NVIDIA has not announced a GeForce RTX 60. That sentence is the load-bearing one in this entire article, and it is easy to lose track of while reading the timeline: the RTX 60 is reportedly built on Rubin, Rubin is a real, shipping, in-production NVIDIA platform, and those two facts are regularly collapsed into “the RTX 60 launches in 2028.” They are not the same fact.
So this is not a review and not a launch piece. It is a timing decision. If you are holding money for the next generation, the only question that matters is whether your current card still does the job — and the answer to that question does not change because a leaker moved a date.
What this covers: what NVIDIA has actually said about Rubin, what the leakers say about an RTX 60 date, and a table for deciding whether to buy now or wait. The gap in one line: press reporting moved the RTX 60 from the second half of 2027 to 2028 — and NVIDIA has said nothing about it either way. Who should care: anyone on an RTX 50 or older card who is putting off a purchase for a GPUGPU. Graphics Processing Unit — the chip that renders the game's visuals; the main driver of framerate and image quality. that has no confirmed price, spec sheet or date.
What NVIDIA has actually said
Start with the official record, because it is much narrower than the rumor cycle implies.
NVIDIA unveiled the Rubin platform at CES in January 2026 as a data-center product line — six chips (later seven) assembled into the Vera Rubin NVL72 rack and the HGX Rubin NVL8 system. The company’s March 16, 2026 announcement says the platform is “in full production” and that “Vera Rubin-based products will be available from partners starting the second half of this year.” Every named customer in that release is a cloud provider, a system builder or an AI lab: AWS, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure, CoreWeave, Dell, HPE, Lenovo, Supermicro. There is no GeForce product, no consumer price and no consumer date anywhere in it.
The one time NVIDIA was asked directly about the RTX 60 rumor, it declined to engage. Tom’s Hardware reported that the company responded to a leaker’s 2028 claim with a spokesperson statement that it “doesn’t comment on rumors or unannounced products.”
What NVIDIA has said about consumer graphics is a direction, not a date. At CES 2026, Jensen Huang repeated that the future of gaming graphics is less focused on raw rasterization and more on neural rendering — which is Tom’s Hardware’s characterization of his remarks, not a product announcement.
For calibration, here is the generation you can actually buy, as NVIDIA announced it on January 6, 2025 and as it shipped:
| Card | Announced price | Availability (per NVIDIA) |
|---|---|---|
| GeForce RTX 5090 | $1,999 | January 30, 2025 |
| GeForce RTX 5080 | $999 | January 30, 2025 |
| GeForce RTX 5070 Ti | $749 | February 2025 |
| GeForce RTX 5070 | $549 | February 2025 |
Those are MSRPs from the launch release. What you pay today is a different number, and that difference is the second half of this story.
What the reporting says, and how thin it is
Everything about an RTX 60 date comes from leakers, and the two most-cited ones have now both landed on 2028 — after first landing somewhere else.
| Claim | Who said it | When | Standing |
|---|---|---|---|
| RTX 60 uses Rubin dies coded “GR20x”, launching in the second half of 2027 | kopite7kimi, reported by Tom’s Hardware | January 2026 | Superseded by the same leaker |
| “GR20x is a product from 2028. That means they have postponed again.” | kopite7kimi on X, reported by Tom’s Hardware, TechSpot, Club386 and others | September 21, 2026 | Current rumor |
| RTX 60 and most of AMD’s RDNA 5 cards arrive in 2028 | Kepler_L2 | September 2026 | Corroborates the 2028 window |
| RTX 6090 possibly in the first half of 2027 | Moore’s Law is Dead | September 2026 | Disputed by other leakers |
| AMD “AT2” chip (70 CUs) still on track for the second half of 2027 | Kepler_L2, via How-To Geek | September 2026 | Separate rumor about AMD |
None of this is confirmed. Wccftech, writing on the same day the 2028 post landed, put the caveat as plainly as anyone: most of the information circulating is “just a guess, and nothing concrete has been set so far,” and using words like “confirmed” or “delayed” is baseless without official backing. Kopite7kimi, for his part, said he would minimize updates until there is “definite news.”
There is also a useful piece of reporting discipline in that same cycle: Notebookcheck noted that the leaker Jukan called an early-2027 claim ridiculous and argued there was “no way Rubin gaming GPUs are coming out that soon.” When a rumor is this contested, the honest move is to plan for the uncertainty rather than pick a side.
Why every date keeps sliding
The mechanism behind the delay is the same mechanism behind a lot of PC pricing right now: memory. ThePCEnthusiast, summarizing the reporting around the RTX 50 Super refresh, put the arithmetic at roughly $60 to $70 for a 3GB GDDR7 module against around $20 for the 2GB module the current cards use. Across six or eight memory chips, that is hundreds of dollars of bill-of-materials cost on a card whose predecessor launched at $999.
That cost pressure is why the mid-cycle refresh you might have been waiting for is also in limbo — we covered the reported RTX 50 Super stack and its slipping window in our RTX 50 Super rumor breakdown. Tom’s Hardware notes the same refresh has been “cancelled or, at the very least, delayed due to supply chain shortages,” and that Huang has floated bringing back older-generation cards to cope with the DRAM shortage.
The street price of the current flagship tells you how tight things are. Tom’s Hardware reported the RTX 5090 selling “above $9,000 — if you’re lucky enough to find one listed online.” TechSpot’s version of the same reporting describes the card often sitting “between $7,000 and $9,000,” against a $1,999 launch MSRP.
Set against that, the length of this generation stops being a surprise. The RTX 50 series arrived in January 2025; a second-half-2028 RTX 60 would make the gap roughly three years, where How-To Geek notes the usual cadence runs “two to two and a half years.” TechSpot’s arithmetic is blunter: 41 to 47 months between generations, against an average gap of about 26 months.
The buy-or-wait table
This is the part you actually came for. Each row states its premise, because the recommendation is only as good as the assumption behind it.
| Your situation | What the evidence says | The call, and its premise |
|---|---|---|
| RTX 40 or 50 owner, happy at your resolution | No confirmed RTX 60 date, price or spec | Keep it. Premise: your card already hits the frame rates you want |
| RTX 20 or 30 owner, or an 8GB card that stutters now | VRAMVRAM. Video RAM — memory on the graphics card used for textures and frames; more of it lets you run higher settings. capacity is the binding constraint on new releases | Buy a 16GB card, not a flagship. Premise: the stutter is real and current, not projected |
| Need a GPU in the next one to three months | The 50 Super refresh is reported on hold | Buy current-generation at a real price. Premise: you have no working card |
| Top-end buyer with an open budget | RTX 5090 is far above its $1,999 launch MSRP | Wait. Premise: you are not being blocked from playing anything |
| Specifically waiting for an RTX 60 | The 2028 window is a rumor from two leakers | Stop planning around it. Premise: a rumor is not a release date |
| Buying for a game launching in 2027 | AMD’s rumored AT2 chip targets the second half of 2027 | Watch that window too. Premise: you are willing to consider Radeon |
The internal logic is worth spelling out, because it is the same in every row. A GPU purchase is a decision about the next two to three years of play, not about which silicon is newest. If your current card covers the games you play at the settings you accept, waiting costs you nothing but the option to buy sooner. If it does not, every month of waiting is a month of not playing the way you want — and the next generation, whenever it lands, is not guaranteed to be cheaper than the current one in a market where the flagship is selling at multiples of its MSRP.
What the leaks claim the cards will be
Treat all of this as rumor, and treat the source of it with more caution than the headline numbers suggest. HotHardware reported on a leak to YouTuber RedGamingTech describing three dies — GR202 for an RTX 6090, GR203 for an RTX 6080 and GR205 for an RTX 6070 — with 32GB, 20GB and 16GB of VRAM respectively. The same leak claims rasterization performance could drop by as much as 35% at similar frequencies, while path-traced performance would at minimum double. HotHardware’s own framing is the correct one: the source itself calls the path-tracing figure “absurdly high,” and the whole thing deserves “a huge grain of salt.”
If that trade-off is real, it changes what “wait for the 60 series” even means. A card that is faster in path-traced workloads and slower in plain rasterization is a different purchase decision than a straight upgrade, and it would depend heavily on how many of your games support the heavier rendering path. Do not build a 2028 plan on a spec sheet that has not been corroborated.
The read, without pretending certainty
Here is the honest summary. NVIDIA has confirmed a data-center Rubin platform shipping to partners from the second half of 2026. It has confirmed nothing about a consumer RTX 60, and it declined to comment on the rumor. Two leakers with strong track records say 2028; one disputed leak said early 2027; every version of the story is a rumor, and the same leaker who moved the date says he is done updating until there is definite news.
That is not enough to schedule a build around, and it is not nothing either. What you can do with it is cheap: if your card works, keep it and keep your money; if it does not, buy the capacity you need at today’s prices rather than paying a premium later for a card that may not arrive on the schedule anyone is currently quoting.