Nvidia is no longer a PC gaming company.

erek

erek

8-D
2FA
Joined
Dec 19, 2005
Messages
18,546
"And honestly, it doesn’t seem that surprising when you consider that the RTX 5090 launched with an MSRP of $1,999 yet regularly sold for upwards of $5,000, due in no small part to massive demand from AI-pilled coders and companies looking to take advantage of that sweet, sweet 32GB of DDR7 memory. A casual search of the major retailers this week turned up precisely zero 5090 cards available for purchase, even at that exorbitant price, and I suspect that will remain true for the foreseeable future.

It can feel heartbreaking if you’re dedicated to the hardcore DIY PC gaming lifestyle, but you can take solace in the fact that the GeForce RTX 5080, RTX 5070 Ti and RTX 5070 are still widely available and plenty good enough for high-res, high-speed PC gaming. Heck, the same leak that suggests the RTX 5090 is done for also claims that Nvidia will launch a new model of RTX 5080 with 24GB of DDR7 memory, which should offer a noticeable improvement over the current 16GB RTX 5080 cards.


But I have to say, the sooner you make peace with the fact that Nvidia is no longer in the business of supporting PC gaming, the better. Don’t get me wrong, Nvidia will continue to make new desktop graphics cards and talk a big game about how it values PC game players first and foremost. But the numbers have shown otherwise for some time, and that’s likely to continue for years to come.

After all, the AI-driven explosion of demand for high-end GPUs and silicon helped Nvidia become the first trillion-dollar chip company, and recent financial reports and SEC filings show the company’s revenue from data center products is now 4-5x that of its consumer hardware revenue. That’s presumably part of the reason why the company recently stopped reporting GPU sales as a discrete category entirely, instead lumping all GPU hardware sales (for both commercial and retail customers) into one category on its balance sheet alongside robotics, automotive and telecom products.



So please, for your own mental health, I’m begging you: do not give in to Jensen’s sales pitches at the next Nvidia press conference. I believe the company will keep making new graphics cards for the foreseeable future, but only to hold on to its commanding hold of the GPU market and keep up appearances while it milks the AI boom for all its worth. If AI demand collapses the way crypto seemed to crash a few years ago, maybe we’ll see Nvidia take up the mantle of being a PC gaming company once again.

But I don’t expect that will happen any time soon, and making peace with that has done wonders for the stress I feel right now as a frustrated fan of PC gaming. It’s helped me realize that I don’t need the latest and greatest hardware to have a blast gaming on my PC, and the same can be true for you. Take this time to play some games in your Steam(ing) pile of shame, get deeper into Linux gaming and explore all the other cool things you can do without an RTX 5090. Because I don’t know that high-end GPUs will be affordable for regular folks like you and me in the next few years, if ever—and I don’t think I’ll be buying an Nvidia GPU any time in the near future."

Source: https://www.pcworld.com/article/3260897/nvidia-is-no-longer-a-pc-gaming-company-get-used-to-it.html
 
I think it's more about not beeing able to get enough chips made, be it GPU's and now also RAM, so obviously as a company you first make what net's you the most money.

If they had access to unlimited chips I'm pretty sure we would see more gaming cards, it's not like AMD and intel are flooding the maket right now with new cards either as they have the same issues.
 
Oh, they'll say they always were a gaming company and never stopped, when the loans supporting the data center craze start to go bad and it all collapses in on itself.

But you got to be resolute and not be like this guy in the article, who will accept scraps from daddy huang.
 
I’ve been saying for nearly two years that gaming is doomed, and there is some evidence for that. It’s more visible now. Nvidia prefers to sell AI accelerators, AMD prefers to sell APUs, and Intel prefers to sell… whatever they have. Microsoft and Epic have bought up most of the big game studios and just fired their staff—so what? Most hardware is getting more expensive now, and the “new lines” are… handhelds with APUs from AMD and Intel (and maybe +Nvidia). Most games are on the UE5 crap that surprisingly runs well on APUs…
 
AVATARAT said:
I’ve been saying for nearly two years that gaming is doomed, and there is some evidence for that. It’s more visible now. Nvidia prefers to sell AI accelerators, AMD prefers to sell APUs, and Intel prefers to sell… whatever they have. Microsoft and Epic have bought up most of the big game studios and just fired their staff—so what? Most hardware is getting more expensive now, and the “new lines” are… handhelds with APUs from AMD and Intel (and maybe +Nvidia). Most games are on the UE5 crap that surprisingly runs well on APUs…
Click to expand...
Gaming is not doomed. It is the ownership of the PC that is. Everything is going as planned. You will literally own nothing and pay a subscription for every little thing and fucking like it.
 
AVATARAT said:
I’ve been saying for nearly two years that gaming is doomed, and there is some evidence for that. It’s more visible now. Nvidia prefers to sell AI accelerators, AMD prefers to sell APUs, and Intel prefers to sell… whatever they have. Microsoft and Epic have bought up most of the big game studios and just fired their staff—so what? Most hardware is getting more expensive now, and the “new lines” are… handhelds with APUs from AMD and Intel (and maybe +Nvidia). Most games are on the UE5 crap that surprisingly runs well on APUs…
Click to expand...
I'm pretty sure it was not even the 2000s when I first heard someone say that PC gaming is doomed. It was probably when the PS1 came out.

I wonder which comes first, the end of PC gaming or the year of linux :p
 
M76 said:
I'm pretty sure it was not even the 2000s when I first heard someone say that PC gaming is doomed. It was probably when the PS1 came out.

I wonder which comes first, the end of PC gaming or the year of linux :p
Click to expand...
Yes, but now the situation is different :) Back then, we owned both the games and the hardware; now that’s about to change.

This! Yes, it’s a mystery to me why we have a “Year of Linux.”
 
I've tried cheap Aliexpress HW before, it could actually bring back the best days of PC building and overclocking, where it is the wild west, nothing is locked down.

If nvidia doesn't want to sell PC hardware someone else will, so anyone who believes this is the end of PCs, doesn't see further ahead than his own nose. nvidia will come crawling back when the AI datacenter scam goes away. And the cracks are already there. Now they are trying to distance themselves from the AI bubble in front of investors by claiming that only one third of their revenue comes from it. Which is obvious BS. Even their own quarterly results say it is BS.
 
This is such a niche situation for the so slim never thought a 9070xt and its successor could be enough to play game population ?

vegeta535 said:
It is the ownership of the PC that is. Everything is going as planned.
Click to expand...
Maybe but the amount of PC Apple sales at a price people can afford right now is still impressive:

https://www.apple.com/ca/shop/buy-mac/mac-mini
Or costco:
https://www.costco.com/p/-/msi-aegi...b-ram-1tb-ssd-windows-11/4201007873?langId=-1

Following spot price in the DIY price will make the situation look more extreme than it is, $650 for a PS5 with 2 dual sense controller is not ideal for an old console, but it is far away from it is hard to own local compute to play games scenario.

M76 said:
Now they are trying to distance themselves from the AI bubble in front of investors by claiming that only one third of their revenue comes from it.
Click to expand...
What 33% of their revenues from AI would mean here ? do you mean the split between the amount of data center compute that inference only vs training new model ? or hyperscaler vs others concentration ? Never heard them say that and I tend to follow-listen their announcement quite a bit since Nvidia ballooned to be a large part of my asset by now. In their latest results announcement it did not felt at all like them trying to distance themselves from AI.
 
M76 said:
I've tried cheap Aliexpress HW before, it could actually bring back the best days of PC building and overclocking, where it is the wild west, nothing is locked down.

If nvidia doesn't want to sell PC hardware someone else will, so anyone who believes this is the end of PCs, doesn't see further ahead than his own nose. nvidia will come crawling back when the AI datacenter scam goes away. And the cracks are already there. Now they are trying to distance themselves from the AI bubble in front of investors by claiming that only one third of their revenue comes from it. Which is obvious BS. Even their own quarterly results say it is BS.
Click to expand...
It would be nice if I were wrong.

The “bubble” in the field of artificial intelligence is set to undergo a change; since it first appeared, nothing different has appeared on the horizon. For now, it remains within predictable boundaries.
And those are that only a few major players will remain in the market; all the others will disappear or be absorbed by the big players. Those that disappear will take the capital invested in them with them, and there will be bankruptcies—exactly as predicted from the very beginning.
Combined with high hardware prices and, on the other hand, supply issues with Chinese hardware, the gaming market will shrink and/or shift to the cloud or some form of subscription service. These are all things that are “simply happening” right now.
 
LukeTbk said:
What 33% of their revenues from AI would mean here ? do you mean the split between the amount of data center compute that inference only vs training new model ? or hyperscaler vs others concentration ?
Click to expand...
They claim a third of their total data center revenue comes from completely non-AI related compute, ie. traditional machine learning and simulation tools. And only a third of it comes from hyperscalers. While the remaining third comes from small startups using open weight models. But revenue generated by open weight models is a drop in the bucket, something ain't mathing here.
LukeTbk said:
Never heard them say that and I tend to follow-listen their announcement quite a bit since Nvidia ballooned to be a large part of my asset by now. In their latest results announcement it did not felt at all like them trying to distance themselves from AI.
Click to expand...

That's exactly the point they boasted about it in their latest report, but now Jensen is trying to downplay it in an interview. Which to me signals that investors are no longer that convinced that the hundreds of billions spent on data center buildout by hyperscalers is going to generate profit.
 
AVATARAT said:
It would be nice if I were wrong.

The “bubble” in the field of artificial intelligence is set to undergo a change; since it first appeared, nothing different has appeared on the horizon. For now, it remains within predictable boundaries.
And those are that only a few major players will remain in the market; all the others will disappear or be absorbed by the big players. Those that disappear will take the capital invested in them with them, and there will be bankruptcies—exactly as predicted from the very beginning.
Combined with high hardware prices and, on the other hand, supply issues with Chinese hardware, the gaming market will shrink and/or shift to the cloud or some form of subscription service. These are all things that are “simply happening” right now.
Click to expand...
Nah, it's the exact opposite. The big players like openAI and Antropic will crash and burn as soon as investors realize they have no hope of generating enough revenue to justify their insane data center demand. And then the market will be dominated by open weight models.

I'm not saying AI will go away, just as the intrenet didn't go away after the web 2.0 crash. But spending on AI will normalize, and nvidia, and all the memory makers will reluctantly come crawling back to consumer electronics.
 
when the AI hype dies Jensen will come back to the gamer's market and say that it was the plan all along
 
M76 said:
They claim a third of their total data center revenue comes from completely non-AI related compute, ie. traditional machine learning and simulation tools. And only a third of it comes from hyperscalers. While the remaining third comes from small startups using open weight models. But revenue generated by open weight models is a drop in the bucket, something ain't mathing here.
Click to expand...
Which is really different than only 33% come from AI, more closer to 66%. That 33% would become blurry, "traditional" machine learning/big data/sql/parrallel compute transistion away from CPU toward GPU are of course a big thing, but they are often AI issh. non generative AI could be a cleaner way to say it.

Non hyperscaler revenues are not necessarily just neocloud or small start-up, a lot are from sovereign AI infracstructure (Saudis for example) or the big enterprise like the Walmart/big pharma/big banks/SpaceX-tesla until recently before they started to rent making some owngrown infracstructure in conjecture of their hyperscaler cloud use. Revenue generated by open weight models are hard to track, if you run open weight model on your own infrastructure it will not show up anywhere and if you rent a private Microsofot/AWS/google compute capacity, they do not know-track much what you do with it exactly and will not show up necessarily as open weight model revenues, those revenues tend to trakc only from services that rent tokens directly.

non hyperscaler can be something like this: https://vision2030.ai/analysis/humain-ai-infrastructure/

M76 said:
The big players like openAI and Antropic will crash and burn as soon as investors realize they have no hope of generating enough revenue to justify their insane data center demand. And then the market will be dominated by open weight models.
Click to expand...
I am really not sure about that, Anthropic just had 2 positive Ebida quarter in a row, their revenues according to their IPO filling are growing 10-12x yearly versus only 3x for their compute, open weight do not require less data center to run usually more, I am not sure how can physically work and not those from Anthropic. If anthropic model does not work, no way the open weight with less ressource and significantly costlier-less efficiant to run can work.
 
LukeTbk said:
Which is really different than only 33% come from AI, more closer to 66%. That 33% would become blurry, "traditional" machine learning/big data/sql/parrallel compute transistion away from CPU toward GPU are of course a big thing, but they are often AI issh. non generative AI could be a cleaner way to say it.
Click to expand...
Not my words, jensen made a point of saying it's not AI, complain to him. And many startups are just buying tokens from the frontier labs and reselling them. So there might be overlap and double counting.
LukeTbk said:
I am really not sure about that, Anthropic just had 2 positive Ebida quarter in a row.
Click to expand...
Nowhere near what is necessary to finance even a fraction of the datacenter commitments. And they are playing fast and loose with extrapolating run rates. Plus their IPO which has been delayed indefinitely, was also recently devalued.
 
You must log in or register to reply here.
Back
Top