There may be a dip in prices for a bit, but since covid, more companies have realized they can get away with manufacturing fewer units and selling them for a higher price.
Unfortunately, this time around the majority of AI build up are GPUs that are likely difficult to accomodate in a random build.
If you want a GPU for graphics, well, many of them don’t even have video ports.
If your use case doesn’t need those, well, you might not be able to reasonably power and cool the sorts of chips that are being bought up.
The latest wrinkle is that a lot of that overbuying is likely to go towards Grace Blackwell, which is a standalone unit. Ironically despite being a product built around a GPU but needing a video port, their video port is driven by a non-nvidia chip.
gaming can work on one GPU and display on another, thunderbolt egpus do it all the time.
My use case is for my own AI plans as well as some other stuff like mass transcoding 200TB+ of…“Linux ISOs” lol. I already have power/cooling taken care of for the other servers I’m running
I’ve already got my gaming needs satisfied for years to come (probably)
I’ll buy a dozen and build my own AI farm tyvm.
it would be amazing to have several models running and have them interact together for some real immersive DnD rp.
something feels off with that picture
Yeah you don’t want those used GPU’s.
When the price of those drop, the price of the ones that werent used for that purpose will also drop
Why not, though? Does the silicium really age?
Supposedly they do but I’ve had surprisingly good luck with used GPUs from ebay. I’m good with warning others against buying used GPUs on ebay though because then costs will stay lower for me.
It depends on how they were deployed. For instance, GPU’s used for crypto mining can be fine unless they were really cranking them (which many do).
Ultimately unless you are buying used GPU’s From a reseller that has gone through them and either refurbished or just done extensive enough testing to show it doesn’t need to be refurbished, you’re taking a pretty big gamble
I thought the efficiency curve for GPUs peaked before 100%. If electricity is your primary cost, driving the GPUs at lower loads saves money.
So you might end up with GPUs that spent their entire life at a steady 80% load or something.
This was my understanding as well - that miners often underclock their GPUs rather than overclock them.
Most crypto mining outfits undervolt their cards for lower power usage. They aren’t cranking them as you say they are. A dead GPU doesn’t produce anything for you; cranking it up the chance that it will fail. You’re better off running it an extra 4 years at a lower voltage than you are cranking it for 1.
Yes.
They don’t exactly age, but top of line chips have very large currents in very small conductors. When you do that with DC current, your conductors deform with time, up to the point that they stop working correctly.
That said, you probably can get plenty of casual use out of them.
Oh but I do, ironically, for the same use cases LMAO. I like to tinker with AI and I like Microshits concept of Recall and similar ideas, like having an AI to be able to search through all my documents with nothing but a sentence or idea of what I’m looking for
But ain’t no fucking way I’m going to give a closed source AI that I’m not running myself that level of access
to be able to search through all my documents with nothing but a sentence or idea of what I’m looking for
Like this?
Makes sense!
Personally, I want to try to have a llama or something rewrite voice to text prompts into Home Assistant commands. Should be very cool.
But yeah, I won’t pay the current prices to run it either, nor use the cloud.
Have you seen the price of new GPUs? Sure ya do. Maybe they only last a few years. That’s alright.
Just deserts.
Just deserts.
We aint found shit
What did OP mean by that?
Onlyflans
Just desserts
Same goes for the death of windows 10. I want me some cheap Linux boxes.
The AI bubble burst is gonna wreck the market and send us into another recession. Especially if the crypto bubble bursts at a similar time.
Those in charge brought it on themselves. They can’t see through their own hype and lies and cult-like worship. It’s the rest of us that will hurt the most though.
It’ll be no worse than the dot-com burst.
That’s the best case scenario. I’m imagining something more like Enron or Bernie Madoff - losing all value at once when people realized it was a ponzi scheme.
Hard disagree.
The dot-com burst would be worst case scenario and I think that’s a long chance.
I understand that some people hate AI or that they feel threatened by it, but I can only imagine that those who say it has no real value are either being facetious or have extremely limited cognitive capacity.
I’m worried their operating costs are higher than they admit. They keep dumbing down the AI to make it use less energy. Right now it’s a free product that uses a crazy high amount of electricity. They don’t seem all that close to profitability.
AI is in it’s “biplane” stage, just as people in the twenties would have found it hard to comprehend the future of commercialised flight or a man on the moon, people struggle to imagine the future of AI.
AI is the very worse it’s going to be today, in the grand scheme of things, it’s only going to get better.
Stop. My penis can only get so erect.
You’re more likely to see the college bubble burst before AI. Colleges are struggling with low admissions.
I believe it is likely that there will be a burst at some point, just as with the dot-com burst.
But I think many people wrongly think that it will be the end of or a major setback for AI.
I see no reason why in twenty years AI won’t be as prevalent as “dot-com’s” are now.
I agree, history always repeats itself. But perhaps the timing is different, it could be 20 years, 10 years, or 50 years who knows
How much if a discount are you expecting to start gaming on a 30k card with no video output?
Not for gaming, for running AI open source models and other AI shenanigans. My 4080 Super has been filling my gaming needs and will for years to come, but it’s not enough for my AI interests lol
The most I can get out of this 4080 is running a ~7B param model, but I want to run cooler shit like that new open source DeepSeek v3 that dropped the other day.
So you’re waiting for the AI bubble to burst because you can’t wait to run all the cool new AI models?
Yea, the underlying tech is what interests me and I have a few potential use cases. Use cases that I would never entrust a random company with. For example, the concept of MS recall is cool, I’d never trust Microshits implementation though. But an open source local version that I’m in control of all the security implementations? Hell yea lol
That’s the problem. If the use case is super cool, and 99% of people have no knowledge (or motivation) to set it up for themselves, the online services will keep existing, and the bubble won’t really burst.
Even if some single companies fail (and they will, there some truly horrendous ideas getting funding), the big players will buy the GPUs wholesale before they hit ebay.
When the “Web” bubble burst, were we left without websites? No.
When the AI bubble bursts, it isn’t going away either. It replaces too much labor. Training the AI models is expensive, but running them isn’t. Especially on Lemmy, too many dullards don’t understand why AI is so popular; it’s because it’s replacing your job - or allowing 1 person to do the jobs of 10.
True, but many MANY companies went under during the dot com burst and made server hardware cheaper on the secondary market for a while
Same will happen with the AI bubble, there will undoubtedly be a few survivors, but many won’t. There will be dead startups everywhere.
When the AI bubble bursts, it isn’t going away either. It replaces too much labor.
The claimed amount of labour it could potentially replace is in fact part of the bubble.
The fact that you can claim this unironically, speaks volumes of your ignorance.
https://huggingface.co/models – Check out all of the things that various AI models can do. They aren’t just LLMs and Diffusors.
I never said that they can’t serve a function, just that corporations like openai have a vested interest inflating their potential impact on the job market.
Why the hell would you want to buy a used gpu that’s been burnt through its lifespan by an AI or crypto bro?
That’s a bad, horrible idea.
Won’t the price of all gpus go down when they’re not all being snatched up for mining?
Absolutely! New ones will be much cheaper, and that’ll be awesome.
But the used gpu market has been trash since crypto mining became a thing. Not worth the money to buy something that’s been worked to within an inch of its EOL.
I’ll never buy used tires, underwear or gpus haha
Cheap is cheap
For a lemon? Cheap is still too expensive.
Nuclear energy could also be a weird side effect of the bubble.
It’s both for me
