• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

[Rumor] Nvidia to cut gaming GPU production by 30-40% starting 2026

It very much is true though. They've not been supplying them very well and instead sending their chips for AI even prior to this RAM situation.
Based on what? Stock outside of the 5090 has been perfectly fine. Cards are frankly everywhere as normal.
 
Based on what? Stock outside of the 5090 has been perfectly fine. Cards are frankly everywhere as normal.
Based on nvidia not meeting demand and having very low volume at their "paper launch". This went beyond the 5090 too. It affected the 5080 as well. The stock situation is a little better now with lower demand (still often limited on 5090) but at launch there was no chip shortage and yet still limited supply because they were just directing their chips to AI. Stores barely got any stock. AMD then came and did a better supplied launch than nvidia and was praised for it if you remember.
 
Last edited:
If the DDR memory shortage pushes the RTX 6000 series into 2028, as RedGamingTech claims, then the RTX 4000 series and perhaps even the 3000 series (models with at least 10 GB of VRAM) will be the longest-lasting generation of GPUs ever. People who bought the RTX 4090 are especially happy now because they have been using a high-end GPU for three years and can still sell it for the same price (or maybe even make a profit) as they originally paid.
 
I mean if you are playing games fine now nothing is going to change in the near future. The consoles set the bar and it seems we have a good few years till the new ones are out. I'm curious to see how much plans change for console makers now. That what we should be looking at as a baseline.
 
Oh boy

I've got bad news for you..

Probably AMD will left too. They will have only apus in the Future is my guess.

At this point, Nvidia could lose gaming and it would barely make a dent in their revenue.
Unfortunately both AMD and Intel are also focusing too much on AI. They don't care all that much about taking market share in gaming from Nvidia.
Taking just 1% of revenue in AI is worth much more than all of the GPU market.

rw7gVfsvlIVf3AIw.jpeg
If they both stopped making gaming GPU's what would Sony do? Go back to cell?
 
If they both stopped making gaming GPU's what would Sony do? Go back to cell?
Sony could just go to Nvidia and switch to ARM like what Nintendo already did since the Switch 1

The only reason Sony is doing what they are doing is because they want to use x86 and if they ever decided to kick x86 to the curb they wouldn't need AMD anymore
 
Last edited:
How do you like it? You had a 9070xt before correct?

I did. Nvidia just wins the image quality matchup every time for me. It's all about DLSS vs FSR. I went from playing Clair Obscur on 9070 XT to 5080 and the difference was huge.

AMD is just in a catch 22, I think. They can't get more FSR 4 adoption until they have a larger base. Can't have a larger base without more devs implementing FSR 4.
 
AMD is just in a catch 22, I think. They can't get more FSR 4 adoption until they have a larger base. Can't have a larger base without more devs implementing FSR 4.
I think this is the main reason for Amethyst. AMD really wants/needs better developer support and will use consoles to get it across both.
 
Monitor I'm playing on a 4K 240Hz at the moment

ah.....of course. Some reason I was trying to translate that to a GPU.

500hz, huh? One day, when they invest bionic eyes and you need them to be able to see, the first thing out of your mouth is going to be "these fucker better have gsync"
 
I did. Nvidia just wins the image quality matchup every time for me. It's all about DLSS vs FSR. I went from playing Clair Obscur on 9070 XT to 5080 and the difference was huge.

AMD is just in a catch 22, I think. They can't get more FSR 4 adoption until they have a larger base. Can't have a larger base without more devs implementing FSR 4.
This has been true since FSR1. Only a fraction of games support FSR compared to DLSS even today after years of DLSS vs. FSR

AMD has it even worse trying to catch up in RT. They let Nvidia have a 7 years head start on getting devs to adopt Nvidia's RT standards so now all they can do is copy exactly what Nvidia does, they will never be able to try and improve on anything RT related when Nvidia had 100% of the market for nearly a decade and tens of millions of products already sold and running Nvidia's RT standards
 
Wasn't cell just a cpu and still relied on nvidia for GPU?
It could be used for graphics too, they put the gpu in there so devs would have something they were more familiar with

Really the best console chip ever made imo, wonder where it would be now if they had kept improving it
 
Last edited:
It could be used for graphics too, they put the gpu in there so devs would have something they were more familiar with

Really the best console chip ever made imo, wonder where it would be now if they had kept improving it
What would it have offered that an APU doesn't today?genuine question. Better chip size?

I think with GPU compute we surpassed it but I don't know the ins and outs of Cell.
 
Part of me wanted to get an RTX 5070 Ti to replace my RTX 3080... but I just couldn't do it.

Firstly, it's no even a 50% increase in framerate. That's after two freaking gens, for essentially the same tier card.

What's more the RTX 5070 Ti costs more than I originally paid for my RTX 3080.

So yeah, even if the ship has sailed and prices increase, it is what it is.
 
Part of me wanted to get an RTX 5070 Ti to replace my RTX 3080... but I just couldn't do it.

Firstly, it's no even a 50% increase in framerate. That's after two freaking gens, for essentially the same tier card.

What's more the RTX 5070 Ti costs more than I originally paid for my RTX 3080.

So yeah, even if the ship has sailed and prices increase, it is what it is.
Huh? Why didn't you just buy a 4080super at msrp? It is 51% faster than the 3080. The 5070ti is 47% faster. The difference is extremely noticeable.
 
Thank goodness i got in on the 5080 fe drop from nvidia marketplace last week. Microcenters in the east coast have a few in stock but they are $200 over msrp.
 
i'm glad i got my new PC a month ago, these raises are insane, i did put my old GPU up for sale but i took it off after this memory situation came out of nowhere, i might need it at some point.
 
What would it have offered that an APU doesn't today?genuine question. Better chip size?

I think with GPU compute we surpassed it but I don't know the ins and outs of Cell.
If used correctly I think it was capable of 2 tflops back when even 1 tflop GPU's didn't even exist yet

Developers started to get the hang of it towards the end of the gen like naughty dog with uncharted 3 and quantic dream with beyond 2 souls
 
Last edited:
If used correctly I think it was capable of 2 tflops back when 1 tflop GPU's didn't even exist yet
Yeah I remember it was pretty insane in theoretical flops for back in the day.

I meant if they kept improving it what would we have now? Didn't APU/GPUs today essentially become what Cell was pushing at the time? i.e. Very good at highly parallel computing tasks, or did it have something we don't really see in todays hardware?
 
Yeah I remember it was pretty insane in theoretical flops for back in the day.

I meant if they kept improving it what would we have now? Didn't APU/GPUs today essentially become what Cell was pushing at the time? i.e. Very good at highly parallel computing tasks, or did it have something we don't really see in todays hardware?
Oh idk about that, just wondered what it would be today if they kept using it
 
Huh? Why didn't you just buy a 4080super at msrp? It is 51% faster than the 3080. The 5070ti is 47% faster. The difference is extremely noticeable.

4080 Super was $1k. Fuck that.

47% is pitiful, especially if that means going from like 60 fps, to not even 90 fps. Again, fuck that.

This is why Nvidia gets away with selling their crap at the prices they do, because people are impressed by mediocre performance increases.
 
4080 Super was $1k. Fuck that.

47% is pitiful, especially if that means going from like 60 fps, to not even 90 fps. Again, fuck that.

This is why Nvidia gets away with selling their crap at the prices they do, because people are impressed by mediocre performance increases.
Well the 3080 to 4080 super was a proper gen on gen improvement. The only issue was the price. The 5000 series brought like a 10-15% improvement outside the 5090. In many ways it's just another super.

I feel you on the price for sure. However, I don't things will get better for Nvidia pc gamers. If you look at a breakdown of their revenue, you'll see where their bread is being buttered. It's not gaming and if they stopped making gaming gpus, it will have an insignificant effect on their revenue.
 
Based on nvidia not meeting demand and having very low volume at their "paper launch". This went beyond the 5090 too. It affected the 5080 as well. The stock situation is a little better now with lower demand (still often limited on 5090) but at launch there was no chip shortage and yet still limited supply because they were just directing their chips to AI. Stores barely got any stock. AMD then came and did a better supplied launch than nvidia and was praised for it if you remember.
Yes, the launch was dismal but you can't just simply extrapolate that to the rest of the year. As I've said, all Nvidia cards have been at or near the MSRP for well over half a year, and with plenty of stock across the board. The only exception being the 5090 (which is also still available everywhere, just at inflated prices). We have multiple data points showing that Nvidia consumer GPU sales havnt dropped. First, Steam survey, the 5070 is already eleventh on the chart, having overtaken the 4070 despite only being on the market for 9 months. Next, data from JPR, sales of dedicated consumer GPUs are up compared to last year, and finally Nvidia own gaming revenue, which is up significantly vs 2024 and 2023.
 
Yes, the launch was dismal but you can't just simply extrapolate that to the rest of the year. As I've said, all Nvidia cards have been at or near the MSRP for well over half a year, and with plenty of stock across the board. The only exception being the 5090 (which is also still available everywhere, just at inflated prices). We have multiple data points showing that Nvidia consumer GPU sales havnt dropped. First, Steam survey, the 5070 is already eleventh on the chart, having overtaken the 4070 despite only being on the market for 9 months. Next, data from JPR, sales of dedicated consumer GPUs are up compared to last year, and finally Nvidia own gaming revenue, which is up significantly vs 2024 and 2023.
Nobody extrapolated though when I clearly said "paper launch" which you replied to. I even still say it is today with how 5090s sell.

What is eleventh place in terms of volume? Of course Nvidia revenue increased they had a product launch and charged way more but what were the amount of units they were providing? What volume were they doing? Especially when datacenter units shot up by something like 140%. Not sure which JPR data you're on about but the one I've seen includes AMD sales and nvidia apparently fell slightly in marketshare.
 
Last edited:
Glad I recently picked up a 5060 TI 16GB.

Didn't upgrade my RAM from 16 to 32 when I had the chance though. I'll be stuck all next year with that by the looks of things.
 
Nobody extrapolated though when I clearly said "paper launch" which you replied to. I even still say it is today with how 5090s sell.

What is eleventh place in terms of volume? Of course Nvidia revenue increased they had a product launch and charged way more but what were the amount of units they were providing? What volume were they doing? Especially when datacenter units shot up by something like 140%. Not sure which JPR data you're on about but the one I've seen includes AMD sales and nvidia apparently fell slightly in marketshare.
You were not just talking about the launch, unless I misunderstood your original comment. And Nvidia didn't charge way more, only the 5090 went up in price, and we all know that is hardly selling that many units. AIB or retail scalping doesn't net them more money. And yes datacenter is doing way better relative to gaming, never claimed otherwise, I'm just using the revenue statement to show that Nvidia is still selling the same (or slightly more) GPUs than they have ever been. Out of the previous seven years, 2025 is the second highest revenue total, only being beat by 2022, and we all know the crypto boom helped fuel that. And yes, according to JPR Nvidia fell in market share slightly, but since GPU sales are up, they still sold more in total.
 
You were not just talking about the launch, unless I misunderstood your original comment.
Yes you did and I already clarified that I had said paper launch a few posts back and said their 5090 supply even today is not great.
And Nvidia didn't charge way more, only the 5090 went up in price, and we all know that is hardly selling that many units.
but you just said it went to eleventh to suggest volume was high when I said volume was low.
AIB or retail scalping doesn't net them more money. And yes datacenter is doing way better relative to gaming, never claimed otherwise, I'm just using the revenue statement to show that Nvidia is still selling the same (or slightly more) GPUs than they have ever been. Out of the previous seven years, 2025 is the second highest revenue total, only being beat by 2022, and we all know the crypto boom helped fuel that. And yes, according to JPR Nvidia fell in market share slightly, but since GPU sales are up, they still sold more in total.
Don't disagree with any of that but you would expect sales are higher with the launch of the much more expenisve (5090) 50xx series cards. The issue is that you've not really provided any evidence of how much volume they sold. Being beat by 2022 is because 2022 was 40xx launch and 50xx you're saying is lower. Even the 40xx cards were selling for higher than their MSRP because the supply situation was shit from nvidia. I'm not sure why you're so adamant to dispute that.
 
Last edited:
Yes you did and I already clarified that I had said paper launch a few posts back and said their 5090 supply even today is not great.

but you just said it went to eleventh to suggest volume was high when I said volume was low.

Don't disagree with any of that but you would expect sales are higher with the launch of the much more expenisve (5090) 50xx series cards. The issue is that you've not really provided any evidence of how much volume they sold. Being beat by 2022 is because 2022 was 40xx launch and 50xx you're saying is lower. Even the 40xx cards were selling for higher than their MSRP because the supply situation was shit from nvidia. I'm not sure why you're so adamant to dispute that.
Your comment made it sound like Nvidia is not selling that many GPUs to consumers anymore, even outside the launch period. And I said the 5070 was the eleventh most popular GPU on Steam. Not the 5090.

And I don't think the 5090 impacts revenue that much. Nvidia sells far more 60/70/80 class cards vs just the 90 series. And the 40xx launch wasn't in the fiscal 2022 report. That ended on Jan 2022 (Nvidia fiscal year is the year the report is made, not the actual year it was revenue from). Fiscal year for the 40xx launch would be 2023, which was one of the lowest revenue years recently due to all the leftover stock inventory from the crypto boom of the previous year. The 50xx GPU launch is in fiscal 2026, which is not over yet, but is already tracking higher than 2025 or 2022 and will be Nvidia highest gaming revenue on record.
 
Brother, don't get your head chopped off with that 8GB VRAM. The time was black friday, but I'd personally take the bet that getting a card now and then flipping your 3060Ti if you need to during shortages is the way to go. Get 16GB VRAM unless you are a 1080p player.

Yeah, I'm looking at new cards this morning, either a 5060Ti (~$450) or a 5070Ti (~$750). Both have 16GB VRAM, which is my main concern and priority for upgrading my 3060Ti which only has 8GB. I think a card with 16GB should future proof me for a long time whereas my 8GB current card will start to limit me more than I'd like.
 
I did. Nvidia just wins the image quality matchup every time for me. It's all about DLSS vs FSR. I went from playing Clair Obscur on 9070 XT to 5080 and the difference was huge.

AMD is just in a catch 22, I think. They can't get more FSR 4 adoption until they have a larger base. Can't have a larger base without more devs implementing FSR 4.
Ah makes sense. Expedition 33 just got FSR4 officially.

I've been lucky so far- games I want to play have it (BL4, Cronos). I made it a habit now to avoid non-FSR4 new games until they patch it in.

I still eye the 5080 to get around the issue though. I just have a hard time justifying the $600 upgrade cost.
 
Sony could just go to Nvidia and switch to ARM like what Nintendo already did since the Switch 1

The only reason Sony is doing what they are doing is because they want to use x86 and if they ever decided to kick x86 to the curb they wouldn't need AMD anymore

I couldn't think of a worse thing for Sony to do with Playstation LOL!
 
So what? If you didn't have a modern gaming GPU by now, you weren't really into modern PC gaming and this makes no difference.
 
Last edited:
Developers will have to adjust to the current situation and not push graphic boundaries while RAM and GPU are at a shortage. I have 32 GB of DDR5 and a 9070 XT, so it's enough to keep me busy while waiting for this AI garbage to calm down.
 
Developers will have to adjust to the current situation and not push graphic boundaries while RAM and GPU are at a shortage. I have 32 GB of DDR5 and a 9070 XT, so it's enough to keep me busy while waiting for this AI garbage to calm down.

Devs need to start optimizing their games properly.
This alone would solve a lot of issues.
 
Top Bottom