Friday, January 16th 2026

NVIDIA Reportedly Ends GeForce RTX 5070 Ti Production, RTX 5060 Ti 16 GB Next

NVIDIA has reportedly stopped production of its mid-range GeForce RTX 5070 Ti graphics card, according to ASUS, one of the largest PC manufacturers. ASUS shared exclusive information with Hardware Unboxed, revealing that NVIDIA has created a supply shortage for the GeForce RTX 5070 Ti, effectively ending its production. As one of NVIDIA's largest AIB partners, ASUS has learned that the model is essentially in end-of-life status. The company noted the supply shortage and offered no further comments on availability or potential improvements in supply. Since retailers and manufacturers can't source enough stock of the GeForce RTX 5070 Ti GPUs, the shortage has led to significant price increases for these models and suggests that the future availability of this GPU SKU might be nonexistent.

Update 2, January 16, 08:30 UTC: ASUS issued an official public statement regarding the "incomplete" information about the RTX 5070 Ti and RTX 5060 Ti EOL.
ASUSWe would like to clarify recent reports regarding the ASUS GeForce RTX 5070 Ti and RTX 5060 Ti 16 GB. Certain media may have received incomplete information from an ASUS PR representative regarding these products. The GeForce RTX 5070 Ti and GeForce RTX 5060 Ti 16 GB have not been discontinued or designated as end-of-life (EOL). ASUS has no plans to stop selling these models. Current fluctuations in supply for both products are primarily due to memory supply constraints, which have temporarily affected production output and restocking cycles. As a result, availability may appear limited in certain markets, but this should not be interpreted as a production halt or product retirement. ASUS will continue to support the GeForce RTX 5070 Ti and RTX 5060 Ti 16 GB and is working closely with partners to stabilize supply as conditions improve.
Below is an older update and the remaining of the original text.

Update 1, 19:45 UTC: NVIDIA issued a statement to HardwareLuxx confirming that the series remains in production, but elevated demand is drying up the inventory.
NVIDIA to HardwareLuxxWhat changes are occurring in the GeForce ecosystem given the current memory market?
Demand for GeForce RTX GPUs is strong, and memory supply is constrained. We continue to ship all GeForce SKUs and are working closely with our suppliers to maximize memory availability.
The next phase of the silent production end is approaching for the GeForce RTX 5060 Ti 16 GB version, with resources reportedly being redirected to the lower-capacity GeForce RTX 5060 Ti 8 GB version. Stocks of the RTX 5060 Ti 16 GB SKU have been dwindling across retailers and distributors, and prices are beginning to reflect this scarcity. The massive industry DRAM shortage has affected even graphics GDDR7 memory. As a result, the lower-capacity 8 GB version will remain available until the supply chain and memory production improve. Yesterday's rumors also confirm that the regular RTX 5060 will be the priority card for NVIDIA going forward, for that segment.

Additionally, ASUS provided information about the previously rumored CES launch of the RTX 50 "SUPER" refresh, which was expected to have higher memory capacity. According to the AIB, this launch did not occur, although the original plan was indeed for a CES launch, which was postponed. Currently, there are no plans to release a series refresh, and some AIBs have reportedly expressed frustration over the sudden change. It seems the rumor about the SUPER launch being postponed was true, with the primary reason being the shortage of high-capacity GDDR7 memory.
Source: Hardware Unboxed
Add your own comment

210 Comments on NVIDIA Reportedly Ends GeForce RTX 5070 Ti Production, RTX 5060 Ti 16 GB Next

#101
Rover4444
Dirt ChipNothing To see here, move along. It is just a temporary 'restocking issue'.
All if fine. Believe us. Really.

Slowly boiling that frog, are you Asus?
Not the first time they've done it. I'm NEVER fucking forgiving them for what they did to the Zenfone.
Posted on Reply
#102
cinemaware
Rover4444What are you even talking about, at worst it was a 4070 Ti Super with a $50 discount. It's the best value card in NVIDIA's stack aside from the 5060 Ti 16GB.
You mean you somehow managed to memory hole all the bile that was spewed here and elsewhere when Blackwell was released? Quite a feat.
Rover4444Yes. And it's billions, actually. They need to keep producing 5080s for GFN, which is why they're not getting cut. They're also the higher value SKU.
They are earning billions from GeforceNow? Who'd've thunk /lol. I guess it's all the part of that dastardly plan to get us all slaved to subscriptions. Any moment now :D

**********************
Meanwhile, in the real world:
wccftech.com/the-current-state-of-nvidia-rtx-50-gpus-our-exclusive-report-on-geforce-production-aic-supply-memory-shortages-prices-eol-rumors/

tl;dr:
Our findings have brought us more information on the matter, and while the memory shortage's impact is real and affecting the entire consumer market, most information is either not passed properly or has been presented in a way that's completely devoid of the truth.
...but don't let it get in the way of this fun pile on....
Posted on Reply
#103
Dragam1337
cinemawareYou mean you somehow managed to memory hole all the bile that was spewed here and elsewhere when Blackwell was released? Quite a feat.

They are earning billions from GeforceNow? Who'd've thunk /lol. I guess it's all the part of that dastardly plan to get us all slaved to subscriptions. Any moment now :D

**********************
Meanwhile, in the real world:
wccftech.com/the-current-state-of-nvidia-rtx-50-gpus-our-exclusive-report-on-geforce-production-aic-supply-memory-shortages-prices-eol-rumors/

tl;dr:

...but don't let it get in the way of this fun pile on....
Yeah, cause corporations are always honest about what they are doing... were you born yesterday, or what ?
Posted on Reply
#104
Bobaganoosh
Rover4444I think they'd definitely need to sell to professionals at least. Enterprise customers would absolutely riot if they didn't get CUDA, NVIDIA would just be leaving money on the table. They also need to keep producing gaming GPUs even if they're not selling them to keep GFN afloat, that's a cash cow.

Yeah, the AI bubble where 8GB GPUs are the norm for yet another gen because the entire supply of DRAM for this year just got allocated this month.
why do they "need" to? People can riot all they want, there's no obligation there. Nvidia's only obligation is to shareholders and they're demanding maximum profits. Maximum profits come from (short-term anyway) putting all semiconductor allotments into data-center products. Sure, people will be pissed. Sure, Nvidia makes money on all those other income-sources, but they don't make as much money and if they actually want to maximize profits, nothing prevents them from walking away from those customers (after existing contracts are finished).
Posted on Reply
#105
LastDudeALive
Rover4444Yes. And it's billions, actually. They need to keep producing 5080s for GFN, which is why they're not getting cut. They're also the higher value SKU.
GeForce Now isn't 5080s in a server, it's enterprise-class Blackwell cards (probably either a Blackwell Ultra or an RTX Pro 6000 Blackwell), with sufficient CUDA cores assigned to your VM to simulate a 5080's processing power. That allows them to use the same GPU for a few 5080 sessions or more lower-tier sessions. It's also why you can only play approved games (it's a very large list, over 5000, but still a limited selection). They need to make custom drivers for each game for the virtual machine GPU.
Posted on Reply
#106
Dirt Chip
BwazeI wonder what will become of such tech review sites, YouTube reviewers and influencers that are now basically just reporting doom in the consumer sector?

I mean, during cryptohighs when GPUs were basically unavailable to mere mortals, most other sectors saw very little disruption, and we all knew every high is followed by a crash. But now? It's anybody's guess how this will turn out, and we're definitely following the more pessimistic outcomes for now.
I hypothetically predicted we will see branching:
Some will be bought by whoever has the big cash and drive for better PR snd we will start to see why all that happens is not that bad, it’s the next chapter in PC, that buying computational power for gaming by monthly subscription is now the thing and it’s also cheaper (after the intonational inflation of all pc parts be whoever they are) and so on.
The other will be the avangard, GN/HUB-Steve style of people who fight the Don-Quixote war with how they are corrupted kink of videos.

More interesting is what will happen to forum like tpu when HW reviews will be non relevant..

Also, a new golden era of retro pixeleized games will have good demand as ‘f#@! You AAA developers’ kind of protest.
Posted on Reply
#107
cfenton
Dragam1337Gddr7 is the primary reason that the 5060 ti is 15% faster than 4060 ti.
I was talking about the cost, not the performance. GDDR7 is definitely faster. The post I was replying to suggested AMD using GDDR6 might allow them to keep prices down. Maybe it will, but I'm skeptical because they both require modern processes and if the cost of the wafers is the primary driver of increased costs, then it probably won't make much difference.
Posted on Reply
#108
Dragokar
Of course, they are not EOL, you just can't buy them anymore or if one escapes the AI you can't afford them XD
Posted on Reply
#109
Tek-Check
wolfAsus got HUB'd. If that becomes a trend then these AIB's will be very hesitant to say anything to them, I suspect many already are and they found one Rep that was a little loose lipped and opionated.
That's ok. There are also anonymous workers from many OEMs who sometimes inform major tech outlets of what is really going on behind the camera that big companies do not want us to know, and on time. For example, it was Microslop that encouraged OEMs to expand BIOS chips to 64MB inn order to get WiFi drivers there and "help" users install Windows with logging into Microslop account. They actively discourage users from installing the OS with a local account. Microslop is already experiencing consequences of this and other actions.
Posted on Reply
#110
chrcoluk
Disgusting behaviour, they obsessed with VRAM starvation, they are rich enough they can subsidise increased VRAM costs on consumer GPUs.
Posted on Reply
#111
Dr. Dro
Hecate91Or they could make even more money selling both Radeon and Geforce cards, but MSI probably wanted that sweet Nvidia exclusive AIB money.
Not necessarily. If the volume is not high enough, the process of research and development, manufacturing and distribution basically becomes not worth it. With the low dGPU market share AMD currently holds plus the incentives you brought up, MSI has likely determined that continuing to work with AMD products is not worth the effort. You have to remember that MSI isn't just the MSI brand, but Micro-Star is a massive OEM that manufactures hardware for a megaton of OEMs and ODMs as well. It's about money and business sense, after all.
chrcolukDisgusting behaviour, they obsessed with VRAM starvation, they are rich enough they can subsidise increased VRAM costs on consumer GPUs.
For-profit corporations do not spend money on charity, and they do enough charity even building these gaming cards. The truth remains, if Nvidia pulled out of the consumer space today and devoted their entire stock of GB203 and GB202 chips to AI products, it's quite likely their stock price would increase.
Posted on Reply
#112
Youseethis
At least on the gaming front, mainly the the 9th gen AAAAAAAA games are affected. For most games a 3080 ti/4070 ti/5070 12gb provide a comfortable 4K@60 experience. Even the 8gb 5060 ti could manage.
Posted on Reply
#113
Dragam1337
Dr. DroNot necessarily. If the volume is not high enough, the process of research and development, manufacturing and distribution basically becomes not worth it. With the low dGPU market share AMD currently holds plus the incentives you brought up, MSI has likely determined that continuing to work with AMD products is not worth the effort. You have to remember that MSI isn't just the MSI brand, but Micro-Star is a massive OEM that manufactures hardware for a megaton of OEMs and ODMs as well. It's about money and business sense, after all.



For-profit corporations do not spend money on charity, and they do enough charity even building these gaming cards. The truth remains, if Nvidia pulled out of the consumer space today and devoted their entire stock of GB203 and GB202 chips to AI products, it's quite likely their stock price would increase.
I am not so sure - hopefully investers will see that nvidia is alot stronger longterm, if it maintains it's existing businesses, rather than just putting all the eggs in the ai basket.
Posted on Reply
#114
Dr. Dro
Dragam1337I am not so sure - hopefully investers will see that nvidia is alot stronger longterm, if it maintains it's existing businesses, rather than just putting all the eggs in the ai basket.
I actually agree, since they hold 94% of this market. It's basically a monopoly on their end. But if they were in AMD's condition, say, 6% of desktop dGPU market share, and holding their AI position? It'd actually be logical to withdraw.
Posted on Reply
#115
Tomorrow
Dr. DroNot necessarily. If the volume is not high enough, the process of research and development, manufacturing and distribution basically becomes not worth it. With the low dGPU market share AMD currently holds plus the incentives you brought up, MSI has likely determined that continuing to work with AMD products is not worth the effort. You have to remember that MSI isn't just the MSI brand, but Micro-Star is a massive OEM that manufactures hardware for a megaton of OEMs and ODMs as well. It's about money and business sense, after all.
What R&D?

Most AIB's use coolers interchangeably on AMD and Nvidia cards with minimal differences to accommodate the die and PCB design.
AMD and and Nvidia both provide kits with GPU+VRAM to AIB's.

The only R&D is for the actual PCB and if they follow the reference design it's an easy thing to do.
We're not talking about a high end cards here with overengineered coolers and complex PCB's.

AMD holds roughly 18-19% of the GPU market according to Steam HW survey (even these numbers are likely under representation).
Clearly both ASUS and Gigabyte find it worth enough to produce AMD cards despite the R&D cost and supposed low AMD market share.
Sapphire is an exclusive AMD partner. If they didn't make a profit then they would have not existed for 25 years.
Dr. DroFor-profit corporations do not spend money on charity, and they do enough charity even building these gaming cards. The truth remains, if Nvidia pulled out of the consumer space today and devoted their entire stock of GB203 and GB202 chips to AI products, it's quite likely their stock price would increase.
Gaming cards are "charity"?

Not with these MSRP's and markups. Far from it. That's just lazy excuse to justify corporate actions to say that they dont do "charity". Selling 70 class dies for 1300+ is by no means barely getting by.

No one knows how long the current AI demand will stick around. It's a bubble. It may burst this year or in five years. Or just level off. Like what happened with crypto mining. Sooner or later everyone will come crawling back.

Gaming will be here in 5, 10 and 50 years. It's a steady baseline income that's guaranteed no matter what. Assuming the products are good and reasonably priced of course. And that the company itself doesn't squander their advantage like Intel did.
Posted on Reply
#116
evernessince
cinemawareIs there a gif which could demonstrate the hypocrisy and crocodile tears people ITT are shedding over 5070Ti? Because I seem to recall that when Blackwell came out it was declared to be overpriced, backwards, fakeframed moneygrab scam, or something to that effect.

It's amazing how easily people can lose their minds over a rumour that predictably turned out be a lot of hot air from, predictably, Hardware Unboxed. But they got their clicks and the believers their narrative reinforced, so it's all good I guess.

Really, if you read some of the more outlandish conspiracies here, like the one where Nvidia kills the Ti to push 5080, it's pure comedy. Do you really think that a trillion $ company which is now all about AI would have time for such silly games, to earn what from this genius gambit...a few million maybe? Gimme a break.

Now, if you hate AI, then sure, it is the real reason for the scarcity. I hate this situation too, but also am aware of the fact that if it was me in the driving seat of a huge corpo in a capitalist system, I would absolutely do the same. You'd be mad not to prioritize your resources this way when being at the forefront of one of possibly greatest advances in tech in decades, if not centuries.

And AI really has the possibility to be just that. I'm not a great fan myself and if I could switch it off forever I probably would. But there’s no denying that if it realizes even a fraction of what the plans are, it'll be a seismic shift. It will be most likely for the worse for most of us (just like the internet turned out to be) but it's also a fact that humanity is sort of stuck at the moment when it comes to tech advances in general. And our long term survival prognosis doesn't look too rosy. If AI could help to propel us in some new directions, then at least that would count for something.
ASUS said they were EOLing certain SKUs on their end and then Nvidia twisted their arm to get them to state otherwise. Says all we need to know about this scenario. It's not like Nvidia suddenly just allocated them a bunch more dies.

The cards will remain "in production" to maintain appearances I'm sure but as for actual stock, I doubt it will be bountiful. Heck I doubt this announcement carries any actual material changes. Nvidia just wants to save face. Much like they tried when they made LHR cards to "help" gamers only to conveniently leak a BIOS that got rid of that restriction (and then they want to make Crypto only GPUs, directly reducing the number of gaming GPUs on the market).
Posted on Reply
#117
Dr. Dro
TomorrowWhat R&D?

Most AIB's use coolers interchangeably on AMD and Nvidia cards with minimal differences to accommodate the die and PCB design.
AMD and and Nvidia both provide kits with GPU+VRAM to AIB's.

The only R&D is for the actual PCB and if they follow the reference design it's an easy thing to do.
We're not talking about a high end cards here with overengineered coolers and complex PCB's.

AMD holds roughly 18-19% of the GPU market according to Steam HW survey (even these numbers are likely under representation).
Clearly both ASUS and Gigabyte find it worth enough to produce AMD cards despite the R&D cost and supposed low AMD market share.
Sapphire is an exclusive AMD partner. If they didn't make a profit then they would have not existed for 25 years.
It's an expense nonetheless and there are costs associated with SKU maintenance, RMAs, etc. - it really must have been deemed not worth it, or they wouldn't have dropped it. Sapphire and XFX have only done AMD cards for a very long time, it's their business. But if you look at the size of ASUS or MSI's operations, vs. the size of Sapphire of XFX, it should be self-explanatory.

And you're counting integrated graphics into this mix. Remember that Nvidia has an effective 0% iGPU market share, the last time they did iGPUs was in early AM3/late 775 days, almost 20 years ago now. I am specifically talking about dGPUs, and the 94/6/0 split report is from JPR, not Steam.
TomorrowGaming cards are "charity"?

Not with these MSRP's and markups. Far from it. That's just lazy excuse to justify corporate actions to say that they dont do "charity". Selling 70 class dies for 1300+ is by no means barely getting by.

No one knows how long the current AI demand will stick around. It's a bubble. It may burst this year or in five years. Or just level off. Like what happened with crypto mining. Sooner or later everyone will come crawling back.

Gaming will be here in 5, 10 and 50 years. It's a steady baseline income that's guaranteed no matter what. Assuming the products are good and reasonably priced of course. And that the company itself doesn't squander their advantage like Intel did.
Yes. Why sell an RTX 5090 at $2,000 if they can use the same core to make a Pro 6000 and sell it for $8,800? And they will.

Your rage, however impotent, is justified, I will acknowledge as much. But it just remains they could almost completely snuff out their gaming business, and neither AMD nor Intel have the production capacity or the mojo to fill up the void.
Posted on Reply
#118
Dragam1337
Dr. DroYes. Why sell an RTX 5090 at $2,000 if they can use the same core to make a Pro 6000 and sell it for $8,800? And they will.
Not exactly - the dies used for 5090 are essentially dies that weren't good enough to become 6000 cards, and has alot of cores and cache disabled compared to the 6000.
Posted on Reply
#119
Tomorrow
Dr. DroIt's an expense nonetheless and there are costs associated with SKU maintenance, RMAs, etc. - it really must have been deemed not worth it, or they wouldn't have dropped it. Sapphire and XFX have only done AMD cards for a very long time, it's their business. But if you look at the size of ASUS or MSI's operations, vs. the size of Sapphire of XFX, it should be self-explanatory.
AMD has fewer number of SKU's and less RMA's due to lower market share. I doubt MSI's decision had anything to do with that. I get the feeling some higher up was simply not in good terms with the Radeon division and hence the decision. They still make AMD motherboards so clearly they have no problem working with the Ryzen team.

And before you say Ryzen has much higher market share - yes it does now, but MSI has made AM4 boards back when Ryzen's market share was abysmal.
XFX has the most SKU's of any AMD card. At least with RDNA4. Clearly the margins must be good enough.
Dr. DroAnd you're counting integrated graphics into this mix. Remember that Nvidia has an effective 0% iGPU market share, the last time they did iGPUs was in early AM3/late 775 days, almost 20 years ago now. I am specifically talking about dGPUs, and the 94/6/0 split report is from JPR, not Steam.
No im not. AMD's market share on steam vs Nvidia actually rises if i discard AMD iGPU's and Nvidia mobile chips. That's because Nvidia mobile chips hold a proportionally higher share in Steam stats. Just look at the top10 and how high the laptop versions are. Take those away and as a consequence AMD share actually increases. Granted it does only so by 1-2% from ~18% that accounts for all GPU's.

JPR's numbers are meaningless to quote for a gamer market. JPR publishes quarterly shipments. They dont account for historical backlog.
Naturally Nvidia dwarfs AMD in quarterly shipments due to server cards. JPR mixes all into one. Realistic Nvidia share in gamer dGPU's is roughly 80%. Not 94%.
Dr. DroYes. Why sell an RTX 5090 at $2,000 if they can use the same core to make a Pro 6000 and sell it for $8,800? And they will.
But they cant. 5090 is a cut down GB202 that cannot be sold for 9k. Hence the 5090. Gamers get scraps and pay trough the nose for it.
At least with AMD you get the full unlocked chip. Their workstation cards using the same die only double the VRAM and enable ECC.
This could change with RDNA5/UDNA if they decide to go the Nvidia route and have one architecture for both.
Dr. DroYour rage, however impotent, is justified, I will acknowledge as much. But it just remains they could almost completely snuff out their gaming business, and neither AMD nor Intel have the production capacity or the mojo to fill up the void.
It's better to keep the gaming business going. No matter how small. Starting up from scratch later is very difficult and requires big expenditure.
That is why i doubt Micron will return to direct to consumer sales any time soon even if the demand dies down.
Posted on Reply
#120
Hecate91
TomorrowWhat R&D?

Most AIB's use coolers interchangeably on AMD and Nvidia cards with minimal differences to accommodate the die and PCB design.
AMD and and Nvidia both provide kits with GPU+VRAM to AIB's.

The only R&D is for the actual PCB and if they follow the reference design it's an easy thing to do.
We're not talking about a high end cards here with overengineered coolers and complex PCB's.

AMD holds roughly 18-19% of the GPU market according to Steam HW survey (even these numbers are likely under representation).
Clearly both ASUS and Gigabyte find it worth enough to produce AMD cards despite the R&D cost and supposed low AMD market share.
Sapphire is an exclusive AMD partner. If they didn't make a profit then they would have not existed for 25 years.
Some of the Nvidia AIB's probably just get by on GPU margins, hence the massive amount of SKU's, and the extra extravagant cards to lure in the wealthiest high end buyers.
Nvidia also said they weren't providing GPU+VRAM as a bundle anymore, AMD promised they would try to keep prices down on 9000 series, GDDR6 may be less affected, and maybe AMD are doing retailer rebates again.
I'm not sure about PCB R&D, but they all have to design coolers, pay for marketing, and that costs extra when Nvidia can write off the cost or just keep supply limited while AIBs have to panic to try to compete with FE cards undercutting them on price.

Asus and Gigabyte still make decent Radeon cards, MSI's Radeon cards were often reusing Geforce coolers in the worst possible ways with poor die contact and thermal pads that didn't do anything.
AMD has more of the market than most people want to give credit for, or simply want to maliciously downplay any other competition in the market because marketshare is everything for some reason, Sapphire, XFX, and Powercolor are AMD only AIBs, if there wasn't any profit to be had with AMD cards then those brands wouldn't exist either.
TomorrowGaming cards are "charity"?

Not with these MSRP's and markups. Far from it. That's just lazy excuse to justify corporate actions to say that they dont do "charity". Selling 70 class dies for 1300+ is by no means barely getting by.

No one knows how long the current AI demand will stick around. It's a bubble. It may burst this year or in five years. Or just level off. Like what happened with crypto mining. Sooner or later everyone will come crawling back.

Gaming will be here in 5, 10 and 50 years. It's a steady baseline income that's guaranteed no matter what. Assuming the products are good and reasonably priced of course. And that the company itself doesn't squander their advantage like Intel did.
Gaming cards aren't a charity, although Nvidia could dump the gaming market, which is only 8% of their overall sales compared to AI datacenter, and still make money.
Or just dump the lower end SKU's and tell the poor gamers to pay up for Geforce Now. As long as Jensen can convince his investors and everyone in the AI circle that AI isn't a bubble, the money will keep getting passed around.
I don't think PC gaming will die off, but it will be very different as people are priced out of dedicated GPU's, and a whole gaming PC as AI is destroying the market more than we've ever seen from shortages or crypto. I'd rather find a different hobby than subscribe to a gaming service or a mini PC with everything soldered onto the motherboard with no potential for upgrades.
Posted on Reply
#121
Dr. Dro
Dragam1337Not exactly - the dies used for 5090 are essentially dies that weren't good enough to become 6000 cards, and has alot of cores and cache disabled compared to the 6000.
But they are better than dies that go on the Pro 5000, and there is room to turn it into a Pro 5500 or something.
TomorrowAMD has fewer number of SKU's and less RMA's due to lower market share. I doubt MSI's decision had anything to do with that. I get the feeling some higher up was simply not in good terms with the Radeon division and hence the decision. They still make AMD motherboards so clearly they have no problem working with the Ryzen team.

And before you say Ryzen has much higher market share - yes it does now, but MSI has made AM4 boards back when Ryzen's market share was abysmal.
XFX has the most SKU's of any AMD card. At least with RDNA4. Clearly the margins must be good enough.

No im not. AMD's market share on steam vs Nvidia actually rises if i discard AMD iGPU's and Nvidia mobile chips. That's because Nvidia mobile chips hold a proportionally higher share in Steam stats. Just look at the top10 and how high the laptop versions are. Take those away and as a consequence AMD share actually increases. Granted it does only so by 1-2% from ~18% that accounts for all GPU's.

JPR's numbers are meaningless to quote for a gamer market. JPR publishes quarterly shipments. They dont account for historical backlog.
Naturally Nvidia dwarfs AMD in quarterly shipments due to server cards. JPR mixes all into one. Realistic Nvidia share in gamer dGPU's is roughly 80%. Not 94%.

But they cant. 5090 is a cut down GB202 that cannot be sold for 9k. Hence the 5090. Gamers get scraps and pay trough the nose for it.
At least with AMD you get the full unlocked chip. Their workstation cards using the same die only double the VRAM and enable ECC.
This could change with RDNA5/UDNA if they decide to go the Nvidia route and have one architecture for both.

It's better to keep the gaming business going. No matter how small. Starting up from scratch later is very difficult and requires big expenditure.
That is why i doubt Micron will return to direct to consumer sales any time soon even if the demand dies down.
Second highest cut of the GB202 available and it's "scraps"? See reply above.

There's no use attempting to skew the statistics to make AMD look better. They lost that battle a long time ago. Outside of iGPUs, their market share is very small right now and they are having major trouble clawing it back, thanks to the series of self-owns they've been inflicting on themselves lately. Mobile dGPUs are another can of worms I haven't even approached, but I can't even remember the last time I saw a laptop with an AMD dGPU, even at the higher volume segments. I'll just mention, XFX and SKU spam, name a more iconic duo?

Agree on Micron - and on keeping the gaming business going - though.
Posted on Reply
#122
Tomorrow
Dr. DroSecond highest cut of the GB202 available and it's "scraps"? See reply above.
The top gamer card has been more and more cut down every generation. Now also 1/4th of the memory capacity. Ok saying it scraps is perhaps too harsh.
Then lets put it this way - gamers are getting a shorter and shorter stick each gen while the cost keeps rising.

Say what you will about AMD , but they have kept their cards below 999. Party trough necessity yes, but in the end the consumer doesn't care why prices are the way the are. They care what the prices are. If i have to choose between two evils then im choosing the lesser of two evils.

Nvidia meanwhile keeps increasing prices far in excess of manufacturing cost. Now 1999 is normalized. If the next gen card in 2027 will be 2099 then people will be cheering that it's only 100 bucks more. Forgetting that before even 1000 was unheard of. Like they cheered 4090 being 1600 instead of 1500.
Posted on Reply
#123
Rover4444
Bobaganooshwhy do they "need" to? People can riot all they want, there's no obligation there. Nvidia's only obligation is to shareholders and they're demanding maximum profits. Maximum profits come from (short-term anyway) putting all semiconductor allotments into data-center products. Sure, people will be pissed. Sure, Nvidia makes money on all those other income-sources, but they don't make as much money and if they actually want to maximize profits, nothing prevents them from walking away from those customers (after existing contracts are finished).
NVIDIA isn't the only corporation capable of lobbying the government to get what they want. Turning enterprise against them is bad for business. It's not like they can't just use the same silicon they're using for GFN and rebadge it as a pro card, they're still thousands of dollars. In fact, they're doing it right now.
LastDudeALiveGeForce Now isn't 5080s in a server, it's enterprise-class Blackwell cards (probably either a Blackwell Ultra or an RTX Pro 6000 Blackwell), with sufficient CUDA cores assigned to your VM to simulate a 5080's processing power. That allows them to use the same GPU for a few 5080 sessions or more lower-tier sessions. It's also why you can only play approved games (it's a very large list, over 5000, but still a limited selection). They need to make custom drivers for each game for the virtual machine GPU.
It's not Blackwell Ultra, they don't have vGPU support. You're probably right though.
Dr. DroI am specifically talking about dGPUs, and the 94/6/0 split report is from JPR, not Steam.
Wow, that actually makes Intel holding %1 of the market at times pretty impressive.
Posted on Reply
#124
Bobaganoosh
Rover4444Turning enterprise against them is bad for business.
I agree completely. I think it would be a terrible idea. I'm quite biased (I'm a gamer who hates AI in it's current usage, and I also hate the massive waste of resources on building out datacenters that won't advance society at all, but more likely contribute to a massive economic depression). I just look at the numbers and the bit market players' recent actions and don't see it as an impossible future.
Posted on Reply
#125
Hilux SSRG
Shame on all the Nvidiots who bought Nvidia graphics cards rather than AMD or Intel when you had an option and they were price competitive/feature similar. Reap what you sow from your overlords.

Note this AI boom does not help but still...
Posted on Reply
Add your own comment
Nov 13th, 2026 19:06 PST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts