The Nvidia RTX 4090 announcement underway as we speak, and we’re here to bring you all the latest as it happens, along with the TechRadar team’s in-depth analysis of what to expect over the next 24 hours.
In case you have no idea what we’re talking about, the Nvidia RTX 4090 is Team Green’s next flagship graphics card, the kind that goes into the best gaming PCs and best workstations for creative content. It’s likely to be the most powerful consumer-grade graphics card ever produced, and it will likely make the best graphics cards currently on the market pale in comparison.
And that’s quite a feat, given how powerful the best Nvidia GeForce graphics cards like the Nvidia GeForce RTX 3090 Ti and even the Nvidia GeForce RTX 3080 actually are.
Nvidia Ampere, as the last generation of graphics processing architecture is called, is no slouch, and we’ll tell just about anyone that the Nvidia GeForce RTX 3070 is really about as much as anyone really needs. Still, Nvidia’s next-generation graphics architecture, Nvidia Lovelace, is shaping up to be downright revolutionary in terms of its gen-on-gen performance.
And now, after much anticipation, it seems like the Lovelace launch is finally here. We can’t say for sure which cards will be announced tomorrow – or even whether any graphics cards are announced at all – but an opening floodgate of rumors, leaked pics, and online benchmarks all point to an announcement on September 20 of at least the RTX 4090, though we suspect that the RTX 4080 and maybe even the RTX 4070 will also make an appearance.
In the lead-up to the announcement, we’ll be giving our expert analysis about what we expect, what we want to see, and finally our live reactions to the announcement as it happens. So stick with the TechRadar Nvidia RTX 4090 live blog throughout the day for all the latest news and analysis.
John (He/Him) is the US Computing Editor here at TechRadar and he is also a programmer, gamer, activist, and Brooklyn College alum currently living in Brooklyn, NY.
Named by the CTA as a CES 2020 Media Trailblazer for his science and technology reporting, John specializes in all areas of computer science, including industry news, hardware reviews, PC gaming, as well as general science writing and the social impact of the tech industry.
You can find him online on Twitter at @thisdotjohn
Welcome to TechRadar’s Nvidia RTX 4090 announcement liveblog! I’m John Loeffler, TechRadar’s US Computing Editor, and I’m kicking things off here less than 16 hours away from what is expected to be one of the biggest announcements of the year for the computing team – and yes, even bigger than a new MacBook.
Later, I’ll be joined by Christian Guyton, UK Computing Editor, and Matthew Hanson, TechRadar’s managing editor of computing and home entertainment, to help cover the lead up to the announcement tomorrow at 8AM PDT / 11AM EDT / 4PM BDT.
There’s a lot to discuss in the leadup to the announcement, so lets dig right in with the very first thing: what we expect to see tomorrow.
Without a doubt, we are expecting to see the Nvidia RTX 4090 tomorrow at Nvidia’s event. First, it’s called “GeForce Beyond”, and since GeForce is Nvidia’s consumer graphics card branding, there’s pretty much no doubt that this will be a consumer products event as opposed to enterprise cloud computing systems or something similar.
Second, in a recent teaser for tomorrow’s event posted to Nvidia’s GeForce Twitter account, Nvidia subtly gave the game away when they shows a computer desk with a monitor displaying a GeForce background, and Ada Lovelace’s computer program for calculating Bernoulli numbers using Charles Babbage’s theoretical Analytical Engine on the desk.
Lovelace, of course, is Nvidia’s code name for its next generation of graphics hardware, so there isn’t much mystery around what will be unveiled tomorrow.
Now, when Nvidia Ampere was first shown off in 2020, it wasn’t for a consumer product, but rather for server and enterprise computing, so there is a precedent for showing off computer architecture before releasing it for a consumer product. But this was also early on in the Covid pandemic of 2020, and Nvidia was specifically highlighting how its new architecture was helping researchers fight the novel coronavirus. Given the circumstances, we don’t think that Nvidia is looking to show off the technology this time around without a product.
A huge part of this is both the product release cadence (it’s been two years since the RTX 3000-series launch), and a number of leaked photos of RTX 4090s with third-party manufacturers and partners, including Gigabyte, Zotac, and Lenovo. These cards are going into systems and retail packaging as we speak, so there’s no reason to think that Nvidia won’t be revealing at least the RTX 4090 tomorrow.
So what about the RTX 4080?
We’ve gotten a lot less info about the RTX 4080 in recent weeks, but there’s still a off chance that we’ll get an RTX 4080 announcement tomorrow as well. The best evidence of this is wholly speculation on my part, which I dived into recently, but the long and short of it is that leaked photos of the RTX 4080 in production look pretty authentic, and so it is likely right behind the RTX 4090 in terms of release.
Staggered releases aren’t anything new, obviously, and there’s every reason to think that Nvidia Lovelace will be the same. If we’re expecting the RTX 4090 in October, the RTX 4080 could show up in late October, early November, so the fact that the RTX 4080 would be in the early stages of rolling out production units tracks.
That could still get us an announcement tomorrow, though the release date might be a later, which is something we saw with Nvidia Ampere.
Now we’re going to talk about the RTX 4070. This is the mystery card. Will we see the RTX 4070? How far along is the RTX 4070? Will the RTX 4070 actually be as powerful as the RTX 3090?
The only card we know less about is the Nvidia GeForce RTX 4060, but that’s for another time. The Nvidia GeForce RTX 4070 is arguably the most exciting card to come from the Nvidia Lovelace line because it is the best graphics card that most people are going to be able to afford, especially with the ongoing cost of living crisises around the globe.
But, we know very little about this card, and considering that it was the last card of the initial launch lineup of the RTX 3000-series to hit the market, we could see something similar here, with the RTX 4070 being announced tomorrow but not actually released until late November at the earliest.
So Videocardz just “leaked” the Nvidia RTX 4090 , and there’s at least something going for it, namely the new font branding that I brought up in my RTX 4080/4090 piece. That said, that pic looks Photoshopped to heck and back.
Yeah, the more I look at it, this RTX 4090 “leak” is fake as all heck.
That is a horrible picture, with the sharpness of the card and the sharpness of Jensen Huang being radically out of step. If this is a legit picture, it is some whack multiverse nonsense that Nvidia is trying to make happen, which even as ridiculous as multiverse stuff being what it is, this is just too crappy to pass. This is a reddit Photoshop fails-level of crap.
Honestly, that RTX 4090 picture is so bad, I’m knocking Videocardz down a peg as a source for even posting this. That pic is awful.
Honestly, WTAF Videocardz?
WHERE ARE JENSEN’S FINGERS???
And, with all that, we’re going to take a brief hiatus while the Uk wakes up and TechRadar’s UK team takes over the live blog.
That said, I don’t want the moment to pass that I don’t mention that Queen Elizabeth II passed away two weeks ago. Clearly, I’m an American, so the Queen’s passing doesn’t much concern me, but a UK colleague told me today that the feeling was as if President Joe Biden went on the TV this morning and said that the US flag was going to be different from that moment forward, for better or worse.
Does it matter in any real way going forward? Not really, no. But it obviously does.
I wish all of you, wherever you are, a beautiful evening. Be kind to each other.
…and we’re back! Thanks to our US colleagues – the UK team will take it from here as we get closer and oh-so-closer to Nvidia’s ‘GeForce Beyond’ presentation at GTC today. Fake GPU leaks aside, Nvidia has let slip some minuscule teasers in the buildup to the RTX 4090 reveal event.
We’ve got a new design for the GeForce RTX logo, which is so minor a change you have to imagine that the graphic design staff at Nvidia aren’t being paid enough. The Nvidia logo has had its colors inverted from white to black, the ‘RTX’ is a bit smaller, and the angled line at the right-hand edge has been straightened to an upright vertical border. This apparently qualifies as news.
Perhaps more excitingly, Nvidia also teased an image of a Discord user (with an Ada Lovelace profile pic) asking ‘how much faster?’ and a sticky note on a computer monitor bearing the hastily-scribbled memo ‘T.I or Tie?’
Now, it seems unlikely that Nvidia is going to launch an RTX 4090 Ti straight out of the gate – this note is probably a humorous nod to the ongoing pronunciation debate surrounding the ‘Ti’ moniker – but we’re amped nonetheless. At present, leaks suggest that what we will see at the event is the RTX 4090 and two different versions of the RTX 4080, one with 12GB of VRAM and one with 16GB; the former of these two cards was previously suspected to be the RTX 4070.
Those Lenovo Legion RTX 4090 cards look pretty darn tasty, if I do say so myself. Sure, they’re ridiculously chunky, but it’s looking like every RTX 4000 GPU is going to be a monster. I’m a big fan of the clean brushed silver finish on these cards, and I really hope Lenovo plans to sell them standalone rather than just putting them in pre-built Legion desktops – more competition on the GPU scene is only a good thing.
If you want to tune into the Nvidia reveal presentation at 8AM PST / 11AM EST / 4PM BST, you can check out our ‘How To Watch’ guide right here.
If you’re worried about a repeat of the previous generation’s horrifying GPU shortages, don’t be. There were a lot of factors at play during the RTX 3000 graphics card shortage, most of which are no longer a major concern.
For starters, the launch of the RTX 4000 series won’t be so badly impacted by COVID-19. A global pandemic saw manufacturing and supply chain issues across the tech hardware industry, with Intel’s troubled Arc GPU series being particularly badly hit. RTX 3000 production was obviously stymied as a result, but businesses have adapted and COVID shouldn’t create such large hurdles this time around.
Crypto mining was also a huge problem, with many GPUs being snatched up by bots, miners, and scalpers. Conventional rigs for mining cryptocurrencies typically use multiple high-end graphics cards to maximize processing power, but since the huge crypto market crash earlier this year, miners have been selling off their GPUs in an attempt to cut their losses.
It’s also worth noting that Ethereum – one of the biggest crypto coins right now – recently switched to a new system called ‘Proof-of-Stake’, which makes mining with GPUs redundant. Our US Computing Editor John Loeffler goes into great detail about it here.
Lastly, Nvidia have returned to the comforting bosom of TSMC for the fabrication of the RTX 4000 series chips. Nvidia is a ‘fabless’ company, which means that it outsources the actual factory manufacturing of its GPU chips. Previously, it used Taiwan-based TSMC, but it switched to Samsung for producing the consumer RTX 3000 GPUs – a move that proved troublesome, as Samsung’s cards reportedly suffered from a higher rate of defects. Nvidia has gone back to TSMC for RTX 4000, which was likely a smart decision.
Where will the RTX 4090 fall in our ranking of the best graphics cards? Well, if we’re talking about raw performance, it’s doubtless going to be the best consumer GPU we’ve ever seen – but performance isn’t everything. According to a recent price leak from a Vietnamese tech retailer, the RTX 4090 could go for as much as $2,500 – an outrageously unobtainable price tag that would mark it as an enthusiasts-only GPU.
Still, that same leak also had models closer to the $2,000 range, and previous-generation pricing would see the RTX 4090 retail at $1,499 in the US; still expensive, but not quite as ludicrous. I’m expecting something between $1,500 and $2,000, personally. Price will be a big factor in review scores here – the GPU will no doubt be amazing, but too high a cost could see Nvidia slammed in the ratings arena.
We’re still speculating among ourselves as to what exactly will show up at Nvidia’s GTC presentation later today. Known leaker @kopite7kimi has stated that it’ll be the RTX 4090 and two flavors of the RTX 4080 (12GB and 16GB models, to be precise). VideoCardz.com and some other sites seem to agree with this analysis.
However, it’s only been a few days since another trusted (well, not trusted by Nvidia) leaker posted an image of a supposed RTX 4090 Ti card. Twitter user @greymon55 claims that the GPU pictured below is an engineering sample with the 4090 Ti name blurred out.
What do you think? You can join me on Twitter if you like speculative yelling and bad memes. I’m still desperately hoping that we’ll get an RTX 4070 too, but it’s not looking likely…
The cooler is actually labeled 4090Ti, which I have blurred because it involves confidential information. But according to Kopite, it is supposed to be a custom AD102 version, which specification is unclear. https://t.co/EJ760AkAXxSeptember 5, 2022
The ‘GeForce Beyond’ event is now less than three hours away, and I can feel the excitement. We’ve got a last-minute tweet from leaker @kopite7kimi, apparently updating a couple of previously-leaked specs for the 16GB model of the RTX 4080.
There’s nothing game-changing, which is hardly surprising this close to the reveal, but the memory speed listed is a little slower than previous leaks for the RTX 4080 suggested: 22.5Gbps as opposed to 23Gbps. It’s a minor change, but it brings the 16GB model closer to the 12GB model (and the RTX 4090) when it comes to memory speed.
Before the keynote, I want to update the spec of RTX 4080 16G. -22.5Gbps-start from 320WAfter so many months of waiting, these details have become unimportant. Enjoy.September 20, 2022
With hours to go before we get our first proper glimpse of Nvidia’s RTX 4000 series of GPUs, a new leak has shown that AMD could be providing some serious competition once its own next-gen graphics cards arrive.
The leak has posited that one of AMD’s RDNA 3 GPUs (presumably the flagship model) will be capable of reaching clock speeds close to a whopping 4GHz. Considering that the RTX 4090 is currently expected to have maximum boost speeds of around 2.8GHz – although this is based on leaks, not direct information from Nvidia – Team Green might need to start worrying about the competition.
In any case, it’s definitely possible that AMD will continue to have Nvidia beat when it comes to pure value for money. It’s unlikely that we’ll see an affordable RTX 4060 revealed today, so AMD still has time to corner the market when it comes to budget GPUs.
Almost 4Ghz GPU😱 pic.twitter.com/CC97YL9NovSeptember 19, 2022
We’re less than two hours away from Nvidia’s big RTX 4090 reveal presentation at GTC 2022, so let’s take a moment to look at what manufacturers we can expect to be making third-party models of the new RTX GPUs.
Gigabyte, MSI, Asus, and Zotac Gaming are all expected to release RTX 4000 cards once they officially hit the market, but EVGA – once a regular partner of Nvidia – has officially terminated its relationship with Team Green, citing “disrespectful treatment” as one of the main reasons. It’s unclear exactly what EVGA meant by this, but don’t expect to see any RTX 4090 cards with EVGA branding.
Naturally, GPU industry stalwart Sapphire Technology isn’t expected to produce any RTX 4000 GPUs either, since it has historically stuck to a long-running relationship with AMD to sell Radeon cards.
If you want to watch the Nvidia GTC event livestream but don’t know how, you can check out our ‘How To Watch’ page, courtesy of our lovely computing writer Allisa.
A little aside from Twitter user @Notanapplefan1 (hah) – a side-by-side comparison of the copper thermal plates that come attached to the RTX 4090, the RTX 3080, and the RTX 2080 Ti.
As ‘Not an apple fan !!’ correctly observes, the die size for the new GPU looks very small in comparison to the other two. The RTX 4090 heat sink pictured is supposedly from a leaked Lenovo Legion GPU we spotted a few days ago; the die slot is notably smaller than the previous-gen models. This is likely due to the new manufacturing process used by TSMC, which will cut the size down from 8nm to 5nm.
In other tweets, @Notanapplefan1 expresses their incredulity towards recent leaks claiming that the RTX 4090 will feature more than 75 billion transistors. I’ll admit that figure does seem high given the apparent size of the die; closer to 50 billion sounds more realistic. Either way, we’re looking at a significant step up from the 28.3 billion transistors found on the RTX 3090 GPU die.
rtx 4090 die size looks smallfirst picture is leaked rtx 4090 cold plate vs rtx 2080ti and rtx 3080 pic.twitter.com/gACiGzMtH8September 19, 2022
Less than one hour to go, folks! I’m running a cheeky poll to guess pricing. Everyone who guesses correctly wins, uh, nothing. I don’t have any money, and that’s why I’m annoyed that we’re probably not going to see the RTX 4070 or RTX 4060 today.
Any bets on RTX 4090 pricing?September 20, 2022
Half an hour left! Let’s take a proper look at the teaser image, while the YouTube livestream chat goes progressively more insane. We already discussed the ‘T.I or Tie’ post-it note, but there are a lot of little nods to past Nvidia products here.
My personal favorite is the reference to the glorious Nvidia G-Assist, seen among the folder names in the open file explorer window. This was an April Fool’s joke from way back in 2017, where Nvidia claimed it was releasing a USB thumb drive (shaped like a tiny GTX 1080 FE, no less) that could supposedly boost gaming performance. As much as I’d like to see this become a reality, I’ll be pretty shocked if it turns up in today’s livestream.
There’s also a nod to SETI@Home, the AI-powered program that utilizes unused resources in people’s home PC to create a network of systems to analyze radio signals from deep space in a huge, collective hunt for extraterrestrial intelligence.
Lastly, there’s a folder on the desktop titled ‘XXXXX with RTX’. I’m going to go out on a limb here and assume that this isn’t a new X-rated adult film business venture from Nvidia, but there’s probably a five-letter word that’ll become suddenly apparent during the presentation itself. ‘Tesla with RTX’ would be a pretty shocking way to announce the upgraded gaming capabilities of the Model S…
Good Morning Folks! It’s John again, big thanks to Christian for holding down the liveblog this morning as we head into Nvidia’s big announcement.
Will Jensen Huang give his keynote from his kitchen? Will he be in the Nvidia Metaverse? God, I hope not, but we won’t know til the event kicks off in about 20 minutes, but TechRadar will be right here to bring you all the latest as it happens, so stay tuned!
And here we go folks!
Nvidia is really leaning into the Omniverse.
The micromachines energy of this presentation is strong.
Lovelace! 76 billion transistors is impressive as hell.
Oooh, new generation ray tracing cores sound impressive, but those tensor cores are arguably even more important. That’s the AI stuff.
Shader Execution Reordering (SER) could be the kind of this that can make native ray tracing possible without DLSS, so DLSS could be even more impressive with DLSS 3.
So a 4X increase in framerate with DLSS 3 over Native rendering.
This was a triumph!
Modders are going to love this.
There it is!
The RTX 4090, 24GB GDDR6X, $1,599, available October 12.
The RTX 4080, 16GB GDDR6X, $1,199, 12GB GDDR6X, $899, available in November.
No word on the RTX 4070, but the price drop on the Nvidia Ampere cards is a huge boon for budget builds. As awesome as the RTX 4090 is likely to be, most people really won’t need more than an RTX 3080 for many years to come, so seeing the price drop on those is huge.
I have yet to hear an industrial designer say that the meta/omniverse is something they’re really excited about, but what do I know. This could be like the biggest thing in engineering, but I’m not entirely sure that you “need” to be using augmented/virtual reality to do this kind of thing.
I have to admit, I am incredibly grateful to Nvidia for frontloading Nvidia Lovelace and the RTX 4090 and RTX 4080 announcement in this presentation.
RIP Atlan, which wasn’t just a chip, they were also the gigantic mechs in Doom Eternal that were powered by Argent Energy that fought back against the forces of Hell and helped defeat the Titans. Never forget.
Fully autonomous vehicles are never going to happen unless they are on specially segregated lanes. We already have these. They’re called trains. Just automate trains.
FYI, if you want to dig deeper into the Nvidia RTX 4090 and Nvidia RTX 4080, make sure to check out our news hubs linked above.
Hello, do you have a minute to talk about the Omniverse today?
It’s absolutely incredible that Nvidia treated the GeForce RTX graphics cards almost like an afterthought in this keynote. Almost all of this has been dedicated to cloud computing and automation. Clearly, you can see where Nvidia sees its future profits coming from.
Nvidia GeForce Beyond? More like Nvidia Beyond GeForce, amiright?
So we have clearly moved on to the business-oriented side of the presentation, so lets recap Nvidia’s RTX graphics card news.
Starting with Lovelace, we’ve got 76 billion transistors, versus Nvidia Ampere’s 54 billion transistors. Ampere was a 7nm chip architecture, though, and Lovelace is a 4nm architecture.
These are the full data center chips, however, and the AD102 in the RTX 4090 will be significantly cut down (for reference, the GA102 in the RTX 3090 has 28.3 billion transistors).
Next, there’s the Nvidia RTX 4090. This is the only graphics card announced today that we have a release date for, October 13. It is also going to retail for $1,599, which is $100 more than the RTX 3090 when it was released in 2020.
We don’t have a price for the UK and Australia, but it should come in around £1,250 and AU$2,600.
It will come with 24GB GDDR6X, which is the same as the RTX 3090 Ti. The biggest implication here is for 8K graphics, namely, will this be enough memory to run 8K gaming. That remains to be seen, and we really won’t be able to tell until we get an RTX 4090 in for review to test 8K graphics.
It also needs to be said that nobody is asking for digital avatars dwelling like trolls in the uncanny valley. No thanks, Violet, can I speak to a human please?
The Nvidia RTX 4080 is going to come in two variants, a 16GB GDDR6X version for $1,199 and a 12GB GDDR6X version for $899, and while there’s no release date per se, Nvidia did say it should be going on sale in November.
The RTX 3080 went on sale for $799, so this is a bit of a bump up in price, which was expected, but this doesn’t bode well for the RTX 4080 Ti, which will probably approach the RTX 3090 in price.