SantaC
Gold Member
the schedule was never more than 2 years, but now we are looking at 3, possibly 4 because they realize that they make more money elsewhere
-
Hey Guest. Check out your NeoGAF Wrapped 2025 results here!
AMD made over 8 times as much revenue on data centers as it did gaming last quarter
- Thread starter cormack12
- Start date
Felessan
Member
What should be the reason for them to care about gaming?
Hadoukens
Member
Well eventually the build out of AI infrastructure will stop. Upgrades and replacements will still be a thing, but the amount of product being sold will go down drastically.
They'll still need customers afterwards.
Last edited:
crackajack
Member
Talking about a bubble before 2024/2025 was probably premature. Since it was backed by cash in the earlier years and growth was somewhat reasonable. Apple, Google, MS have tons of cash. They burned in the past a lot on other things and can burn a lot on AI. Now though it is clearly a bubble that matches the dot com bubble insanity with way too many too big projects everywhere. Stocks might still go up, maybe even a lot, but there is no substance, yet. Since AI is still not actually intelligent, still just programms with now extreme energy hungry computations. Especially Oracle is being way too invested, on borrowed money, for its comparably small size. Ruining the consumer market for entertainment, mobile phones, but also skyrocketing budgets in every company for HW. So far the cost to benefit relation seems just out of whack and a incestous pyramid game hoping for the end goal to be reached.
The winner takes all might happen once AGI is actually achieved which means even the second and third place will lose a ton of money. And if its a chinese AI, which seem to do similar results with far less waste of everything to begin with, the end results will be funny.
Last edited:
Hudo
Gold Member
What really pisses me off that both Nvidia and AMD (but mostly Nvidia) are spamming their AI stuff at Siggraph, where they force their AI stuff on graphics. Only a fucking handful of their papers and talks are actually interesting, the rest is just essentially spam and sales bullshit. It affects the whole "seriousness" (it's not fucking easy to get a paper accepted there) of the conference, imho.
LordCBH
Gold Member
And not only building them, but convincing local and state governments to eminent domain people out of their homes for the benefit of private organizations. And those who didn't get put out of their homes but have to now live next to one of these centers have a drastically reduced quality of life.
I would end up in prison if I lived next to that monstrosity.
Felessan
Member
Realistically it'll be a point when they start to care
Not token presence to keep market share for future ifs, but something meaningful
Unless we migrate to cloud gaming of course, than they will never care about consumers products ever again
neocycle
Member
Maybe in 50 years. New, better chips and hbm are built. And AI will be built into lots of electronics devices, including new device categories like humanoid robots.
GoldenEye98
posts news as their odd job
Yeah I certainly wouldn't be assuming this...
That talk about a bubble prior to 2024 is unrelated to AI.
If you don't recall people were saying the entire market is in a bubble and we're going to have a recession. It's five or so years later and I'm still waiting. I should have invested the funds I had ready to scoop up deals. But I didn't and left some money on the table.
Varteras
Member
manlisten
Member
And NVIDIA, as usual, is way ahead of AMD on this front.
Bojji
Gold Member
It's also much harder to get more performance right now vs. for example 10 years ago. Die shrinks are expensive and provide laughable results (gains were massive in the past).
Between Ada and BW there was almost 3 years, now we will wait the same amount of time for Hopper (or whatever it will be called).
Felessan
- they still make new architectures every few years, might as well sell them to gamers (billions of "free" profit).
Last edited:
YCoCg
Gold Member
Shut up communist! Capitalism has no roof, you can never have enough money.
Dorfdad
Gold Member
They will. They are going to jump on the AI datacenter bandwagon. Everyone is abandoning gamers in droves. Selling 1000 video cards is now considered primitive when you can add some basic AI stuff and sell to companies for 3x per card.
If you got a 5000x hang on tight resist selling it will last you 10 years now
Intel stagnated for years. However, NVIDIA's latest AI server tech averages a 2-3x uplift and the new hardware was released in 12 months from the previous generation with 70% of the components redesigned.It's also much harder to get more performance right now vs. for example 10 years ago. Die shrinks are expensive and provide laughable results (gains were massive in the past).
Between Ada and BW there was almost 3 years, now we will wait the same amount of time for Hopper (or whatever it will be called).
Quantum leaps are still happening. Just not for the consumer.
CLW
Member
Just "hide" that prompt like that one professor then fail EVERYONE when they "write" nonsense and turn it in.
Trogdor1123
Gold Member
With this kind of money you really can't blame them for chasing it. There is just too many incentives to go after it.
Bojji
Gold Member
You talk about multi GPU on one part (racks)?
On PC we are comparing one GPU chip vs. one GPU chip, so you only get higher performance by adding more cores, better IPC, higher power usage etc.
BW is almost the same as Ada in those aspects, 5090 is better than 4090 only because it has more raw specs and much higher power usage.
I'm talking about the Vera Rubin servers. Same number of cores, etc. But 50% more compute and bandwidth. Which is leading to 2-3x training, inference, etc at the same power draw.
I was incorrect on the release, it was a 70% redesign that happened in 12 months. It's be produced right now. I'm not sure if any companies have them yet.
Bojji
Gold Member
Ok but try to achieve the same results with a single GPU.
And I'm talking about gaming stuff, for ML they just add more and more units with each gen.
Game performance does not scale like that.
Last edited:
But the Blackwell and Vera Rubin have the same number of CPU and GPU cores with that gain.
Agreed when it comes to games, we hit a point of diminishing returns. I'm still of the belief the RTX 5000 series will have some legs as renders move away from raw raster and integrate more ML into the pipeline.
dgrdsv
Member
The schedule is dictated by the ability to produce a lineup which will be worthy of buying, and that is dictated by the availability of better production tech at suitable prices.
What was previously is irrelevant, the reality of silicon production has changed, and we've not seen the worst of it yet.
Team Andromeda
Member
People like you don't see to want to accept that billions of people use AI and that's why corporations are spending silly amounts of money on it. I've only got to go out or look at Facebook to see people using AI for their photos or TikTok videos. Every girl in my workplace uses AI to touch up the photos they choose to upload
Labour replacement always happens, people like you would rather see farmers use shovels to dig rather than machines or have 13-year-old kids clean the chimney
IDKFA
I am Become Bilbo Baggins
Are you okay? I explained in my previous post that these AI corporations are not spending this sort of money because people are using free AI tools.
You didn't read my previous post, did you.
Labour replacement always happens, people like you would rather see farmers use shovels to dig rather than machines or have 13-year-old kids clean the chimney
I see people compare job losses from AI to the industrial revolution a lot.
However, the comparison to the industrial revolution is wrong because although mechanical automation replaced physical labor, humans were still needed to operate these machines and work in factories etc. This forced populations to migrate to cities and towns so they could work in these factories where human labor was required, which in turn saw cities and town experience mass population growth. Human cognitive and dexterity skills remained irreplaceable, allowing workers to pivot into manufacturing, logistics, and eventually the service sector.
Now look at AI. Instead of creating a massive surge of physical hiring hubs like the factory boom towns of the 19th century, AI is instead dissolving jobs and not replacing them. Rather than pulling workers into centralised physical hubs for new roles, AI models and agents can scale with near zero marginal labor costs, bypassing the need for a proportional human workforce. That's a massive difference. It's no where near comparable to the industrial revolution.
Team Andromeda
Member
You keep on blaming evil corps, not looking to accept they are only providing a servce billion of people want and use, whether you like it or not.
Also, I don't need to look at the Industrial Revolution, I can see a marked difference from the time of my Grandad to when I left school, to now.
I live in a rural village of just over 2,000 people in South Wales. In my grandad's day that same village had 4 working pits employing over 3,200 people, all gone by the time I was born.
When I left school, you didn't need qualifications you had a pick of factory work with the Hitachi factory alone, employing over 800 people. Today, all gone and most of the Japanese factories have pulled out of South Wales completely. It's so hard to get a job. That happened long before AI became a thing, was more to do with Labour bringing in the minimum wage & heavy industry being killed in the UK, like no doubt in other parts of the western world
Last edited:
IDKFA
I am Become Bilbo Baggins
I'm not calling them evil. Why are you so hell bent on defending them?
Why would these companies spend this sort of money, or why would investors front the cash, for people to use free AI tools? Where is the RoI? How is it being generated if the tools are free? Go back and read my original post and either address those points or move on.
From AI?
None of the above has anything to do with AI. You mention the working pits of Wales. In the same era my grandad worked in railway factory in Swindon that built trains. It closed for the same reason the pits were closed in Wales.
Thatcher.
But this has nothing to do with AI replacing jobs. AI creates massive jobs losses and doesn't replace them with any alternative. This is happening on a larger scale then anything else in history.
FingerBang
Member
Another great hate-thread full of economic illiterates. Can we do something about this shit? It's the equivalent of politics.
Last edited:
winjer
Gold Member
Another great hate-thread full of economic illiterates. Can we do something about this shit? It's the equivalent of politics.
You have to be extremely naive if you think this AI push is for the benefit of humankind, and not for the good of corporations and shareholder value.
Team Andromeda
Member
That's my point about AI. Millions of working-class people lost their jobs, nothing to do with AI and all to do with the outsourcing of heavy industry to Asia & far more use of robots so we could all buy cheap TVs & Microwaves. Even in the 90's we had robots more and more build our cars rather than it all being built by hand, like inthe Henry Ford days. The way you act, it's like now it's only an issue that millions of people will lose their jobs over AI
All that's to overlook how much people use AI, especially for videos & photos, since the modern nation is addicted to the likes of TikTok
Last edited:
IDKFA
I am Become Bilbo Baggins
The problem with white collar displacement is the sheer mismatch in velocity and scale. When a factory worker was displaced during in the past ,retraining took months, but they could find another role.
When a senior data analyst, copywriter, or mid level accountant loses their livelihood to an AI model, the market is flooded with thousands of peers experiencing the exact same shock, which in turn crashes the value of those specific skills. Also, the companies deploying these AI systems are doing so precisely to cut labor costs, meaning they are rarely hiring at a volume that matches what they are laying off.
Again, go back and read my original post. AI companies are not spending billions for people to use free AI tools. They're not making a RoI on free software. Sure, that's a nice thing to have, but that is certainly not the main reason so much is being spent on AI.
Bojji
Gold Member
Another great hate-thread full of economic illiterates. Can we do something about this shit? It's the equivalent of politics.
My country already tried (it was forced to use it) communism and reject it in 1989.
But unchecked capitalism is not much better for common folks. Corporations are ruining the world right now...
Team Andromeda
Member
My village has been decimated with the move from coal & heavy industry. I hate the move to cashless, but not many care, even if it comes at the cost of banks up and down the land closing, hell, even cash machines (ATMs) are getting harder to find now in my town.
And read what I said, If there wasn't demand, we wouldn't be seeing the move to AI, Millions/billions of people use AI on their phones; it's not all just about the evil tech world wanting no one to have a job. But the ever- increasing need for vast data centres so people can prat around on Tik-Tok, Snapchat & waste their day on Facebook.
You know I grew up in the era when you had to pay people to develop your own photos for you, The move to digital killed those jobs, hell my 1st job was as a paper boy. The move to digital killed that too. Let's hate on Digital, shall we?
Last edited:
Remember that a big part of AMD's Gaming Business is Radeon. And Real PC Gamers are boycotting RDNA4 (especially N44) for one reason or another and buying Blackwell instead. Resulting in a catastrophic all-time low share of 7% for AMD in 2025.
IDKFA
I am Become Bilbo Baggins
I don't think you're reading my posts because you're not addressing the points I raised, or just purposely ignoring them. Instead, you're just repeating the same things I've already addressed, or more making nonsensical comparisons.
Let's just agree to disagree and call it a day. Thanks for your time.