Key takeaways
- Confirmed release date: None. Nvidia has not announced any GeForce RTX 60 product.
- Latest rumoured window: 2028, per kopite7kimi (September 21, 2026) and Kepler_L2 (September 16, 2026).
- Earlier rumoured window: First half of 2027, per a Moore’s Law Is Dead source (September 15, 2026); second half of 2027, per kopite7kimi’s January 2026 estimate.
- Architecture: Rubin. Confirmed by Nvidia for data-centre products; reported by kopite7kimi for gaming as the “GR20x” family.
- Rumoured flagship: RTX 6090 on a GR202 chip, with GR203, GR205, GR206 and GR207 also reported.
- Should you wait? No, unless you are happy waiting more than a year. Buy for your needs today.
Nvidia has not announced the GeForce RTX 60 series, so there is no official release date, price or specification. What is known for certain is that Nvidia’s next architecture is called Rubin and that its data-centre version, the Vera Rubin platform, was shown at CES 2026 and is now shipping to AI customers. For gaming cards, the latest leaks from kopite7kimi and Kepler_L2, both from September 2026, point to a 2028 launch, while Moore’s Law Is Dead’s source claims the first half of 2027. Everything about consumer RTX 60 specs remains rumour.
This page separates confirmed information from reports and rumours, and names the source for each claim. Leaks about unreleased GPUs change often, so treat any specific number below as provisional.
Verdict at a glance
- Confirmed release date: None. Nvidia has not announced any GeForce RTX 60 product.
- Latest rumoured window: 2028, per kopite7kimi (September 21, 2026) and Kepler_L2 (September 16, 2026).
- Earlier rumoured window: First half of 2027, per a Moore’s Law Is Dead source (September 15, 2026); second half of 2027, per kopite7kimi’s January 2026 estimate.
- Architecture: Rubin. Confirmed by Nvidia for data-centre products; reported by kopite7kimi for gaming as the “GR20x” family.
- Rumoured flagship: RTX 6090 on a GR202 chip, with GR203, GR205, GR206 and GR207 also reported.
- Should you wait? No, unless you are happy waiting more than a year. Buy for your needs today.
What is confirmed
| Fact | Status | Source |
|---|---|---|
| Nvidia’s architecture after Blackwell is called Rubin | Confirmed | Nvidia roadmap (GTC 2025) and CES 2026 keynote |
| Vera Rubin data-centre platform shown in full at CES 2026 | Confirmed | Nvidia CES 2026 keynote |
| Rubin data-centre GPU uses TSMC 3nm and HBM4 | Confirmed for data-centre parts | Nvidia platform details, reported widely |
| HBM4 suppliers Samsung, SK hynix and Micron qualified for Rubin | Confirmed | Jensen Huang, Seoul, June 2026 |
| Any GeForce RTX 60 product, spec, price or date | Not announced | No Nvidia announcement as of October 2026 |
| No new GeForce GPUs announced at CES 2026 | Confirmed | Nvidia’s CES 2026 statement, reported by VideoCardz |
What Nvidia has shown: Vera Rubin for data centres
At CES 2026, Nvidia presented the Vera Rubin platform, a set of six chips: the Rubin GPU, the Vera CPU, the NVLink 6 switch, the ConnectX-9 SuperNIC, the BlueField-4 DPU and the Spectrum-6 Ethernet switch. Widely reported specifications for the Rubin data-centre GPU include a TSMC 3nm multi-chip design with two compute dies, 336 billion transistors, 288GB of HBM4 and about 22 TB/s of memory bandwidth. The Vera CPU has 88 custom Arm cores. Nvidia says the Vera Rubin NVL72 rack combines 72 Rubin GPU packages with 36 Vera CPUs.
Sources disagree slightly on when “full production” was declared. Some say CES in January; others point to Computex-season statements around the start of June 2026. Either way, Rubin data-centre systems are shipping to cloud and AI customers in the second half of 2026.
Why this matters for gamers: GeForce cards normally share an architecture with Nvidia’s data-centre chips, but they are separate dies with different memory. The data-centre Rubin GPU uses HBM4; gaming cards have used GDDR memory. So these figures tell you the architecture is real and in production, not what an RTX 6090 will look like.
RTX 60 release date: what has been reported
| Date of report | Claimed launch | Source |
|---|---|---|
| 2025 | Late 2026 | Early industry speculation, repeated by several explainer sites |
| January 2026 | Second half of 2027 | kopite7kimi |
| Early 2026 | Pushed back due to the DRAM crisis | TweakTown and TechPowerUp rumour coverage |
| September 15, 2026 | First half of 2027 (2027 release “approved months ago”) | Moore’s Law Is Dead, citing a source inside Nvidia |
| September 16, 2026 | 2028 (“everything else from both AMD and Nvidia is 2028”) | Kepler_L2, AnandTech forums |
| September 21, 2026 | 2028 (GR20x is “a product from 2028”; Nvidia “postponed again”) | kopite7kimi |
| September 2026 | No next-generation information at board partners | BenchLife, checking with Nvidia’s partners |
The 2028 reports
The two most-followed leakers on GPU roadmaps both moved to 2028 in September 2026. Kepler_L2 wrote that apart from a few products, everything else from both AMD and Nvidia is 2028. Five days later, kopite7kimi, who had previously estimated the second half of 2027, said the GR20x gaming chips are now a 2028 product and that Nvidia had postponed them again. Wccftech argued that a 2028 date fits the current memory and pricing trajectory, saying the issue is cost rather than Nvidia’s ability to build the chips.
The 2027 report
Moore’s Law Is Dead, citing a source inside Nvidia, said a 2028 launch was wrong. According to that source, gaming Rubin was pushed back from 2026 to let AI products use as much TSMC capacity as possible, but a 2027 release had been approved months earlier, with the first half of 2027 as the target. This report came out a day before Kepler_L2’s and six days before kopite7kimi’s, so it is the outlier among the most recent claims.
Our read
A 2026 launch is effectively ruled out. Between the two camps, the 2028 claims are more recent and come from two leakers with long track records, so we consider 2028 the more likely window for now. That could change, and Nvidia has not commented.
Rumoured RTX 60 series specifications
Very little about consumer RTX 60 specs is solid. VideoCardz has described many detailed spec sheets as educated guesswork rather than information grounded in company guidance. Here is what has been reported, with how much weight each claim deserves.
| Item | Rumour | Source | Confidence |
|---|---|---|---|
| Chip family name | GR20x (GR202, GR203, GR205, GR206, GR207) | kopite7kimi, January 2026 | Moderate to high |
| Flagship | RTX 6090 on GR202 | kopite7kimi and wider reporting | Moderate |
| Process | TSMC 3nm-class, likely a custom variant | Multiple reports, consistent with data-centre Rubin | Moderate |
| Memory type | GDDR7, possibly faster speeds | Multiple reports | Moderate |
| RTX 6090 memory bus | 512-bit | Reports summarised by TechSpot and others | Low to moderate |
| RTX 6090 SM count | 192 SMs on the full die | Reports | Low |
| RTX 6090 memory size | 32GB (down from earlier 48GB rumours) | Reports | Low |
| Performance | About 30% to 35% more raster, about 2x path tracing versus RTX 5090 | RedGamingTech (single source) | Low |
| Price | Unknown | None credible | None |
Which numbers to ignore
Two widely shared figures are almost certainly wrong for the gaming card:
- About 28,000 CUDA cores or 256 SMs. According to kopite7kimi, these come from analysis of the Rubin CPX chip Nvidia showed, which he identifies as GR212 and not the gaming variant.
- 24,576 CUDA cores. This figure comes from a placeholder entry in TechPowerUp’s GPU database, not from a leak.
Claims of a 2nm process, 48GB of VRAM and an 800W power rating for the RTX 6090 come from a single outlier site and are not supported by any established leaker.
The RTX 6090 versus the RTX 5090 (rumoured)
The RTX 5090 has 170 SMs enabled on its GB202 chip, 32GB of GDDR7 on a 512-bit bus and 1,792 GB/s of bandwidth. If the rumoured 192-SM full die for GR202 is accurate, a full chip would have about 13% more SMs than the 5090’s enabled count, and shipping cards would likely have some disabled. Any real performance gain would also depend on clocks, architecture changes and memory speed. The RedGamingTech claim of 30% to 35% more raster performance and roughly double path-tracing performance is a single-source rumour.
Why the RTX 60 series keeps slipping
Reports point to economics rather than technical trouble.
- AI demand for TSMC capacity. The Moore’s Law Is Dead source said gaming Rubin was delayed to let AI products use as much TSMC capacity as possible.
- Memory shortage. Memory makers are prioritising HBM and server DRAM. TrendForce expects conventional DRAM contract prices to rise another 10% to 15% in Q4 2026 and the share of capacity for PC memory to shrink further in 2027.
- Cost of new nodes. Wccftech and others note high costs for leading-edge TSMC processes.
- Little pressure to launch. BenchLife found board partners had heard nothing about next-generation cards, and some felt a new generation was not needed yet. Kepler_L2’s comment suggests AMD’s next generation has also slipped to 2028, which reduces competitive pressure.
The RTX 50 Super refresh, which was expected to fill the gap, has also been delayed. Reports range from a CES 2027 reveal to outright cancellation.
What about AMD and Intel?
Kepler_L2’s September post suggested most of AMD’s next-generation products also land in 2028. AMD’s recent desktop launches have been extensions of its RDNA 4 lineup: the RX 9070 GRE went global in June 2026 and the entry-level RX 9050 launched in July 2026. Intel’s Arc B770 gaming card did not launch; Intel released the same chip as the Arc Pro B70 and B65 workstation cards in March 2026. So no competitor is currently forcing Nvidia’s hand.
Should you wait for the RTX 60 series?
Reasons to wait
- You already own a capable card, such as an RTX 4080, RTX 4090 or RX 7900 XTX, and are happy with it.
- You want the biggest possible generational jump and do not mind waiting one to two years.
Reasons not to wait
- The latest leaks put the RTX 60 series in 2028, which is a long time to go without an upgrade.
- There is no reason to expect lower prices at launch, given memory cost trends.
- Current cards are strong: the RTX 5080 delivers 128.3 fps at 1440p Ultra and the RX 9070 XT 116.5 fps in Tom’s Hardware’s hierarchy.
- New flagship launches tend to be hard to buy at first, and early stock often sells above list price.
What to buy instead
For 1440p, the RX 9070 and RX 9070 XT are the value picks, and the RTX 5070 Ti is the choice for ray tracing. For 4K, the RTX 5080 is the sensible high-end option and the RTX 5090 remains the fastest card available. A card bought now will likely be supported with driver updates and features for years.
What features might the RTX 60 series bring?
Nvidia has not announced any GeForce RTX 60 features. Based on the direction of recent generations and Nvidia’s public statements about Rubin, it is reasonable to expect a few broad themes, but none of the following is confirmed.
- More AI hardware. Every recent GeForce generation has expanded Tensor core capability, and DLSS has grown from upscaling into frame generation and multi-frame generation with the RTX 50 series. The data-centre Rubin GPU emphasises very low-precision AI formats such as NVFP4, so improved AI throughput is a likely focus.
- Stronger ray and path tracing. The single-source RedGamingTech rumour of roughly double path-tracing performance fits this pattern, but is unverified.
- Faster memory. Reports suggest GDDR7 at higher speeds than the RTX 50 series uses. The memory market, not the GPU, may decide how much VRAM each tier gets.
- A more advanced process. Moving from Blackwell’s TSMC 4nm-class process to a 3nm-class node would typically improve efficiency, which could help keep power draw in check.
The catch: These are educated expectations, not leaks. Treat any article that presents specific RTX 60 features as fact with caution until Nvidia announces them.
RTX 60 timeline in context
Nvidia’s GeForce generations have historically arrived roughly every two years: the RTX 20 series in 2018, RTX 30 in 2020, RTX 40 in 2022 and RTX 50 in early 2025. On that rhythm, an RTX 60 launch around 2027 would have been expected. A 2028 launch would make the RTX 50 generation one of the longest-lived in GeForce history, especially if the RTX 50 Super refresh never appears.
| Generation | Architecture | Launch |
|---|---|---|
| RTX 20 series | Turing | 2018 |
| RTX 30 series | Ampere | 2020 |
| RTX 40 series | Ada Lovelace | 2022 |
| RTX 50 series | Blackwell | 2025 |
| RTX 60 series (rumoured) | Rubin | 2027 or 2028 (unconfirmed) |
For buyers, that matters in a positive way: a card bought in 2026 is likely to remain current-generation for longer than usual, with continued driver support and new DLSS or FSR features.
How to spot a reliable RTX 60 leak
- Check who said it first. Many articles repeat the same few leakers. Trace the claim back to kopite7kimi, Kepler_L2, Moore’s Law Is Dead or another named source.
- Watch for data-centre confusion. Specs for Rubin, Rubin CPX and Rubin Ultra data-centre parts are often misapplied to gaming cards.
- Be wary of precise numbers far from launch. Exact core counts and prices more than a year before launch are usually guesses or placeholders.
- Look for supply-chain signals. Board partner briefings, regulatory filings and shipping manifests usually appear a few months before a real launch.
Products Mentioned in This Guide
These are the specific current cards we suggest above as what to buy instead of waiting for the RTX 60 series.
ASUS Prime Radeon RX 9070 16GB OC Edition
ASUS Prime Radeon RX 9070 16GB GDDR6 OC Edition Gaming Graphics Card
Included for 'RTX 60 Series Release Date and Specs: What Is Known' as a representative home product in this roundup; selection reflects the product set returned for this topic.
View on AmazonOne of our two 1440p value picks for buyers who do not want to wait until 2028.
GIGABYTE Radeon RX 9070 XT Gaming OC 16G
GIGABYTE Radeon RX 9070 XT Gaming OC 16G Graphics Card, PCIe 5.0, 16GB GDDR6, GV-R9070XTGAMING OC-16GD Video Card
Included for 'RTX 60 Series Release Date and Specs: What Is Known' as a representative home product in this roundup; selection reflects the product set returned for this topic.
View on AmazonThe other 1440p value pick, delivering 116.5 fps at 1440p Ultra in Tom’s Hardware’s hierarchy.
GIGABYTE GeForce RTX 5070 Ti Gaming OC 16G
GIGABYTE GeForce RTX 5070 Ti Gaming OC 16G Graphics Card, 16GB 256-bit GDDR7, PCIe 5.0, WINDFORCE Cooling System, GV-N507TGAMING OC-16GD Video Card
Included for 'RTX 60 Series Release Date and Specs: What Is Known' because the listing adds cooling or gel-infused layers, which is the attribute this article compares.
View on AmazonOur choice for 1440p ray tracing among cards available today.
GIGABYTE GeForce RTX 5080 Gaming OC 16G
GIGABYTE GeForce RTX 5080 Gaming OC 16G Graphics Card, WINDFORCE Cooling System, 16GB 256-bit GDDR7, GV-N5080GAMING OC-16GD Video Card
Included for 'RTX 60 Series Release Date and Specs: What Is Known' because the listing adds cooling or gel-infused layers, which is the attribute this article compares.
View on AmazonThe sensible high-end option for 4K, delivering 128.3 fps at 1440p Ultra in Tom’s Hardware’s hierarchy.
ASUS TUF Gaming GeForce RTX 5090 32GB OC Edition
ASUS TUF Gaming GeForce RTX 5090 32GB GDDR7 OC Edition Gaming Graphics Card
Included for 'RTX 60 Series Release Date and Specs: What Is Known' as a representative home product in this roundup; selection reflects the product set returned for this topic.
View on AmazonThe fastest card available and the baseline any rumoured RTX 6090 will be measured against.
How we compared
We compiled this article from Nvidia’s public statements and keynotes, independent reporting and leaks from the sources named above. Confirmed facts are separated from reports and rumours throughout. We did not bench-test any hardware ourselves; no RTX 60 hardware has been released. Performance figures for current cards come from the Tom’s Hardware GPU Benchmarks Hierarchy. We will update this page when Nvidia announces the RTX 60 series.
Sources
- Nvidia: CES 2026 keynote and Vera Rubin platform announcements
- VideoCardz: GeForce RTX 60 reportedly uses GR20x Rubin GPUs; Nvidia confirms no new GPUs at CES 2026
- kopite7kimi: posts on GR20x naming (January 2026) and 2028 timing (September 21, 2026)
- Kepler_L2: AnandTech forums post (September 16, 2026)
- Moore’s Law Is Dead: report citing an Nvidia source (September 15, 2026)
- BenchLife: board partner checks (September 2026)
- Wccftech: The 2028 launch rumour for RTX 60 Rubin GPUs
- TweakTown: RTX 60 series release date rumours
- TechSpot: RTX 60 series leaks are everywhere, but Nvidia has not finalised the GPUs
- Igor’s Lab: GeForce RTX 60, Rubin, GR20x and the AI economy
- The PC Enthusiast: RTX 60 series 2028 delay coverage
- TrendForce: 4Q26 memory contract price forecast
- Tom’s Hardware: GPU Benchmarks Hierarchy 2026
Frequently Asked Questions
When is the RTX 60 series release date?
Nvidia has not announced one. The most recent leaks, from kopite7kimi and Kepler_L2 in September 2026, point to 2028. A Moore’s Law Is Dead source claims the first half of 2027.
Will the RTX 60 series come out in 2026?
Almost certainly not. No leaker or report currently points to a 2026 launch, and Nvidia has announced no new GeForce GPUs this year.
What architecture will the RTX 60 series use?
Rubin. Nvidia has confirmed Rubin for its data-centre products, and kopite7kimi reports the gaming chips will be called GR20x.
How much VRAM will the RTX 6090 have?
Unknown. Recent reports suggest 32GB of GDDR7 on a 512-bit bus, down from earlier 48GB rumours, but none of this is confirmed.
How fast will the RTX 6090 be?
There is no reliable figure. A single-source rumour from RedGamingTech claims about 30% to 35% more raster performance and roughly double path-tracing performance compared with the RTX 5090.
Is the 28,000 CUDA core RTX 6090 leak real?
According to kopite7kimi, that figure comes from the data-centre Rubin CPX chip, not the gaming GR202, so it should not be applied to the RTX 6090.
Why is the RTX 60 series delayed?
Reports blame AI demand for TSMC manufacturing capacity, the global memory shortage, high costs for new process nodes and limited competitive pressure.
Should I buy an RTX 50 series card now or wait for RTX 60?
If you need a GPU, buy now. With RTX 60 likely more than a year away and memory prices still rising, waiting is only sensible if your current card already meets your needs.
