NVIDIA RTX 5090 and 5080: Next-Gen GPUs Rumored for CES 2025 Launch with Impressive Specs

Nvidia RTX 5090 rumor claims PCIe 5.0, DisplayPort 2.1 UHBR20, and only a single 12V-2x6 power connector

Now that preliminary specs for the RTX 5090 and RTX 5080 have leaked, including the GPU core counts and die models, other rumored specs for Nvidia's next-generation Blackwell RTX 50-series GPUs are flooding the internet. Benchlife reports that Nvidia's flagship RTX 5090 will come with PCIe 5.0 connectivity and a single power connector, as well as adding DisplayPort 2.1 UHBR20 outputs. The news claims that the RTX 5090 will have a single 12V-2x6 power connector, noting the GPU "can only" use a single connector. This could be some sort of limitation Nvidia is implementing, but we have seen overclocking-focused RTX 4090s with dual 16-pin connectors before. If the 5090 has a 600W TGP, there's little doubt in our minds that some AIB partner will release a factory overclocked halo model with dual power connectors at some point. The RTX 5090 will purportedly finally have DisplayPort 2.1a connectors -- Nvidia has been shipping DP1.4a outputs since the GTX 10-series launched in 2016, even though DisplayPort 2.0 was formally released by VESA in 2019. DP2.1 in 2022 didn't change much, just cleaning up things to align with USB4. In fact, AMD revised its RDNA 3 DP2.0 support to be DP2.1 compliant with the RX 7000-series parts. Nvidia will also reportedly support the full bandwidth UHBR20 spec -- so far, only AMD's Radeon Pro W7900 offers UHBR20 on a single output; the rest of the ports as well as other RDNA 3 GPUs use UHBR13.5. Elsewhere, it's claimed that the 5090 will have a 14-layer PCB to help with signal integrity. Air cooling will remain the primary cooling solution for the RTX 5090, despite its ultra-high 600W TDP. However, the media outlet suggested that "various companies" will also launch liquid cooling solutions -- which isn't much of a stretch, considering we've seen such GPUs for the past decade or so. If correct, these specs represent a couple of firsts for Nvidia. The RTX 50-series will be the first Nvidia client GPUs to be offer PCIe 5.0 capabilities (the data center H100/H200 have already had PCIe 5.0 support). That double the amount of external bandwidth for the GPUs, from 32GT/s (PCIe 4.0 x16) to 64GT/s (PCIe 5.0 x16). It's questionable how beneficial this will be in gaming scenarios, but it could be advantageous for certain non-gaming workloads that do more communication with the host system, as well as potentially in multi-GPU scenarios (like AI). This isn't the world's first GPU with PCIe 5.0, however, as the Nvidia Hopper H100 already had that in 2022 -- and even Moore Thread's GPU lineup also has PCIe 5.0 support. The RTX 50-series will also be the first Nvidia GPU lineup to sport a revision of DisplayPort beyond the now-ancient 1.4a version. DisplayPort 1.3 through 1.4a have offered up to 32.4 Gbps of bandwidth (25.92 Gbps with 8b/10b encoding for data) since 2014, with 1.4 adding DSC (Display Stream Compression) support in 2016. DP1.4a revised the DSC support from 1.2 to 1.2a in 2018, as earlier DSC revisions seemed to have had some issues. Now that Nvidia is finally moving to the newer DisplayPort standard, it's going for the highest bandwidth variant: UHBR20. That supports up to 20 Gbps per lane (up from 8.1 Gbps with version 1.3 through 1.4a) for 80 Gbps of total bandwidth. DP2.1 also uses 128b/132b encoding to improve the efficiency of transmission, so the maximum data bandwidth ends up at 75.37 Gbps (with 0.20 Gbps reserved for other purposes). That means the actual video data bandwidth is nearly triple that of the aging DP1.4a spec. With DSC, UHBR20 can even support up to 8K output at 240 Hz or 16K at 60 Hz -- or a theoretical 960 Hz for 4K video. Last but not least, Benchlife claims the RTX 5090 and RTX 5080 are going to be released at CES 2025. We have our doubts about Nvidia doing a CES launch, as that's not traditionally a PC hardware tradeshow where the company has major announcements. The RTX 40-series Super refresh was the exception in early 2024, but that was a mid-cycle update rather than a major new architecture. Still, CES has also become a major venue for AI discussions and hype, and there's little doubt Nvidia will be banging the AI drum for the RTX 50-series when it arrives.

Leak claims RTX 5090 has 600W TGP, RTX 5080 hits 400W -- up to 21,760 cores, 32GB VRAM, 512-bit bus

Preliminary specifications of Nvidia's GeForce RTX 5080 and GeForce RTX 5090 graphics cards have been published by @kopite7kimi, a reputable hardware leaker who tends to have access to accurate information about Nvidia's upcoming products. If the specifications are correct, then Nvidia's forthcoming GeForce RTX 5090 will be a monster with a 600W total graphics power rating (TGP). Many of these specs line up with previous leaks, just with a bit more detail in some areas. You can see our Nvidia Blackwell RTX 50-series GPUs guide for more details. Nvidia's GeForce RTX 5090 -- the flagship Blackwell GPU for desktop PCs -- is expected to be based on the GB202 graphics processor with 21,760 CUDA cores, mated with 32GB of GDDR7 memory using a 512-bit interface. That massive number of FP32 cores will consume enormous amount of power, up to 600W if the leak is accurate. The card will certainly rank among the best graphics cards when it's released, even if the final specifications end up being slightly less impressive. Just don't ask about pricing... By contrast, Nvidia's GeForce RTX 5080 is said to be powered by the GB203 GPU and will feature 10,752 CUDA cores, which is roughly half of the range-topping offering. RTX 5080 graphics cards are now projected to come with 16GB of GDDR7 memory with a 256-bit interface, with a TGP rating of 400W. With a decent bandwidth uplift enabled by GDDR7, the RTX 5080 should significantly outperform its predecessor in high resolutions and should be a potent graphics card. While the potentially massive performance of the GeForce RTX 5090 certainly draws attention, another thing that strikes the eye is the huge performance gulf between the flagship RTX 5090 and its smaller RTX 5080 sibling. It's almost exactly half of the range-topping graphics processor in terms of the number of stream processors and memory interfaces. The TGP will be two-thirds of the top-tier card, so clocks might be higher to try to narrow the gap, but this represents an even bigger difference than with the 40-series GPUs. That potential performance disparity looks strange to say the least, and suggests Nvidia may want to try and create a new tier of performance -- or perhaps limit the appeal of certain consumer cards as AI compute alternatives. The RTX 4090 on paper offers 68% more GPU cores, 50% more VRAM, 41% more memory bandwidth, and 13% more L2 cache than the RTX 4080. In practice, CPU limits hold the 4090 back at lower settings, but at 4K ultra it ended up being about 35% faster than the second tier 40-series GPU -- and the RTX 3090 was only about 15% faster than the RTX 3080. But these specs, if correct, suggest the 5090 could be up to twice the performance of the 4080. While we can't say for certain why Nvidia decided to build its next-generation lineup this way, one of the technical explanations could be that Nvidia's GB202 processor may consist of two GB203 dies. Using a multi-chiplet design for Blackwell GPUs has been rumored for a while and the GB100/GB200 datacenter GPUs indeed adopt this architecture. However, using CoWoS-L packaging to enable the high-speed (~10 TB/s) interconnect between dies for a consumer-grade product seems like a very expensive idea. An alternative to building a multi-chiplet GPU would be to build a monolithic graphics processor with over 21,760 CUDA cores at TSMC's 4nm-class process technology, which would result in a circa 650 mm^2 die. Such a design is hard to yield because of the large die size, which is why redundancies are usually present -- the 4090's AD102 chip for example has a maximum of 144 Streaming Multiprocessors (SMs), but only 128 are enabled. So it's not impossible for Nvidia to go that route if it wants to. A monolithic chip would also be very expensive, however, and it would be weird to have such a huge gap between the RTX 5080 and RTX 5090. Other GPUs could try to plug the holes, though, and we could eventually see lower tier parts that might have something like 18,000 functional CUDA cores. What we do know is that Nvidia uses the same chips for a variety of products: desktop, mobile, professional, and data center GPUs are all based on the same silicon designs. With AI being such a hot item right now, Nvidia might be creating a massive data center part as the first priority, and then productizing it as a consumer offering as well. If that's the case, don't be surprised if pricing ends up being quite a bit higher than the already exorbitantly priced RTX 4090 -- and we could even see AI variants arrive before the consumer models. For now, all the information we have about Blackwell-based graphics cards for client PCs is strictly unofficial. Apply the usual skepticism and know that, until Nvidia says something directly, things can and likely will change. There are still conflicting rumors on the release date as well, with some saying the RTX 50-series won't arrive until early 2025. If correct, that gives ample time for continued tweaking ahead of the launch. Until the official announcement, we can expect the rumor mill to stay busy churning out various theories and specifications.

NVIDIA GeForce RTX 5090 & RTX 5080 Graphics Cards To Debut At CES 2025: 14-Layer PCB, Single 12V-2x6 Power, PCIe 5.0 & DP 2.1a Support

NVIDIA GeForce RTX 5090 & RTX 5080 graphics cards are set for a CES 2025 introduction with brand-new features for gamers. Yesterday, we got our taste of the first specifications of the upcoming high-end options with the RTX 50 "Blackwell" GPU lineup, the GeForce RTX 5090, and the GeForce RTX 5080. These two cards will be the top of the line for gamers around the world when they debut and it looks like Benchlife has some more information giving us an idea of what to expect from the next-generation offerings. Starting with the launch, it looks like Benchlife states that the NVIDIA GeForce RTX 5090 & RTX 5080 are going to debut at CES 2025. This is a follow-up to our previous report and also aligns with what Kopite7kimi had stated earlier. But we also know that NVIDIA plans to launch a China-specific GPU variation of the RTX 5090 known as the RTX 5090D. If there are no surprises, GeForce RTX 5090 and GeForce RTX 5080 will be officially released at CES 2025; at the same time, we may also see the GeForce RTX 50 series for notebook computers. via Benchlife Since the Chinese New Year takes place on the 29th of January, that would be the most ideal timeframe to introduce/launch the graphics card. NVIDIA's GeForce RTX 50 "Laptop" GPUs are also said to make a showcase at CES 2025 with the OEM market anxiously anticipating the availability of new GPUs to power their latest AMD & Intel-powered PCs for gamers and enthusiasts. Considering that NVIDIA has reportedly finalized the design, it will take a few months before the products are ready to be shipped for retail and it makes sense to introduce the card at January's CES 2025. The current generation RTX 40 series was announced in September and launched in October with the RTX 4090 hitting retail shelves first followed by the RTX 4080 in November. The green team is also reportedly preparing the discontinuation process of its RTX 4090 & RTX 4090D GPUs next month to make room for the newer flagships. The NVIDIA GeForce RTX 50 series GPUs are also going to feature a brand new design which will encompass the PCB, Cooling solution, and power delivery. The GeForce RTX 5090 in particular is expected to utilize three individual PCBs, each housing different components of the chip. We've previously detailed what the PCB might look like and from what we know, the primary PCB that houses the GPU will consist of 16 GDDR7 memory modules, each with 3 GB of VRAM capacity & running at speeds ranging between 28-32 Gbps. This will provide the highest VRAM solution ever on a consumer-tier graphics card, totaling 32 GB across a 512-bit interface and reaching up to 2.0 TB/s of bandwidth. Other changes will include the new Founders Edition design which is said to utilize a compact dual-slot cooling solution for the RTX 5090. It is not clear if the RTX 5090 and RTX 5080 would share the same Founders Edition cooling as that was the case with the RTX 4090 and RTX 4080. But overall looks and presentation will be different from the RTX 30 and RTX 40 series. Also, there have been recent rumors that NVIDIA's GeForce RTX 5090 might end up with dual 12V-2x6 power connectors but that's not going to be the case. Talking to a few AIB partners ourselves, we were told that there might be certain enthusiast designs made for LN2 overclocking that come with dual 16-pin connectors but the Founders Edition and majority of AIB models will feature a single connector. There are already dual 16-pin RTX 4090 cards on the market and those are again geared towards enthusiast overclockers so it won't be much of a surprise as being made out. On the other hand, whether it is the 400W GeForce RTX 5080 or the 600W GeForce RTX 5090, they can only use a single 12V-2×6 connector. There are no two sets of 12v on the GeForce RTX 5090 as reported by some media. -2×6 connector. via Benchlife With that said, the 600W/400W TBP figures don't represent actual power consumption which will be lower than we have seen in previous generations. The RTX 4090 was rated at a much higher TBP of 450W than the RTX 3090 but it ended up consuming less power in games than the RTX 3090 Ti & the RTX 3090. Besides the launch and design of the next-gen GeForce RTX 50 series, NVIDIA is also planning to introduce new features within its Blackwell Gaming GPU lineup. Some of the features highlighted are support for PCIe 5.0 specifications. The current iteration of AMD, Intel, & NVIDIA GPUs are all PCIe 4.0 compliant with RTX 40 only featuring PCIe 5.0/5.1 CEM support through their power port interfaces. The only GPU that has PCIe 5.0 support through the slot is the Moore Threads MTT line which is confined to Chinese marketplaces and has very poor gaming performance. Our news mentioned that the GeForce RTX 5090 will use a 14-layer PCB and will also introduce the PCIe 5.0 specification and HBR20's DisplayPort 2.1a. via Benchlife NVIDIA's GeForce RTX 50 graphics cards will end up being the first to support the new standard, offering double the bandwidth and transfer speeds versus PCIe 4.0 and taking advantage of the latest motherboards from Intel and AMD which are housing PCIe 5.0 slots for a while now. We don't expect the next-gen lineup to fully saturate the PCIe 5.0 bandwidth or even PCIe 4.0 x16 but there can be certain workloads and applications, such as AI, that may take advantage of the extra BW. Lastly, the next-gen lineup is also expected to feature support for DisplayPort 2.1a (UHBR20). This will allow the NVIDIA GeForce RTX 50 graphics cards to be certified for compatibility with the latest DP80 displays, providing up to 80 Gbps of bandwidth or double that of HDMI 2.1 interfaces. With UHBR20, the GPUs will be able to support up to 16K displays at 60Hz with HDR and 10-bit color while offering insane refresh rates at lower resolutions. The launch of the NVIDIA GeForce RTX 50 graphics cards including the GeForce RTX 5090 & RTX 5080 will mark a new beginning for the PC gaming segment. While the competition has moved to compete in the mainstream and entry-level segments, NVIDIA will be seen as the one that caters to all segments from enthusiasts to budget users. Expect more information to roll out in the coming weeks.

Nvidia RTX 5080 and 5090 again rumored for CES 2025 launch - possibly alongside next-gen laptop GPUs

Nvidia's GeForce RTX 5090 and RTX 5080 will be introduced at CES 2025, we've again been told - alongside Blackwell notebook GPUs, possibly. The latest rumor on the next-gen GPUs from Team Green comes from the Chinese tech site Benchlife (add seasoning), and follows the detailed spec spillage about the incoming desktop graphics cards from Kopite7kimi we saw yesterday. Indeed, Benchlife doubles down on that previous leak, noting that they have received the same info aired by Kopite7kimi on X regarding the spec of the RTX 5090 and 5080 (including the latter getting just 16GB of video RAM). We're told that these GPUs will be officially revealed at CES 2025, and that at the same time, we could see Blackwell mobile GPUs to power up gaming laptops - but the site sounds more unsure about the latter. We also get a clarification on another couple of points, namely that the RTX 5090 will use a 14-layer PCB (printed circuit board) and will leverage PCIe 5.0. Furthermore, Benchlife pours cold water on a previous rumor that the RTX 5090 will use two 12V-2×6 connectors - one is going to be standard on both the RTX 5090 and 5080, the site asserts. It's good to hear that the 'double power connector as standard' rumor for the RTX 5090 is a false one - at least in theory. However, as Wccftech, which spotted the Benchlife article, elaborates, there may actually be premium Blackwell flagship graphics cards designed for overclocking enthusiasts that have two 12V-2×6 connectors. The point is that the vast majority of RTX 5090 models won't, though. As for the purported launch timeframe, the leaks are increasingly pointing to a CES 2025 debut, rather than a late 2024 reveal for the new Blackwell GPUs. All in all, we're inclined to believe this is the case - as there's hardly a rush for Nvidia to get the RTX 5090 and 5080 out. Team Green already owns the top-end of the graphics card market, and AMD isn't even going to challenge in this space with RDNA 4, which will top out at mid-range (if the many rumors about this are right). It's also not a surprise to launch somewhat later given that Nvidia would want to prioritize AI graphics cards - which make far more money than GeForce models - with the initial top-end Blackwell silicon rolling off the production lines. While the RTX 5090 appears to be shaping up nicely - for those who want a ridiculously powerful GPU, albeit guzzling a ridiculous amount of power, possibly, and likely ridiculously expensive to boot - the revelation that the RTX 5080 only has 16GB of VRAM (if the spec leak is right) did not go down at all well. There were cries aplenty of Nvidia 'cheaping out' on the video memory once more, as you might imagine - and given that the RTX 5080 is expected to be more powerful than the RTX 4090, it does feel like an underpowered VRAM configuration. (Particularly when on the other side of the GPU fence, AMD equips a current-gen mid-range graphics card, namely the 7800 XT, with 16GB). There were hints that Nvidia's looking at 16GB for the RTX 5080 going back quite some way, mind you, but we still can't leap to conclusions yet (though currently, plenty of folks are bounding across conclusion chasm like there's no tomorrow, it must be said).

Leaks suggest the RTX 5090 will pack 32GB VRAM, but it's the RTX 5080 I'm worried about

There's more to graphics cards than just memory, but if this generation has proven anything, lower VRAM can come with ramifications. It feels like only yesterday that Nvidia unannounced its 12GB 4080 card in response to the backlash, leaving just the original RTX 4080. In truth, even the 16GB model's reputation ended up in dispute, only slightly recovering thanks to the GeForce RTX 4080 Super and its enhancements. With that in mind, I do find it a bit weird that new leaks from Kopite7Kimi allude to the RTX 5080 coming with 16GB GDDR7 VRAM. Admittedly, I was distracted from that part by the fact the RTX 5090 allegedly uses 32GB, as that's the kind of generational jump that'll put smiles on enthusiast's faces. But, despite the fact we're talking about faster memory than GDDR6X, I think the 5080 is playing with fire. It's well worth noting that the reliable insider has shared more than just memory specs, as they claim the RTX 5090 will come armed with 21,760 CUDA cores, a 512-bit bus, and a monstrous 600W TDP. On the flip side, the RTX 5080 seemingly packs 10,752 cores, a 256-bit memory bus, and sits at 400W. On paper, the flagship feels like it's going to be much more of a leap than the 80-series card, and that in itself is intriguing. Before I start fleshing out my thoughts, it's worth once again gulping down the usual pinch of salt. While Kopite7kimi has been a reliable source, Nvidia likes to chop and change its next-gen GPU plans. That said, the green team might already have some generational plans up its sleeve that have led to the RTX 5080 sticking with 16GB, and on paper specs aren't necessarily an indication of performance. Let's break down the leaked specs - the apparent RTX 5090 is naturally a huge jump from the Nvidia GeForce RTX 4090, but it doesn't use a "full" GB202-300. Instead, the chip has been reduced by around 13%, leaving room for a potential Super card with an enhanced core count. The RTX 5080 appears to be in the same position with its GB203-400 GPU, but the fact its specs are halved means that there's absolutely room for a model with a wider memory bus. In turn, this could result in Nvidia releasing an RTX 5080 Super down the road with more VRAM, providing an opportunity for a mid-generation refresh. The TLDR of all of this boils down to the RTX 5080 appearing pretty underwhelming compared to the RTX 5090. I'm not personally convinced that the 90-class card won't show up in place of an ultra-enthusiast TITAN card this time around, which may have huge implications in terms of price. Ultimately, my prediction is that Nvidia will pitch the RTX 5080 as a gaming component while the flagship will cater to AI usage. But hey, that's just a guess, and we won't know anything for certain until Nvidia reveals its next-gen GPU plans.

I'm trembling at the potential power of a 21,760-core RTX 5090. And the price. Actually mostly the price

Nvidia will announce the next generation of GeForce graphics card at some point. This we know. What we can't say for sure is just what that next generation will look like. We're not completely in the dark, however. Recent leaks claiming insight into Nvidia's mammoth machinations suggest we're eyeing up a graphics card of herculean proportions. On the word of a reliable leaker over on X, kopite7kimi. They claim the card will offer 32 GB of GDDR7 memory, across a 512-bit memory bus. Once up-and-running, said card will gobble up 600 watts -- right at the maximum output available from a single 12v2x6 connector and PCIe connection. Moreover, the reported RTX 5090 will consist of 21,760 CUDA cores -- that's 33% more cores than the RTX 4090. The GB202-300-A1 would need to be absolutely massive to house that many CUDA cores, too. If you scaled up the AD102 GPU found inside the RTX 4090 to the size of Nvidia's best datacentre chip, Hopper's GH100, you'd probably fit just enough SMs inside it to make the RTX 5090. That makes for a die size around 814mm2, or conveniently just over 33% larger than the AD102 at 609mm2. Clearly, this proposed graphics card would make mincemeat of 4K. It might even lend credence to Nvidia's claims of 8K gaming, too. Though my initial reaction is split between feelings of awe, wonder, excitement, and some degree of fear. Is the RTX 5090 going to be the most expensive gaming graphics card yet? The initial leak has led many onlookers to expect such a thing, including KitGuru's Leo Waldock asking whether we'd get any change from our $2,500 down payment. Thankfully, Leo, I dare say that feels too much, even for this monster card. Let's break it down into simple mathematics. The RTX 4090 started out at $1,599 in the US, though it was relatively more expensive in most of Europe. We'll take the US MSRP for ease, anyways. With 33% more cores, lending to a presumed 33% or more fps gain in games, the RTX 5090 is already coming close to $2,149. Perhaps $2,099 on a good day. Would Nvidia go as simple as matching the price to the performance like that? Nvidia has often priced its graphics cards more or less directly in line with their performance uplift, even to the point of some frustration with the RTX 40-series. That said, the RTX 4090 was a much better deal than most. Expensive, yes, but a good deal for what you got. The RTX 4090 trounced the RTX 3090 Ti. I mean, really trounced it. And when the price tag was a mere $1,599, that's actually a significant price reduction versus the RTX 3090 Ti at $1,999. This tells us two things: firstly, that Nvidia doesn't mind pricing a graphics card at a whopping $2,000; secondly, that Nvidia isn't necessarily pricing graphics cards in tandem with performance. So, maybe we're actually looking at a $1,899/$1,999 card in the RTX 5090. That'd be a pretty good scenario, and anything less than that would be a brilliant one. What gives me serious hesitation, however, is just how expensive the RTX 4090 is right now. You can chalk up the RTX 4090's inflated price tag to its AI ability, ban in China, or assumed end of life status, but it's never really been a card that's sold at its MSRP. You'd have been lucky to pay just $1,599 for one -- they've been upwards of $1,650, if not $2,000, at frequent times throughout its two-year lifespan. So, the card that's potentially two thirds of what this proposed RTX 5090 can manage is already selling for $2,000, as the market demands, and Nvidia's previous history suggests it's not a company that likes being cut out of its own profit. If the demand is there for extreme-high-end graphics cards such as the RTX 4090, and to such an extent for board partners and retailers to happily charge upwards of $2,000 for one, that's going to give any business-savvy product manager at Nvidia a bright idea: 'If they can charge $2,000 for the RTX 4090, why can't we charge $2,200 for the RTX 5090?' We saw this sort of thinking back during the cryptocurrency mining craze, which is what seemingly led to such extremely high, but unsustainable, prices for cards such as the RTX 3090 Ti and later the unlaunched RTX 4080 12GB. I remember saying back then I sure hope these inflated prices don't act as an excuse to keep prices high once the craze is over, and they did, for a bit. But then Nvidia saw sense and realised it had to bring its prices within reason for us mere mortals. But the point stands: the roof has been blown off graphics card prices in the past half decade, and we're slightly at the mercy of the market to tell us what to spend on these cards. The super-duper overclocked Asus Strix or equivalent RTX 5090 is sure going to sail past $2,000 whether that's the MSRP for the Founders Edition or not. For now, we don't know what Nvidia would like to price the RTX 5090 at, nor will we find out for quite some time. The RTX 5090 is currently rumoured to be announced at CES 2025 and planned prices can, and very likely will, change between now and then.

Nvidia may release the RTX 5080 in 24GB and 16GB flavors -- the higher VRAM capacity will come in the future via 3GB GDDR7 chips

RTX 5080 could come with two different VRAM capacities thanks to GDDR7. A shipping manifest has cropped up featuring different SKUs for Nvidia's upcoming Blackwell RTX 50-series GPUs. Discovered by harukaze5719 on X, the manifest shows seven variants, meaning we could see multiple iterations of the RTX 5090 and RTX 5080 at the very least, possibly with varying VRAM capacities. The manifest showcases six GPU SKUs aimed at "development and testing purposes." All seven feature varying codenames: 699-1G144-0030-TS1, 699-1G144-0050-TS1, 699-1G144-0045-TS1, 691-1G145-2030-TS1, 699-1G147-0070-TS1, 699-1G147-0050-TS1, 699-1G147-0045-EB1. It's impossible to tell what these models are specifically, but given Nvidia's past track history, it usually launches its highest-tier models first, those being the RTX 5090 and RTX 5080. As a result, these seven models could be multiple iterations of the RTX 5090 and RTX 5080 featuring either different variants and/or different board versions, with possibly an RTX 5070/5070 Ti SKU or two thrown in for good measure. We expect at least two of the SKUs to potentially correlate to RTX 5080 24GB and 16GB flavors with the same bus width. To recap, preliminary specs for the RTX 5090 put it at 32GB of VRAM capacity, while the RTX 5080 will stick with 16GB. That would mean no 24GB card this generation, which has existed for the past three generations (Titan RTX, RTX 3090 / 3090 Ti, and RTX 4090). It would be odd at best for Nvidia to ignore the 24GB capacity, as the gap between the RTX 5080's purported 16GB capacity and the 5090's 32GB is simply too big to ignore. An AI/prosumer-focused RTX 5080 with 24GB of VRAM would be the perfect middle-ground for users who might not need or cannot afford the RTX 5090 but still want the workflow capabilities that 24GB of VRAM offers. Note that this isn't the same as what happened with the 'unlaunched' RTX 4080 12GB (which eventually became the RTX 4070 Ti). The 4080 12GB used a different die, AD104, with a 192-bit interface and fewer GPU processing cores. The RTX 5080 would likely have the same (or very similar) specs in both VRAM capacities, with only the amount of memory being different. All this comes thanks to the GDDR7 standard and its new capacity options. GDDR7 will be the first graphics memory solution to offer capacities beyond 2GB per IC, with potential of 3GB variants coming in the future -- and capacities up to 64Gb / 8GB could eventually become available. Higher capacities could prove hugely advantageous for next-gen GPUs, as it means manufacturers have more flexibility in choosing various memory capacities without being restricted by the bus width of a GPU's memory subsystem. With 3GB ICs, the RTX 5080 would be able to offer 24GB of VRAM without adding or reducing the memory chip count, enabling it to retain its 256-bit wide interface. Depending on how Nvidia wants to operate in the age of GDDR7, it could also play the same trick with the RTX 5090 and give it 48GB of memory with 3GB ICs. Of course, that's less likely as 48GB could put the 5090 into direct competition with Nvidia's future workstation graphics cards. More likely is that Nvidia will eventually offer a professional RTX GPU with 48GB on a 512-bit interface, and probably even 96GB by putting memory on both sides of the PCB. AI workload have proven to be voracious in their appetite for VRAM, after all.