Nvidia’s next GeForce generation just lost another year. On September 21, 2026, hardware leaker Kopite7kimi posted that “GR20x is a product from 2028,” referring to the Rubin-based chips expected to power the RTX 60 series, and added that “they have postponed again.” Tom’s Hardware reported the claim the same day, noting Nvidia’s only response was its standard line that the company does not comment on rumors or unannounced products.
That single social post would be easy to dismiss on its own. But it lands on top of a pile of consistent, and consistently worsening, leaks stretching back to early 2026, and it isn’t only about Nvidia anymore. According to leaker Kepler_L2, AMD’s next Radeon lineup is caught in the same freeze, with only one new die, codenamed AT2, expected to reach the market in 2027. Everything else from both companies, per the current run of leaks, is sliding into 2028. That’s the real story: an entire generation of PC gaming hardware, from both major GPU vendors, stalling out at roughly the same moment, for roughly the same reason. This piece looks at what’s driving the freeze, what it means for anyone holding an RTX 50 or Radeon RX 9000 card today, and why the GPU upgrade cycle gamers grew used to over the past decade may not come back on schedule.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Kopite7kimi actually said about RTX 60
Kopite7kimi is one of the most closely watched Nvidia leakers on social media, with a track record that includes early and accurate calls on RTX 40 and RTX 50 series specifications before official announcements. TechPowerUp had already flagged the architecture behind the shift back in March 2026, identifying Rubin as the silicon tied to the next-generation RTX 60 lineup, and by July 2026 TechPowerUp reported that Kopite7Kimi himself had moved his RTX 60 estimate out to 2028. On September 21, the account described the upcoming GR20x platform, the internal codename tied to those Rubin-based consumer GPUs, as slipping to 2028 once again. That’s a downgrade from the leaker’s own earlier position, which had placed the same hardware in late 2027.
The timing matters because it isn’t an isolated data point. As far back as December 2025, Tom’s Guide reported leaks pointing to a possible RTX 60 announcement in Q1 2027, and by February 2026 Tom’s Hardware was already reporting that mass production might not start until 2028, delayed beyond the 2027 window many had assumed. A week before the latest Kopite7kimi post, on September 15, a separate report citing Moore’s Law Is Dead placed the RTX 60 family in the first half of 2027, after what MLID described as an earlier slip from an original 2026 target. Within six days, the expected window moved by roughly a year, from early 2027 to sometime in 2028. Whether that reflects an actual schedule change inside Nvidia or simply multiple leakers converging on different snapshots of an unstable roadmap, the direction of travel across every report since late 2025 is the same: later, not sooner.
Tech Insider covered the RTX 60 series skipping 2027 entirely when that framing first surfaced, and the newest Kopite7kimi post is the strongest single confirmation yet that the later date is the one sticking.
AMD’s next Radeon generation is caught in the same freeze
The angle that hasn’t gotten as much attention is what this means for AMD. Leaker Kepler_L2 reported that AMD’s AT2 graphics die is expected to be the only major new gaming GPU the company brings to market in 2027. The rest of AMD’s next-generation Radeon lineup, according to that same reporting, is pushed into 2028 alongside Nvidia’s RTX 60 series.
If that holds, it means PC gamers won’t get a meaningful new-generation choice from either of the two dominant GPU makers until 2028, regardless of which brand they prefer. That’s a different problem than an Nvidia-specific delay. A single vendor slipping a launch is a company-level story. Two competing vendors slipping in the same window, for overlapping reasons, points at shared, structural constraints in the supply chain both companies depend on, namely advanced node capacity at TSMC and high-bandwidth memory production.
Intel, the third player in the discrete GPU market with its Arc lineup, has not had a comparably specific 2028 claim attached to its next-generation Celestial architecture in current reporting. That leaves an open question about whether Intel could use a slower Nvidia/AMD cadence to close the gap, though nothing in the available leaks confirms an accelerated Arc timeline either.
Why AI is eating the GPU roadmap
The explanation showing up across multiple outlets is consistent: Nvidia is prioritizing data center silicon over consumer graphics cards. MLID’s reporting described the delay in blunt terms, framing it as a decision to let “AI gobble up as much TSMC capacity as possible this year.” Tom’s Hardware’s own headline made the same point directly, describing gaming as taking “a back seat as company focuses on delivering data center products.”
That prioritization isn’t a mystery from a business standpoint. Nvidia’s data center segment, anchored by Blackwell and the upcoming Vera Rubin platform, commands vastly higher margins per wafer than consumer GeForce cards. When advanced packaging and leading-edge node capacity is the scarce resource, and every customer from OpenAI to Meta to Amazon is trying to buy more of it, a company allocating that capacity toward its highest-margin, highest-demand product line is following the money. Gaming GPUs, however central they are to Nvidia’s brand and to PC building culture, are a smaller fraction of the company’s revenue than they were five years ago.
Memory supply compounds the problem. Reporting on the broader 2026 GPU memory crunch has pointed to constrained DRAM and high-bandwidth memory production as a second drag on consumer GPU output, independent of the AI capacity question. Tech Insider previously covered the HBM4 crunch that has left DRAM inventory sitting below ten days across parts of the supply chain, a level that leaves almost no buffer if any single supplier hits a snag.
The RTX 50 Super refresh appears to be a casualty
One of the more concrete pieces of collateral damage in this story is the RTX 50 Super refresh, the mid-cycle update that would normally have bridged the gap between RTX 50 and RTX 60. PC Gamer reported in February 2026, citing The Information, that Nvidia might ship zero new gaming GPUs that entire year, and as of August 2026 Nvidia still had not announced an RTX 60 model, price, or launch date of any kind. Multiple reports describe the Super lineup as finished from a design standpoint but deprioritized, with some coverage describing it as postponed indefinitely and other coverage describing it as effectively scrapped. Neither outcome has been confirmed directly by Nvidia, and the difference between “postponed” and “canceled” matters enormously to anyone deciding whether to wait for a refresh or buy now.
What is consistent across the reporting is a production cut on the existing RTX 50 lineup itself: multiple sources, including original claims traced to China’s BoBantang forum and corroborated by Benchlife citing AIC partners and component suppliers, point to RTX 50-series production being reduced by 30 to 40% in the first half of 2026 compared with the same period the prior year. Tech Insider tracked what that supply cut did to RTX 5090 pricing, which climbed as availability tightened even before the RTX 60 delay news broke.
Breaking a decade of GeForce cadence
Nvidia’s consumer GeForce lineup has followed a rough two-year cadence for most of the past decade: the GTX 10 series launched in 2016, RTX 20 in 2018, RTX 30 in 2020, RTX 40 in 2022, and RTX 50 in early 2025. As recently as August 2025, PC Guide was still pegging RTX 60 for a 2026-2027 window with no indication of an earlier release, before the run of 2026 leaks pushed that estimate out repeatedly. Even accounting for the pandemic-era supply chain chaos that stretched some of those gaps, a jump from RTX 50’s 2025 launch to an RTX 60 launch in 2028 would represent roughly a three-year interval. GamesRadar put a sharper number on it in September 2026, arguing that a 2028 launch would hand Nvidia its longest GPU-generation gap in 20 years. That’s the framing Tom’s Hardware used too, describing the wait as three years between generations rather than Nvidia’s historical two.
The table below shows how the reported timeline has shifted across just the past several months of leaks, none of which have been confirmed by Nvidia itself.
| Report date | Source | Claimed RTX 60 window |
|---|---|---|
| Early 2026 | Industry roadmap chatter | Originally targeted for 2026 |
| February-April 2026 | Multiple supply chain reports (DRAM shortage coverage) | Pushed to “at least 2028” |
| September 15, 2026 | Moore’s Law Is Dead | First half of 2027 |
| September 16, 2026 | Kepler_L2 / Notebookcheck | 2028, alongside most of AMD’s next-gen Radeon |
| September 21, 2026 | Kopite7kimi / Tom’s Hardware | 2028, “postponed again” |
The pattern in that table is the real headline: the target date has only moved in one direction since early 2026, and the two most recent, most specific reports both land on 2028.
What this means for anyone holding an RTX 50 or RDNA 4 card
For PC gamers, the practical upshot is straightforward: if these reports hold, the graphics card sitting in your PC right now is going to be the top of the market for longer than any GPU generation in recent memory. Anyone who bought an RTX 5090, RTX 5080, or a high-end Radeon RX 9000-series card expecting a two-year window before the next big jump is now looking at closer to three, based on current reporting.
That changes the calculus in a few concrete ways. First, resale and secondary-market value for current-generation flagships is likely to hold up better than it would in a normal cadence, since buyers looking for the fastest available card won’t have a newer option to wait for anytime soon. Second, it puts more pressure on GPU makers to fill the gap with incremental refreshes, memory-capacity bumps, or professional-tier spillover products rather than a clean-sheet architecture change. Tech Insider covered the broader GPU relief timeline sliding to 2028 in the context of DRAM pricing, and the RTX 60 delay is effectively the GPU-specific expression of that same supply squeeze.
Third, this extends beyond desktop cards. Handheld gaming PCs have already been hit by the same DRAM crisis driving up component costs, and a delayed GPU architecture transition means the underlying silicon powering next-generation handhelds is also likely to arrive later than enthusiasts had hoped.
How the GPU delay compares with CPU roadmaps slipping too
The RTX 60 slip isn’t happening in isolation. AMD’s own Zen 6 desktop CPU lineup has reportedly been delayed into 2027, according to separate leaks around the “Powderhorn” chiplet design, pointing to the same foundry capacity and packaging bottlenecks rippling across CPUs as well as GPUs. When both halves of a gaming PC’s most expensive components hit the same wall at roughly the same time, it stops looking like a product-specific hiccup and starts looking like an industry-wide capacity ceiling tied to how much of TSMC’s advanced-node output is being absorbed by AI accelerators.
The table below lays out how several major hardware roadmaps compare against their originally expected timing, based on the leaks and reporting gathered so far.
| Product line | Originally expected | Currently reported | Primary cited cause |
|---|---|---|---|
| Nvidia RTX 60 (GeForce, Rubin-based) | 2026 | 2028 (per Kopite7kimi, Sept. 21) | AI/data-center capacity priority, memory constraints |
| Nvidia RTX 50 Super refresh | 2026 | Deprioritized / unclear status | Reallocated design resources, supply constraints |
| AMD next-gen Radeon (ex-AT2) | 2027 | 2028 (per Kepler_L2) | Shared foundry/memory constraints with Nvidia |
| AMD AT2 die | 2027 | 2027 (largely on track) | Smaller die, less capacity pressure |
| AMD Zen 6 desktop CPUs | 2026 | 2027 (per Powderhorn leak coverage) | Chiplet/packaging capacity |
Why leaker consensus doesn’t equal Nvidia confirmation
It’s worth being precise about what has and hasn’t actually been confirmed here. Nvidia’s only on-record response, given to Tom’s Hardware, was a refusal to comment on rumors or unannounced products, which is the company’s standard answer regardless of whether a leak is accurate. That is not a denial, and it is not a confirmation. Tech Insider’s earlier coverage of the top-leaker claim and the separate report of a second leaker backing the same 2028 window both rely on the same category of evidence: consistent, credible, but unofficial sourcing.
Kopite7kimi and MLID both have real track records, and having two independently sourced leaks converge on the same year adds weight. But neither leaker has produced a document, a roadmap slide, or an on-record Nvidia statement. Specific claims like an HBM4 shortage being the exact mechanism behind the delay remain plausible but unconfirmed by any Nvidia or TSMC statement found in current reporting. Readers should treat 2028 as the best current estimate from credible sources, not as a locked-in release date.
Market and investor angle
From a pure business standpoint, the delay is not necessarily bad news for Nvidia’s bottom line, even if it frustrates PC gamers. Nvidia’s gaming segment has become a smaller share of total revenue as data center sales built around Blackwell and the coming Vera Rubin platform have scaled dramatically. Every wafer and every allocation of CoWoS advanced packaging capacity that goes toward an AI accelerator instead of a GeForce card is, in the current pricing environment, a wafer generating substantially more revenue per unit. That reality is the throughline connecting the RTX 60 delay, the RTX 50 Super’s uncertain status, and the broader restructuring of Nvidia’s product priorities that outlets have been documenting since early 2026.
For AMD, the calculus is different, since AMD doesn’t have nearly the same scale of AI accelerator business subsidizing a gaming delay. If AMD’s next Radeon generation really is also sliding to 2028 largely because of shared foundry constraints rather than a deliberate reallocation toward AI products, that points to TSMC capacity itself being the binding constraint across the whole industry, not just an Nvidia-specific business decision.
What to watch next
A few developments would meaningfully change this picture. An official Nvidia roadmap disclosure, even a vague one at an event like GTC or CES, would carry far more weight than any leaker’s post. A confirmed cancellation or revival of the RTX 50 Super refresh would clarify whether Nvidia plans to offer any mid-cycle bridge product before 2028. And any specific reporting on Intel’s Arc Celestial timeline would clarify whether Intel sees the Nvidia/AMD slowdown as an opening to gain share in the discrete GPU market, something current reporting does not yet establish either way.
It’s also worth watching TSMC’s own capacity disclosures. Since the RTX 60 delay is repeatedly tied to advanced-node and packaging capacity being consumed by AI accelerators, any signal that TSMC is expanding CoWoS or 2nm-class capacity faster than expected could pull GPU timelines back forward, just as continued AI demand growth could push them out even further than 2028.
Predictions for how this plays out
- Nvidia will likely stay silent on an official RTX 60 date through at least its next two quarterly earnings calls, continuing to decline direct comment on the leaks while emphasizing data center results.
- Expect at least one more mid-cycle GeForce SKU or memory-capacity variant on the RTX 50 architecture as a stopgap, given the reported deprioritization of the original RTX 50 Super plan.
- AMD is more likely than Nvidia to attempt an earlier partial launch, given reports that its AT2 die is still tracking toward 2027, potentially giving AMD a marketing window before Nvidia’s next generation arrives.
- Used and current-generation flagship GPU prices, including RTX 5090 and high-end Radeon cards, are likely to stay elevated or rise further through 2027 given tightened production and a longer wait for replacements.
- Intel’s Arc Celestial timeline will become a more closely watched story over the next two quarters, as press and analysts probe whether Intel can use the Nvidia/AMD gap to gain ground, even without confirmed leaks pointing that direction yet.
Frequently asked questions
Is the Nvidia RTX 60 delay to 2028 officially confirmed?
No. It comes from leaker Kopite7kimi’s September 21, 2026 post, reported by Tom’s Hardware. Nvidia has only said it does not comment on rumors or unannounced products, which is neither a confirmation nor a denial.
Why is Nvidia reportedly delaying its next GeForce generation?
Reporting points to two overlapping causes: Nvidia prioritizing TSMC advanced-node and packaging capacity for its data center Rubin AI accelerators, and broader memory supply constraints affecting the whole GPU industry.
Is AMD’s next Radeon generation also delayed to 2028?
According to leaker Kepler_L2, most of AMD’s next-generation Radeon lineup is also tracking toward 2028, with only the AT2 die expected to arrive in 2027.
What happened to the RTX 50 Super refresh?
Multiple reports describe the RTX 50 Super as design-complete but deprioritized, with some coverage saying it’s postponed indefinitely and other coverage suggesting it may not launch at all. Nvidia has not confirmed either outcome.
How long is the gap between RTX 50 and RTX 60 expected to be?
If the 2028 timeline holds, the gap would stretch to roughly three years from RTX 50’s 2025 launch, longer than Nvidia’s roughly two-year cadence maintained through most of the past decade.
Should I buy an RTX 5090 or wait for RTX 60?
Based on current reporting, waiting means a significantly longer hold than in past GPU generations. As recently as November 2025, PC Guide had pegged a potential RTX 6090 launch window of January-March 2027, but that estimate has since slipped sharply, and anyone needing a high-end GPU now is unlikely to see a next-generation replacement before 2028 at the earliest.
Is an HBM4 shortage confirmed as the cause of the RTX 60 delay?
Not directly. HBM4 is primarily tied to Nvidia’s data center Rubin platform, and while broader DRAM and memory shortages are widely reported across the GPU industry, no Nvidia or supplier statement has confirmed HBM4 availability as the specific mechanism delaying GeForce RTX 60.
Could Intel benefit from Nvidia and AMD both delaying next-gen GPUs?
It’s possible, but current reporting does not include a specific timeline change for Intel’s Arc Celestial architecture, so any competitive advantage remains speculative rather than confirmed.
Related Coverage
- Nvidia's Top Leaker Confirms RTX 60 Delay to 2028 [2026]
- Second Leaker Backs RTX 6000 Delay to 2028 [2026]
- RTX 6090 Release Date Slips to 2028: 3-Year Wait [2026]
- AMD Hits $1T: Wall Street Splits $76 on Next Move [2026]
- AMD Joins $1T Club, Stock Jumps 10% Past Intel [2026]
