Nvidia’s next flagship graphics card just got a lot more complicated to plan around. Following a fresh wave of leaker claims this week, the picture for the GeForce RTX 6090 has shifted from “sometime in 2027” to a 2028 launch window, according to reporting from VideoCardz, Notebookcheck, and TweakTown. The story itself, the back-and-forth between leaker kopite7kimi and YouTube channel Moore’s Law Is Dead (MLID), has already made the rounds on this site (see our coverage of the second leaker backing the delay). What hasn’t been fully unpacked is why Nvidia would walk away from its own two-year cadence, what it means for the RTX 6090 specifically as the halo card of the lineup, and what buyers sitting on an aging RTX 40-series or RTX 50-series card should actually do about it.
That’s the gap this piece fills: not another recap of who-said-what on X, but a look at the economics behind the delay, the historical pattern it breaks, and the decisions it forces on gamers, system builders, and anyone tracking the RTX 6090 release 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.
What actually changed this week
For most of 2026, the working assumption among hardware leakers was that Nvidia’s Rubin-based gaming architecture, the silicon underpinning the GeForce RTX 60 series and its flagship RTX 6090, would ship in the first half of 2027. That assumption was already baked in well before September: RedGamingTech predicted in November 2025 that RTX 60 could arrive no earlier than January 2027, leaker kopite7kimi posted “2027H2” for RTX 60 in January 2026 (PCWorld reported), and The Information reported in February 2026 that no new Nvidia gaming GPUs were planned for 2026 and that RTX 60 mass production was targeted near late 2027. Itechguides echoed in August 2026 that Nvidia had no new gaming GPUs planned for 2026 at all, and a separate Komputronik roundup from that same month pegged the earliest rumored release at “late 2027.” The timeline traced back largely to Moore’s Law Is Dead, whose reporting was summarized by TweakTown: “According to a new rumor, NVIDIA’s GeForce RTX 60 Series powered by RTX Rubin is reportedly on track for a first-half of 2027 launch,” the outlet wrote on September 14, 2026. Notebookcheck’s own summary of the MLID claim, published around the same window, framed it the same way: the Rubin-based RTX 60 series was “currently targeting a launch in the first half of 2027, after reportedly being delayed from an original 2026 launch window,” according to Notebookcheck.
That timeline held for roughly a week. On September 21, 2026, kopite7kimi, a leaker with a long track record on Nvidia’s internal codenames and die names, posted a correction to his own earlier forecast. As VideoCardz reported, kopite7kimi wrote: “Unfortunately, GR20x is a product from 2028. That means they have postponed again.” GR20x is the internal designation tied to the Rubin gaming die family that would power the RTX 6090 and the rest of the RTX 60 lineup. The same leaker, quoted separately by PC Guide, put it more bluntly: “RTX 60 is not coming next year.” A second leaker, Kepler_L2, corroborated the shift independently, telling outlets that both Nvidia’s Rubin gaming cards and AMD’s RDNA 5 lineup were sliding to 2028. We covered that specific corroboration in detail in our Second Leaker Backs RTX 6000 Delay to 2028 report, so this piece won’t re-tread that ground.
What matters for the discussion below is the magnitude of the swing: in the space of about a week, the consensus leaker timeline for the RTX 6090 moved by a full year, from H1 2027 to sometime in 2028. That later window has only hardened since: Baike reported in October 2026 that additional coverage continues to push the release toward 2028, and Traictory, also reporting in October 2026, put it more specifically, noting that consumer Rubin GPUs may not arrive any earlier than 2028. Nvidia itself has not commented on either timeline, and a report dated October 3, 2026 claimed the RTX 6090 could be delayed until 2028, citing alleged pressure from data-center demand, GDDR7 costs, and advanced-wafer allocation. As multiple outlets have noted, there is no official roadmap for RTX 60 or Rubin gaming silicon; everything currently circulating is sourced to supply-chain leakers and board-partner chatter, not a GTC keynote or an investor call.
Why the RTX 6090 release date keeps sliding
The explanation offered across the leak chain isn’t a yield problem or a design flaw. It’s an allocation problem. Nvidia’s Rubin architecture is not a single product line; it spans both the data-center AI accelerators that generate the bulk of Nvidia’s revenue today and the gaming-focused Rubin die (GR20x) that would eventually become the RTX 6090. Both draw from the same pool of advanced-node wafer capacity at TSMC, and reporting on the roadmap shift has consistently pointed to Nvidia prioritizing AI accelerator volume over gaming silicon capacity as the reason the RTX 60 series has moved twice in the space of a year, first from an original 2026 target to 2027, and now, per the latest leaks, into 2028.
That prioritization is not a mystery given Nvidia’s current business mix. Data-center demand for AI accelerators has outpaced gaming GPU demand by a wide margin over the past two years, and every wafer allocated to a gaming die is a wafer not available for a data-center part carrying a dramatically higher margin per unit. For a company managing genuinely scarce advanced-node capacity, delaying a consumer product by a year to keep AI accelerator supply flowing is a defensible, if unpopular, decision. It’s also not unique to Nvidia. The same reporting that pushed RTX 60 to 2028 describes AMD facing a nearly identical squeeze with its RDNA 5 gaming lineup, with only a single die, codenamed AT2, still tracking for a 2027 release.
How unusual is a three-year GPU generation gap?
Nvidia’s GeForce lineup has followed a roughly two-year cadence for most of the last decade. The GTX 10 series (Pascal) launched in 2016. The RTX 20 series (Turing) followed in 2018. RTX 30 (Ampere) arrived in 2020, RTX 40 (Ada Lovelace) in 2022, and RTX 50 (Blackwell), including the RTX 5090, launched in January 2025 (and the RTX 5090 is still the newest GeForce desktop flagship on sale, with no RTX 6090 announced), itself already a few months later than the historical two-year rhythm would have suggested. If the RTX 60 series, and the RTX 6090 within it, now lands in 2028 as the latest leaks suggest, the gap between RTX 50 and RTX 60 would stretch to roughly three years, the longest gap between mainstream GeForce generations in the RTX era.
| GeForce generation | Architecture | Approximate launch | Gap from prior gen |
|---|---|---|---|
| GTX 10 series | Pascal | 2016 | – |
| RTX 20 series | Turing | 2018 | ~2 years |
| RTX 30 series | Ampere | 2020 | ~2 years |
| RTX 40 series | Ada Lovelace | 2022 | ~2 years |
| RTX 50 series | Blackwell | January 2025 | ~2.3 years |
| RTX 60 series (RTX 6090) | Rubin | Reportedly 2028 | ~3 years (reported) |
A three-year wait changes the calculus for anyone budgeting a GPU upgrade around a “next generation is coming soon” assumption. It also means the RTX 5090, already Nvidia’s most expensive mainstream gaming card at launch, is likely to remain the top consumer SKU for far longer than any prior flagship in the Ampere-to-Blackwell stretch. We’ve tracked what that extended shelf life has already done to RTX 5090 street pricing in our report on RTX 5090 pricing climbing well past its original MSRP as the delay narrative firmed up.
What this means for the RTX 6090 specifically
The RTX 6090 sits at the top of whatever the RTX 60 stack ends up looking like, the direct successor to the RTX 5090 and, before that, the RTX 4090. Nvidia has not confirmed the RTX 6090 name, its specifications, or any pricing, and no leak in the current chain includes verified specs for the card. What the delay does change is the RTX 6090’s competitive position at launch, whenever that turns out to be. A GR20x-based flagship arriving in 2028 would be facing off against a far more mature AI PC and handheld ecosystem, a TSMC process node that has had additional years to mature, and a base of RTX 5090 owners who, by then, will have owned their card for three years rather than the usual two. That’s a longer hold period than Nvidia has asked enthusiast buyers to sit through in the RTX era, and it raises the bar for how large a generational leap the RTX 6090 needs to deliver to justify the wait.
It also means anyone hunting for concrete RTX 6090 specs, memory configuration, or pricing this year is going to come up empty. The numbers that do circulate are purely speculative: a Komputronik roundup from August 2026, for instance, floated figures of roughly 24,576 CUDA cores, 32GB of GDDR7 memory, a TSMC 3-nanometer manufacturing process, and pricing exceeding $2,000, but every one of those figures is extrapolated from the RTX 5090’s specs rather than a leaked spec sheet. A May 2026 report from Szwecent likewise described rumored RTX 6090 specs of 32GB GDDR7 and 192 SMs, plus a rumored 512-bit memory bus and 1.79TB/s of bandwidth, and it too remains unconfirmed by Nvidia. Readers should treat any RTX 6090 memory, core count, process node, or price figure circulating before an official Nvidia roadmap as unverified.
Leaker and outlet claims, side by side
The timeline shift is easier to follow laid out chronologically. Here’s how the RTX 60 series / RTX 6090 launch window has moved across the leak chain over the past several months:
| Source | Claimed RTX 60 / RTX 6090 window | Reported by | Date |
|---|---|---|---|
| Moore’s Law Is Dead | H1 2027 (originally aimed at 2026) | TweakTown, Notebookcheck | Sept. 14-16, 2026 |
| kopite7kimi (earlier estimate) | 2H 2027 | VideoCardz | Prior to Sept. 21, 2026 |
| kopite7kimi (revised) | 2028 (“GR20x is a product from 2028”) | VideoCardz | Sept. 21, 2026 |
| Kepler_L2 | Not 2027; postponed to 2028 | Reported alongside kopite7kimi’s revision | Sept. 2026 |
| Nvidia (official) | No public roadmap or date | – | As of Sept. 22, 2026 |
The pattern across the table is consistent: every revision has moved the date later, never earlier, and Nvidia has offered no counter-statement in either direction. That asymmetry, dates only sliding out, is itself a signal worth weighing when deciding how much confidence to put in any single leaker’s window.
The memory market makes the timing worse
The RTX 6090 delay isn’t landing in a vacuum. It arrives on top of a DRAM and GDDR memory shortage that has already pushed component and finished-card prices well above their launch levels across the current RTX 50 generation, a dynamic we detailed in GPU Relief Pushed to 2028 as DRAM Jumps a Record 95%. A longer gap before the next generation means the current memory-constrained pricing environment for RTX 5090 and RTX 4090 cards has more time to persist rather than being reset by fresh supply tied to a new launch. For buyers, that combination, an extended wait plus a tight memory market, is the least favorable pairing the GPU market has seen in years.
AI accelerators are absorbing the capacity gamers used to get
The RTX 6090’s delay is best understood alongside what Nvidia is building instead. The company’s data-center Rubin accelerators, the AI-focused sibling to the gaming die powering the RTX 6090, are reportedly the priority recipient of the wafer capacity gaming Rubin isn’t getting. We’ve covered how that AI-first allocation shows up elsewhere in Nvidia’s lineup, including the outsized resale value of data-center cards like the B200, which we reported on in Nvidia B200’s 58% Premium Fuels GPU Loan Boom. When a used AI accelerator can resell for a premium over its original price because demand so badly outstrips supply, it’s not hard to see why a company managing finite wafer starts would rather ship more of those chips and fewer gaming dies, even at the cost of delaying a marquee consumer product like the RTX 6090.
How AMD’s roadmap compares
Nvidia isn’t alone in this squeeze. The reporting that surfaced Kepler_L2’s claim about the RTX 60 series also described AMD’s RDNA 5 lineup sliding into the same 2028 window, with the exception of a single die, reportedly codenamed AT2, that remains on track for 2027. That parallel matters because it undercuts any narrative that this is an Nvidia-specific execution problem. If both of the two companies that actually build consumer discrete GPUs are pushing their next architectures out by roughly the same amount, at roughly the same time, the more plausible explanation is a shared, structural one: TSMC advanced-node capacity and HBM/GDDR memory supply are both bottlenecks the whole industry is fighting over, not just Nvidia’s own scheduling. Our comparison of the current professional-grade Blackwell, RTX 5090, and Radeon Pro lineups gives a sense of how tightly matched, and how supply-constrained, the current generation already is heading into this extended gap.
What Nvidia has, and hasn’t, confirmed
It’s worth being precise about what’s actually established here versus what’s still leaker chatter. As of October 2026, no official RTX 6090 exists, as Traictory’s ongoing coverage of the rumor mill makes clear, and Cloudspress reported that Nvidia had still announced no RTX 60-series model of any kind as of August 2026. Nvidia has not confirmed an RTX 6090 name, a 2027 window, a 2028 window, specifications, or pricing. Every date discussed in this piece, and in the broader leak chain it draws on, traces back to individuals with track records on Nvidia’s internal roadmap, not to an Nvidia press release, investor call, or GTC keynote. That distinction matters because leaker accuracy varies by claim type. Codename and die identification (what kopite7kimi is generally credited with getting right) is a different category of claim than calendar-year launch timing, which depends on decisions, like capacity allocation, that can and do change close to the date. Readers should treat the 2028 window as the current best-supported leaker consensus, not a confirmed release date.
What buyers should actually do right now
If the 2028 timeline holds, anyone currently deciding between buying an RTX 5090 now or waiting for the next flagship is facing a genuinely different decision than they would have been six months ago, when a 2027 launch still looked plausible. Waiting a year for a next-gen card is a reasonable bet. Waiting two to three years, while paying elevated prices for a current-generation card in a tight memory market, is a much harder case to make for most buyers. For gamers and creators who need a high-end GPU in the next 12 months, the practical read from this week’s reporting is that the current RTX 50 and RX 9000 generations are the products you’ll actually be living with for the foreseeable future, not a stopgap before an imminent successor. That doesn’t mean prices will stay where they are; it means the case for holding out for RTX 6090 specifically has weakened considerably.
Historical context: Nvidia has stretched cadences before
This wouldn’t be the first time Nvidia’s generational cadence has bent under external pressure. The GTX 10 to RTX 20 transition in 2018 came with a significant price jump tied to the addition of ray-tracing hardware, and some of the RTX 30-to-40 gap was shaped by the crypto-mining boom-and-bust cycle that distorted GPU availability and pricing through 2021 and 2022. What’s different this time is the cause: a mining boom is a demand-side shock that eventually fades, while AI accelerator demand pulling capacity away from gaming silicon is a structural shift in what TSMC’s advanced nodes are being built for, one that shows no clear sign of reversing before 2028.
Predictions: where the RTX 6090 story goes from here
Based on the current leak trajectory and the underlying capacity dynamics, here’s how this is likely to play out over the next year:
- The 2028 window will likely firm up further rather than reverse, since every revision in the leak chain to date has pushed the date later, never earlier.
- RTX 5090 pricing is likely to stay elevated well above its original MSRP through 2027, given the combination of a longer hold period and an ongoing DRAM/GDDR shortage.
- Nvidia will probably use 2027 to expand its data-center Rubin AI accelerator lineup and possibly introduce a mid-cycle refresh of the RTX 50 series (a “Super” or “Ti” style update) to keep consumer revenue flowing without touching scarce advanced-node capacity earmarked for AI chips.
- AMD’s RDNA 5 delay will likely track Nvidia’s roughly in parallel, keeping the competitive gap between the two vendors’ consumer lineups roughly where it is today rather than letting either pull decisively ahead.
- Expect continued leaker disagreement and revision through 2027, with the RTX 6090’s exact launch quarter unlikely to firm up until Nvidia itself confirms a roadmap, which historically happens roughly six to nine months ahead of an actual launch.
The bigger picture: gaming silicon is now a secondary priority
Strip away the leaker drama and the RTX 6090 story is really a story about where Nvidia’s manufacturing capacity is going. For most of the RTX era, gaming GPUs were the company’s flagship product and its primary revenue driver. That’s no longer true. Data-center AI accelerators now generate the majority of Nvidia’s revenue, and every leak in this chain points to the same underlying decision: when advanced-node wafers are scarce, Nvidia is choosing to protect AI accelerator supply over its two-year GeForce cadence. The RTX 6090 isn’t being cancelled or redesigned; it’s being deprioritized. That’s a meaningful shift in how Nvidia’s flagship consumer product fits into the company’s broader business, and it’s likely to shape how the next several GeForce generations are scheduled, not just this one.
Frequently asked questions
What is the current RTX 6090 release date?
There is no confirmed release date. As of September 22, 2026, the most recent leaker reporting, from kopite7kimi and Kepler_L2, points to a 2028 launch, revised from an earlier first-half-2027 estimate tied to Moore’s Law Is Dead. Nvidia has not issued an official roadmap or date.
Why did the RTX 6090 timeline move from 2027 to 2028?
Reporting attributes the shift to TSMC advanced-node capacity being redirected toward Nvidia’s data-center AI accelerators rather than a design or yield issue with the gaming-focused Rubin die.
Is the RTX 6090 name officially confirmed?
No. Nvidia has not confirmed the RTX 6090 name, specifications, or pricing. It’s the widely used shorthand for the expected flagship of the Rubin-based RTX 60 series, following the RTX 5090 naming pattern.
Should I buy an RTX 5090 now if the RTX 6090 is delayed to 2028?
If you need a high-end GPU within the next year, the extended delay makes waiting for the RTX 6090 a much longer bet than it looked six months ago. Prices for current-generation cards may stay elevated due to an ongoing memory shortage, but a multi-year wait is a harder case to justify than the shorter delay previously expected.
Is AMD’s next GPU generation also delayed?
Reporting on Kepler_L2’s leak indicates most of AMD’s RDNA 5 lineup is tracking toward the same 2028 window, with one die, reportedly codenamed AT2, still expected in 2027.
How reliable is kopite7kimi as a source?
Kopite7kimi is widely described by hardware outlets, including Notebookcheck and VideoCardz, as one of the more consistently accurate Nvidia leakers, particularly on codenames and die identification. Calendar-year launch timing is a separate category of claim that can shift with Nvidia’s internal capacity decisions, which is exactly what happened between the leaker’s earlier 2027 estimate and the revised 2028 one.
Will Nvidia release a mid-cycle refresh before the RTX 6090?
Nothing has been confirmed, but a longer-than-usual gap between RTX 50 and RTX 60 increases the odds Nvidia introduces some form of mid-generation update to keep the current lineup competitive, based on the pattern of past extended-cadence periods.
