They are selling on EBay for over $20k, used.
The 0-reputation account in Spain selling an M3U 512GB for $4200 is 100% fraud.
And the fact of the matter is that in 2026, all electronics has gone up, not down, and sought-after GPUs have gone up in price in the used market.
Computers depreciate because they are obviously being supplanted by newer better models—until they become vintage and then move into collectibles.
Doing this particular one is definitely expecting the market squeeze to continue. "Worst case" is back to more "normal" depreciation. Where I'd expect to only be able to recoup more like 18k. But... if you look at GPU prices the last 3 years... it's not a crazy assumption that it won't drop that fast.
iPhone example since those are easiest to find in quantity: new iPhone 16 Pro Max for $1200, Gazelle would want $866 for "execllent" condition. Lost ~28% for one-model-back. iPhone 15 Pro Max, though: excellent priced at $667 here, only down another 23%, and gives you basically half-priced-upgrade if you can sell it for that and roll into the newest.
So to have never-more-than-one-model-old rough estimate at today's value-holding you'd be out $3600 for three new phones, with getting 1732 of that back, or 1868 for it (with a $334-per-year incremental cost of upgrade).
For never-more-than-two-models-back you'd be out $2400, getting back $866, for net $1534 spend, with a $167 incremental per-year upgrade cost once you buy the first one. Pretty good if you keep the phone in excellent condition and are happy to budget a bit over $10/month to be on a every-two-year upgrade train.
Well, you'd also eat the tax...
https://buy.gazelle.com/products/iphone-16-pro-max-256gb-unl...
https://buy.gazelle.com/products/iphone-15-pro-max-256gb-unl...
I would absolutely not count on that, if and when it drops it will drop hard.
We aren't exactly in "standard" times and haven't been for quite a while. Even five year old graphics cards are worth more today than they were just a year ago. Things will obviously depreciate at some point, but you gotta throw your existing notions of how quickly and how much hardware will depreciate out the window. There's just been too much money dumped into AI for a "well I guess this won't ever pan out, let's dump all this hardware to recoup our costs" moment to happen and tank the price of everything suddenly IMO.
And that's not even getting into the other geopolitical stuff going on right now. Strange times.
If you are able to tie up $25k for a few years just for shiggles, you clearly are able to make do fine without that money and if lost it would be at worst annoying, not catastrophic.
"The DGX GB200 NVL72 AI server costs approximately $3 million per unit. This system includes 72 Blackwell GPUs and 36 Grace CPUs, making it one of the most powerful AI servers available."
The search assist actually credited a source used with: https://www.tweaktown.com/news/98292/nvidias-new-gb200-super...
That $25k spend by GGGP seems like nothing in comparison. That's ~1/3 of one chip in that cabinet. God gawd I'm old and out of touch with modern AI data centers.
We've been in a centralised phase for longer than usual - first cloud everything, then AI - but at some point in the next decade prices will crash and a market will appear for personal, local intelligence.
There are bigger data centers than Colossus 1 around too.
There is a reason NVidia is the most valuable company on the planet.
https://en.wikipedia.org/wiki/Colossus_(supercomputer)#Curre...
A better way of putting it is that you can run plenty of things on a single ordinary system, but you may be disappointed at the performance. Generally, you can't expect inference to be as quick as with cloud for SOTA-like models. You have to run smaller models for quick replies, and large models with a lot of real-world knowledge for less time-critical inference, possibly batching many requests simultaneously to improve throughput.
Yes this is exactly what I'm doing. I isolated the actual math question, and then sent it to my two servers to process and that's what's taking 10m+ to return. I'm asking them to solve the question and return the full answer along with their steps. I care about correctness so taking time is okay but I can't use 10m per solution.
if some have more than one layer it could fewer but that's the order of magnitude
This is taking a hobby to its extremes, in much the same way that a $5k boat and $500k boat let you catch the same fish.
Remember: one year showed up to be a gigantic leap in regards to quality of results and innovation in the AI space. Agents weren't really a thing and vibe coding wasn't even invented as a term because the top notch tools at the time were lousy, with lovable being the frontrunner with its - in my view - sorry Tailwind recombination tool shaming AI to do the work.
Then fall hit 2025 hit us, new year's eve and suddenly there was such a massive surge of innovation and competition with ChatGPT Codex suddenly showing up.
Remember: one year ago many now commonly used tools weren't yet available like Nano Banana or Codex.
"The 25k are so vast" - Yes, and no. For example, if the machine is bought for business usage I can deduct the costs from taxes. This roughly amount for 50% of the financial burden.
So I jokingly use to say, that I pay only half the price for my Apple business machines. And yes, I am strict in this regard. Business means business. No private emails etc. nothing on my company computers.
Maybe there are other options as well to reduce the financial expenses the dude mentions, but it doesn't seem so.
I would also go for leasing, this way already the monthly payments can be deduced and I don't need to buy and maybe resell the machine.
Apple is a luxury good. Without business usage or at least partly using it for business as well as private (mixed usage in tax reports) I wouldn't buy the devices or think twice.
Apple under Cook evolved into a Gucci like luxury brand, that is more and more a rip off than quality delivered, especially considering the latest OS updates for Mac, iOS and iPad. Apple is a mess, following Microsoft Windows' footsteps happily, because the CEO is as has been correctly assessed, no product guy.
But I stop with my rant here.
Always try to use tax deduction as leverage for your computer expenses. Every citizen should invest in basic knowledge about that.
Even a 10-20% professional usage for work (mixed usage) gives you a noticeable advantage over normal pay.
It's not financially a good idea: renting really does beat owning, and cloud beats both if you're only running inference on these machines. But I'm not just doing inference, and as a thing I can do silly stuff on to learn, it's hard to beat!
I do still use Vast and Runpod for things too, but it’s much nicer to test a fine tuning run here to make sure I’m in the ballpark
I also did literally say “It's not financially a good idea, renting is better than owning” so I’m confused why I have two people telling me that
Also it’s just far more fun to play with something tangible to me :)
It’s also annoying because then I need to make sure my little “lab” setup is well automated, and I’m lazy :)
Also, I literally said “ It's not financially a good idea” so I’m confused why you think I don’t know that.
That money could have been spent on way more bang/buck performance in the form of a set of 4 graphics cards.
Also I would probably put the odds 70:30 that Apple marketing is astroturfing on HN from the amount of posts about running llms on Macbooks, because in reality, the inference speed of any decent llm is unusable on a Macbook despite the ability to fit it into RAM.
If you like having a box with 8-12 fans blasting hot air and noise into your office all day, nobody's stopping you.
the 40-80 tok/sec is only for initial prompt processing, and with the "medium" models, like Qwen3.6:27b. The actual token generation is in the 10 token/second Thats very slow. And your Macbook pro will stop being a LAP-top, because it will get very warm.
Meanwhile, my 2x3090s happily crank out ~100 tok/sec generation. Oh and I can run 100 tok/sec on my phone as well, because I can just access ollama on my home desktop over ssh from termux.
I saw your heat comments about the RTX 6000 Pro as well. I bought a few of them recently and I'm running 2 of them in a 2U case in a colo. You need a lot of active airflow to keep them cool. Mine range from 23 C to 80 C.
After my last run, I'm going to wait for the new case I ordered to come in and cannibalize my kid's PC that we built beginning of this year to form an entirely separate computer. And then figure out better ways to deal with the heat, especially with summer coming up. I'll have to play around with undervolting and running vents directly outside my house to see if that helps.
Used to overclock back in the day during winter with an intake duct rigged to suck in outside air, best thing about -30c :)
Not consistently, I did start using petroleum jelly till I upgraded and found out that wasn't very fun to clean up.
But the trend here is interesting. I think by 2030 you'll be able to buy fairly cheap hardware that is currently $10k+. I don't know what this does to the trillions invested in AI data centers because the next NVidia architecture after Blackwell will essentially half the value of purchased cards overnight.
I'm not convinced Apple has yet pivoted the Mac Studio line towards this market and the expected M5 Ultras in Q3 2026 will likely be an incremental improvement rather than big leap forward but I'd like to be proven wrong.
I feel that the open weight models pale in comparison to the frontier models, and I believe that if the gap closes quickly, that the open weight vendors will stop releasing it for free.
Higher radiation, space insulations, etc.
Underwater data centers provide a lot of the same benefits and can (much more) easily be hauled to the surface
Point being, once the M5 Ultra is available, I suspect a lot of people will get very serious about making Macs work with RTX GPUs because that will yield an inference platform with a good bang:buck ratio. If so, you may find that your existing hardware is more powerful than it seems today. And it may be a lot more expensive to replace later if you sell it now.
An RTX 6000 pro Blackwell is a pretty good card
Unless you’re doing others things with that gpu such as gaming, video, video AI etc.
This is a real problem and why I've just about given up on ebay or fb marketplace, esp for computers. If you are in Canada though sellit9.com is a great solution to having to deal with sketchy buyers.
YMMV but between your nearest PD office and Library, you should be able to use one or the other for your exchange of goods/money. The biggest thing I've sold is a mid-range video card during late covid (I managed to get a better one via newegg shuffle) so I sold the old one (RX 5700XT -> RTX 2080) to make up the difference a bit. I just did the exchange at the Starbucks near me for that.
Police "safe trade zones" are basically a parking space outside a police station, with a sign.
The buyer doesn't know who the seller is, and vice-versa... the level of trust you can bear depends on how much you're willing to lose. My advice is only in that there are safe venues you can use to make such an exchange.
If it is an electronic payment, I'm not sure how completing the transaction in front of a police station will help any. Well, it will help the buyer to see it working, but the seller gets no additional protection besides seeing "a person."
That's not the point of going to the "police safe exchange zone".
The point is to hopefully prevent the possibility of the buyer showing up with a .38 in hand, and demanding to be given the easy to fence "item" unless the seller wants to get a .38 slug embedded in their gut.
The risk of a "hold up" increases with dollar value and with items that are easier to fence.
The problem is that while one these gpus is a huge improvement over a laptop or a single 3090, you very quickly wish you had more. I would buy a second one, but I did the math and realized that with the current crop of models, 2 Blackwells doesn't buy me any new capability that I didn't have with one. So I would need a 3rd one. And when I buy a 3rd one I will feel like I want to running a higher quant, so then I will want a 4th.
Also, the 4-bit quants of MiniMax 2.7 will run at 100 tps or so with two cards, which is pretty decent. It doesn't go any faster at all with 4 GPUs from what I've seen, so if you don't actively need 384 GB of VRAM, 2x RTX6000 is a good place to be.
It's going to be a non-trivial haircut. This stuff depreciates pretty fast.
Of course, this is an unusual state of affairs; I see my GPU purchase as consumption, not investment.
The two major drivers of inference costs are GPUs and electricity. You can't get cheaper GPUs, but you can make existing GPUs not sit idle, and you do that by utilizing them 24/7, processing user B's request when user A is thinking, and handling many requests in parallel, neither of which you can do as an individual. You can get cheaper electricity... by moving, and it's much easier to move your AI workload than to move yourself.
This is a completely different dynamic than renting houses or apartments, as you can't really rent out the same house to different people at different times of day.
I'm very pro local models, but not to have parity with SoTA frontier models. Just contextually trained small models doing smaller specific tasks.
Trying to run bigger LLMs for an individual user to do big tasks is not going to be a good time.
Idk what people were thinking. I’ve never seen anyone offer a plausible way to sidestep batch processing for example.
A GPU is much like a hotel or an airplane. To make hotel and airplane economics work out, you need to keep your seats / rooms filled at all times. Because the number of spots is limited, an empty seat, room or request slot today is an opportunity forever wasted.
Unlike hotels and airplanes, though, GPU requests are short-lived. This means you need somebody to provide you with useful work to do 24/7. If you're an individual who needs to sleep or a single-country business that works 9 to 5, most of your request processing opportunities will go to waste. You're much better off renting your GPU from somebody who can provide it with useful work 24/7, and hence get much better per-request prices.
This is the type of place one might be “waiting for the other shoe to drop.” Which carries a variety of potential meanings in this moment of AI.
Tangentially related: Mack and the boys lived in the “Palace Flophouse and Grill” in Cannery Row.
I suppose I must have looked up flophouse when reading all the Steinbeck I could get my hands on and it’s stuck w me.
You'll realize real quick its not profitible. You cant just say things you don't like to hear are unsubstantiated without verifying.
Not to mention, subscriptions.. $2mm in GPUs being given out for 5 hrs a day at a cost of $200 a month.
I could easily say that everyone who says its profitible is msking unsubstantiated claims lol.
Shouldn't we compare the API pricing, where we pay per token? The whole point of local inference is that we don't have any restrictions regarding product use or time limits, so it would only be fair if we compare it to a plan that offers the same. And even that is only a first approximation, because the commercial models are usually much more capable than the open weight models.
Yes, once you have modeled the problem correctly and you know all the input parameters. This is not that: Session# * tps * 86400 (secs in a day) * 30 days.
I don't think there is enough public information to check Anthropic's claims regarding inference profitability. It depends not just on unknown technical factors but also on agreements they have with other companies.
Its possible you could pay off hardware for Kimi 2.6 after maybe 2-3 yrs (by providing low tps / high concurrency) but you're now out of warranty and have been running your machines full throttle 24/7 for 2-3 years.
This is why moonshot attempted to double the price when they released 2.6 but then it got driven down by North American capital subsidies.
If the companies as a whole are destined to be profitable, or worth their valuations is a very different question. The only people who can truely answer that have time machines.
These 1T param models running at <$3.00 per 1mm are certainly not profitable.
And people who don't understand the difference between capex and opex are making uneducated claims. It's not basic math.
Running an inference data center is a mix of variable and fixed costs. The fixed costs are currently in the billions of billions of dollars for pretty much any investment in this space. Many of those fixed costs have (currently) unknown refresh cycles. So, unless you have access to the financial books of these companies it's currently just speculation whether inference is profitable.
As long as the power users are paying per token, everything is good.
Anthropic told the Department of War-nee-Defence that they'd made $5bln total, which is a lot LOT less than what they're spending.
We'll see what's in OpenAi's IPO later this year I guess. I'll be very surprised if they're losing less that $100bln a year.
Batch size is what you should look at. If a cluster is running and processing one request, filling the batch has almost no marginal cost (kv cache creation/storage/fetch costs aside). But if the concurrent requests exceed batch size, one extra request would cost basically the rent cost of entire new cluster. APIs have the bursty nature so companies would plan to price it such that they are profitable / break even at 40%-50% utilization (% of filled batch for simplicity). so any extra request would not have the same cost as long as they are alongside an api request. you might think it degrades teh performance. easy: just assign a priority tier to api requests, and a lower tier to subscription requests.
its even more effective and powerful now that you have continuous batching. so likely if the api is being used, they are not eating any loss, let alone "big loss"
You absolutely can. OpenAI et al are paying a fortune for GPUs but they are not paying retail prices.
The entire business model of retail is to sell above cost.
> if I can find a reliable place to sell it without getting ripped off by scammers.
I don't follow this last part. What is the scam they try to run?For something listed at $25k I would not list on eBay at all. eBay corporate will pocket $3400 in fees and will also dock you local taxes on the $25k.
Slow? Yes, but ... private. Unconditionally.
And recently with https://github.com/antirez/ds4 one can use just one system to a very, very decent speed and ttft for chat inference. Again, private.
In comparison to just spending for tokens, the tokens would have been much cheaper and much much faster. I've been running against Gemma4:31b, Qwen3.5 and 3.6, and getting local LLMs to solve AMC 8/10 math questions and it's about 10-100x slower than just doing it online. When I tried it with ChatGPT late last year, it took about one night and $25 to solve about 1000 questions. Using my RTX 6000 and M3 Ultra and Gemma4:31b on both, it answered about 40 questions in 7 hours and I haven't checked how good the answer is yet. At 800 watts (600 for RTX and 200 for M3 Ultra) and running for 7 hours, it solved around 40 questions.
At the very least I'm going to try to sell my M3 Ultra if I can find a reliable place to sell it without getting ripped off by scammers.