlet has published a confirmed MSRP for any Super model, which is notable given how detailed the spec leaks have become.
415 Watts: What the Power Jump Means for Your PC
Power draw is the other headline number. The RTX 5080 Super is reportedly rated at 415W TGP, up from 360W on the current RTX 5080, an increase of about 15 percent. The RTX 5070 Ti Super reportedly climbs to 350W from 300W, roughly 17 percent higher, while the RTX 5070 Super moves to 275W from 250W, about 10 percent higher. Those aren’t trivial jumps for anyone running a mid-tower case with a 650W or 750W power supply, especially once CPU headroom and case airflow are factored in.
It’s also why the Seasonic PSU calculator leak was such a clean source for Tom’s Hardware. PSU sizing tools need accurate TGP figures to function correctly, which means Seasonic’s backend arguably carried more precise numbers than the spec-sheet leaks culled from partner roundups. Anyone planning a RTX 5080 Super build should expect PSU and case-cooling advice to be a recurring theme once official reviews eventually land.
From Seasonic Calculators to Board Partners: How the Leak Surfaced
Unlike a typical pre-launch leak that traces back to a single insider account, the RTX 50 Super story has emerged from an unusually wide spread of sources. Tom’s Hardware’s Seasonic calculator find gave the clearest power-draw numbers. VideoCardz-linked spec-sheet leaks, repeated by Wccftech and other roundup sites, supplied the VRAM, memory bus and CUDA core figures. Separately, European reporting including Basic Tutorials described finished boards already sitting with add-in card partners, waiting on a Nvidia go-ahead that hasn’t come. Taken together, the sourcing suggests the hardware itself is real and largely finalized, while the commercial decision about when, and at what price, to sell it remains unresolved inside Nvidia.
Timeline Confusion: 2026 Launch Hopes vs a CES 2027 Reality
Timing has been the least stable part of this story. Earlier 2026 coverage treated a Super refresh as likely sometime within the year, in keeping with Nvidia’s history of mid-cycle updates roughly 12 to 18 months after a series launch. By August 2026, that uncertainty has hardened into consensus: market researcher TrendForce, as relayed through VideoCardz and Tom’s Hardware, now reports the refresh sliding out of 2026 entirely and into early 2027, with CES 2027 in January repeatedly cited as the leading launch window rather than one option among several. Tom’s Guide’s rumor tracker and Digital Trends have both logged the same back-and-forth: a launch window that keeps sliding rather than firming up.
No Official Word From Nvidia
It’s worth stating plainly: everything in this leak cycle, model names, specs, pricing expectations and the delay itself, comes from third-party reporting, partner-facing tools and supply-chain sourcing, not from Nvidia directly. That’s normal for a Super-style refresh, which Nvidia has historically confirmed only close to launch, but it also means every figure here should be read as the current rumor baseline rather than a settled specification.
Historical Context: How the RTX 40 Super Refresh Compared
Nvidia has run this playbook before. The RTX 40 Super series, RTX 4070 Super, RTX 4070 Ti Super and RTX 4080 Super, launched officially in January 2024, about 15 months after the RTX 40 series debuted. That refresh combined modest core-count increases with sharper pricing and, on the 4070 Ti Super, a VRAM bump that addressed a specific complaint about the original card. It was an official, date-certain launch with full reviews on day one. The rumored RTX 50 Super cycle looks different in almost every respect.
| Aspect | RTX 40 Super (Jan 2024) | RTX 50 Super (Rumored, Early 2027) |
|---|---|---|
| Status at this stage | Officially launched | Unconfirmed, reportedly delayed |
| Time since base series launch | ~15 months | ~20 months and still slipping |
| Main change | Modest core increase + pricing tuning | Large VRAM increase, mostly flat cores |
| Power impact | Minimal | +10% to 17% TGP across three models |
| Reported driver | Stack rebalancing | GDDR7 memory cost pressure |
The contrast matters for buyers. When the RTX 40 Super launched, it arrived, got reviewed, and settled the “wait or buy” question within days. The RTX 50 Super rumor cycle has instead stretched across most of 2026 without a confirmed date, a very different experience for anyone trying to time a GPU purchase around it.
Market Impact: Should You Buy an RTX 50 GPU Now or Wait
For shoppers, the practical question is simple and the answer is frustratingly not: buy a current RTX 5070, 5070 Ti or 5080 now, or wait for a Super version that might not arrive until 2027. The VRAM increases being rumored are large enough to matter for anyone running 4K textures, frame generation, or newer AAA titles that lean on video memory more than earlier generations did. At the same time, an indefinite delay means waiting could mean going months without a GPU upgrade for a refresh with no confirmed price or ship date.
Retailers and system integrators face a related bind. Continuing to sell current RTX 50 stock at full price is harder to justify publicly once VRAM-boosted versions are known to exist, even unofficially, which can quietly soften demand for the exact cards Nvidia still needs to move through the channel before any Super launch. That dynamic is part of why some analysts expect Nvidia to keep the delay quiet rather than acknowledge it outright, since an official “it’s coming, just later” statement risks freezing sales of current inventory almost immediately.
Competitive Landscape: Where AMD and Intel Stand
The delay also changes the competitive picture, at least temporarily. AMD’s current RX 9070-series cards compete directly with the non-Super RTX 5070 and 5070 Ti on price, and a stalled Nvidia refresh gives AMD more runway to hold its position in that segment without an imminent VRAM-boosted response to plan around. Intel’s Arc lineup remains focused on the budget-to-midrange tier and isn’t a direct competitor at the 5070 Ti Super or 5080 Super level, but a delayed Nvidia refresh similarly reduces the pressure on Intel to accelerate any higher-end Arc plans.
Neither AMD nor Intel has announced a direct answer to the rumored RTX 50 Super specs, and neither company has publicly commented on Nvidia’s reported GDDR7 cost pressure. That said, both companies buy from the same limited pool of memory suppliers and would likely face similar cost dynamics on any VRAM-heavy refresh of their own, which is one reason the entire GPU industry is watching how Nvidia resolves its Super pricing math.
Analyst and Reviewer Reaction: A Capacity Fix, Not a Performance Leap
Coverage across the enthusiast press has converged on a similar read: this is a memory-capacity correction, not a generational performance jump. Because CUDA core counts are reportedly flat on the 5070 Ti Super and 5080 Super, the expected gains come almost entirely from having more VRAM headroom rather than more raw shading power. That’s being framed positively by outlets that criticized the original RTX 50 cards’ VRAM as tight for the price tier, and more skeptically by those noting that a 10 to 17 percent higher power draw is a real cost for a refresh that isn’t delivering a matching jump in rendering performance. Digital Trends’ ongoing RTX 50 tracker has described the pattern as Nvidia fixing its own segmentation rather than pushing the generation forward.
What This Means for PC Gamers and Builders
If the rumored specs hold, the practical impact splits by budget tier. Builders eyeing a 5080-class card for 4K gaming get the clearest win: 24GB is enough headroom for texture-heavy modern titles and leaves room for future games, at the cost of needing a beefier PSU and better case airflow to comfortably handle 415W. Buyers at the 5070 tier get a smaller but still meaningful bump, useful for 1440p gaming with frame generation enabled. Anyone building a new PC this summer faces the classic refresh dilemma without a firm end date to plan around, which is arguably the most concrete impact of this leak cycle so far: it’s already shaping purchase decisions before Nvidia has said a single word publicly.
5 Predictions for the RTX 50 Super Launch
- A CES 2027 reveal is more likely than a standalone 2026 announcement. Given the reported delay and the marketing value of a CES keynote slot, Nvidia has strong incentive to save the reveal for January rather than a quiet mid-cycle press release.
- Pricing lands at or slightly above current RTX 50 launch prices, not a steep hike. Nvidia has historically avoided sticker shock on Super refreshes even when its own input costs rise, preferring to protect the perceived value of the upgrade.
- The RTX 5080 Super and RTX 5070 Ti Super ship closer to the currently leaked specs than the RTX 5060 Super and RTX 5050 variants. The top two cards have far more cross-source corroboration; the lower tier could still change or get dropped entirely.
- PSU and case-cooling advice becomes a recurring review theme. A 415W flagship all but guarantees that reviewers spend real column space on power-supply headroom once cards actually ship.
- Further slippage into spring 2027 remains a real risk. If GDDR7 pricing eases before early 2027, a faster announcement becomes plausible; if memory costs stay elevated, expect another quiet delay past CES.
Related Coverage
- Nvidia, AMD Hike GPU Kit Prices: China Retail Up 22% [2026]
- RTX 5090 vs RTX 5070 Ti for AI: 2x VRAM, 2.7x Price [2026]
- RTX 5060 Ti 8GB vs 16GB: The Quick Answer
- AMD RX 9070 GRE Ends China Exclusivity at $549 [2026]
- RTX 5070 Ti vs RX 9070 XT: The Quick Answer
- RAM Shortage 2026: AI Chips Hit 5 Gaming Platforms’ Prices
- How to Undervolt a GPU: 12 Steps, 45 Min [2026]
- Nvidia RTX Spark Debuts: 128GB Memory, 1 Petaflop AI [2026]
For broader coverage of AI chips, GPUs and hardware pricing trends, see the AI chips and hardware hub.
Frequently Asked Questions
When will the RTX 50 Super series launch?
Nvidia hasn’t confirmed a date. Rumors point to a delay from an expected 2026 window to CES 2027 in January, now the leading launch window according to TrendForce, Tom’s Hardware and Basic Tutorials, driven by reported memory cost pressure.
How much VRAM will the RTX 5080 Super have?
Leaked specs point to 24GB of GDDR7, up from 16GB on the current RTX 5080, a 50 percent increase, according to spec roundups from Wccftech and other outlets.
Why is Nvidia delaying the RTX 50 Super launch?
Reports point to the rising cost of higher-density GDDR7 memory chips. The 3GB modules needed for the bigger VRAM pools reportedly cost around $60 to $70 each, compared with about $20 for the 2GB modules used in current cards.
Will the RTX 50 Super cards use more power than current models?
Yes, based on the leaks. The RTX 5080 Super is rumored at 415W TGP versus 360W on the current RTX 5080, and the RTX 5070 Ti Super is rumored at 350W versus 300W on the current RTX 5070 Ti.
Is the RTX 5060 Super confirmed?
No. A 12GB RTX 5060 Super has appeared in some leak roundups, including from Tom’s Hardware, but it’s less consistently corroborated than the RTX 5070 Super, RTX 5070 Ti Super and RTX 5080 Super.
How does this compare to the RTX 40 Super launch?
The RTX 40 Super series launched officially in January 2024 with modest core increases and pricing adjustments. The RTX 50 Super cycle has stretched much longer without an official date and centers on a far larger VRAM increase instead.
Should I buy a current RTX 50 GPU now or wait for the Super refresh?
It depends on your timeline. If the Super cards slip to CES 2027 as rumored, buyers who need a GPU now may be better off purchasing current RTX 50 stock rather than waiting for a refresh with no confirmed price or ship date.
Where did the RTX 50 Super leaks come from?
Multiple sources: unreleased cards appearing inside Seasonic’s PSU calculator tool, as reported by Tom’s Hardware; spec-sheet roundups linked to VideoCardz and repeated by Wccftech; and European supply-chain reporting from outlets including Basic Tutorials.
