Everyone: things suck, better move my stuff on a small home server.
The hyper-scaler mafia: NOT ON MY WATCH!
The only silver lining is that newer devices will have to scale down memory, so developers will have to ditch memory-sucking frameworks and start to optimize things again.
Given this has been going on for years at this point, the high prices of graphics cards through crypto and now AI, it feels like this is the new normal, forever propped up by the next grift.
I don't think this ideology and investment strategy will survive this grift. There's too much geopolitical instability and investment restructuring for it to work again. Everyone is looking at isolationist policies. I mean mastercard/visa is even seen as a risk outside US now.
Well a risk has an abstract level and is either increasing or decreasing. You can look at your risk profile over time and work out how to define policy going forwards. It takes a long time to make changes at country level.
US is medium risk and increasing rapidly. Run away quickly.
> I’m just afraid that prices of $everything will go up soon and will not come down anymore, like they did after Covid.
That's how inflation works. In this case it seems more narrow though, there's hope the prices will go down. Especially if the AI hype finds a reason to flounder.
> I’m just afraid that prices of $everything will go up soon and will not come down anymore, like they did after Covid.
Just like the price of labour. Your salary went up and doesn't come down
In the UK weekly earnings increased 34% from December 2019 to December 2025.
CPI went up 30% in the same period.
Obviously that CPI covers things which went up more, and things which went up less, and your personal inflation will be different to everyone elses. Petrol prices end of Jan 2020 were 128p a litre, end of Jan 2025 they are 132p a litre [0]. Indeed petrol prices were 132p in January 2013. If you drive 40,000 miles a year you will thus see far lower inflation than someone who doesn't drive.
How many useless living humans do you know? They go somewhere. Something happens to them. Whatever it is it’s about to happen to 30% of the population.
Capitalism relies on consumers. It's pretty much central to the idea. That's why the world on average (median) has gotten wealthier and better off over the decades.
Do they? When I was at AWS in 2019-2021 there were some servers in the fleet— not many, but they definitely existed— which dated to before 2010. Maybe it depends on the type:class, but I think hyperscalers run hardware until it dies, then junk it. Where have you seen hyperscaler hardware for sale?
I've seen plenty of stuff from google and meta datacenters on the market.
Had some weirdo AMD GPUs that were GCP-only that were offered to folks if you bought in bulk quantity. A whole ton of whitebox Ethernet switches from Facebook were on the market a number of years ago.
Lots of weird stuff like OCP 25G NICs that only were bought in quantity by hyperscalers as well come out in waves. You don't see a lot of these things hit eBay and traditional "used marketplaces" but they are out there if you know where to look and have connections with liquidators. You're unlikely to get deals on a few here and there though unless the market becomes absolutely saturated - usually hundreds are MoQ for a lot of these things.
Storage will be unlikely to see as the risk:reward for a giant company usually is too great. If data was ever written to them, typically they will be shredded vs. secure wiped. Very few large corporations want to take the risk of some liquidator either not running a proper wipe process or something slipping through the cracks. Even with encrypted drives I haven't seen much in quantity hit unless they were quite literally never utilized and straight pulls from large JBOD orders or the like. However, ServerDirect and other places seem to have cracked this nut a little bit considering the number of refurb drives they run through - not sure what their sources are though. Certainly not hyperscaler levels in any case.
The main thing that the powers that be have always underestimated is the insane creativity the common people have when it comes to wanting things, but being forced to use alternative ways. Not going to say it won't suck, but interesting ways will indeed be found.
You’re going to find what, ways to make hand crafted survival RAM and drives in your backyard chip foundry?
Call me cynical if you like, but I don’t see this optimism that assumes the banal idea that somehow good always wins, when that’s simply not possible and in fact bad-guys have won many times before, it’s just that “dead men tell no tales” and the winners control what you think is reality.
The Chinese have end-to-end production capacity for lower capacity, lower performance/reliability consumer HDDs, so these are quite safe. Maybe we'll even see enterprise architectures where that cheap bottom-of-the-barrel stuff is used as opportunistic nearline storage, and then you have a far lower volume of traditional enterprise drives providing a "single source of truth" where needed.
In the same way that China is stepping into RAM production, I suspect they will step into the gap for high capacity drives as well. The market abhors a vacuum and China is eager to fill it, even at minimal levels of profit. Chinese manufacturers have become very good at providing an acceptable level of quality at a good price, maybe not the highest quality, but acceptable for consumer use.
I can understand that incumbents may not want to overinvest in capacity, which could be financially precarious, but they are also putting themselves in danger by opening up avenues for competition. One more thing ruined by AI mania, I suppose.
I come from a time when people had to use 1 to a maximum of 48 kilobytes for their entire computer. Later on i once went to Helsinki to watch with my own eyes what people can program in a restriction of 4 to 64 kilobytes. Computers have amazed me, but people who used them, have amazed me a lot of more times.
I wasn't saying it'll be good and that the good guys win, but a lot of insane creativity to circumvent restrictions will pop up.
> The only silver lining is that newer devices will have to scale down memory, so developers will have to ditch memory-sucking frameworks and start to optimize things again.
No. Prices will just go up, less innovation in general.
Saw this one coming and got my personal stuff out. It's running on an old Lenovo crate chucked in my hallway.
Work is fucked. 23TB of RAM online. Microservices FTW. Not. Each node has OS overhead. Each pod has language VM overhead. And the architecture can only cost more over time. On top of that "storage is cheap so we won't bother to delete anything". Stupid mentality across the board.
It is one tiny sliver of silver lining that “storage/memory/compute is cheap” nonsense that has produced all kinds of outsourced human slop code. That mentality is clearly going to have to die.
It could even become a kind of renaissance of efficient code… if there is any need for code at all.
The five guys left online might even get efficient and fast loading websites.
Honorable mention of the NO-TECH and LOW-TECH magazine site because I liked the effort at exploring efficient use of technology, e.g., their ~500KB solar powered site.
We went from using technology to solve problems to the diametric opposite of creating new problems to solve with technology. The latter will have to contract considerably. As you say, many problems can be solved without code. If they even need to be solved in the first place.
On the efficiency front, most of what we built is for developer efficiency rather than runtime efficiency. Also needs to stop.
I'm a big fan of low tech. I still write notes on paper and use a film camera. Thanks for the link - right up my street!
Unless people notice that they just built lots of useless datacenters and push back towards a mainframe + terminal setup, because ah sorry, modern software just runs much better that way, and you can save money on our inexpensive laptop with subscription model
> so developers will have to ditch memory-sucking frameworks
Since when have developers ever lowered hardware requirements? Prosumers will just cough up the extra money while casual users will continue to be left in the dust, like they have been for practically the last decade (or longer).
After, RAM, SSD, GPUs, now HDDs what else is there left to sell out? Power supplies, fans?
In a way this feels a bit absurd for these AI centers to hog HDDs.
As pointed by others neither training nor inference require HDDs and storing raw data should not require that much.
So my hypothesis is that it is a double whammy of overall declining consumer sided HDD demand, leaving data centers as main source of demand and additional demand from the new AI centers.
I feel like the AI centers are just buying HDDs because why not throw a HDD in each server blade even if there is no need? The money is there to be spent and it must be spent.
As someone who has been building computers since 1989 it feels like end of personal hobby casual building.
I will end with an imperfect analogy with multiplayer gaming. It is quite common in multiplayer games for higher level players to wish to acquire some tradeskill they neglected to acquire earlier. maybe a new quest appears, or new "must have" item that requires such skill.
They (past me included) have too much game money and no wish to acquire tradeskill items slowly. So the "rich" will overpay by 2x or 10x or even 100x the usual price.
That is free market at work right?
In the process whole low level economy is destroyed due to 2nd order effects.
Meaning a new player starting out can only be a farmer.
So if a student comes to me wishing to start building computers what advice do I give them? Farm something?
> As someone who has been building computers since 1989 it feels like end of personal hobby casual building.
We have a long way to go before the average PC costs even half as much as it did in 1989 (adjusted for inflation). And of course the performance for typical consumer use is orders of magnitude better than it was back then.
My parents love to tell me how in either late 1997 or early 1998 they bought the first PC for our family, a Compaq with a Pentium 3, 12 GB hard drive, 128MB of RAM, and no graphics acceleration at all beyond whatever was integrated on the Pentium 3. It cost $2000 back then so probably almost $4000 today. My high end 4090 rig cost a bit more than that to build for comparison and that machine is better than 98% of machines out there today.
>So if a student comes to me wishing to start building computers what advice do I give them? Farm something?
Buy used stuff? 99.9% of consumers have no need for anywhere near the cutting edge tech. I do far more than most people and get by just fine with a workstation I bought used in 2014. My newest Laptop is ~2018 and that was only because I wanted something with 4K that I could flip to tablet.
Raspberry PI's, SOCS, Microcontrollers, there's a million things today that are awesome. Are hobbyist students needing to build datacenters!?
Most of the computers I buy are refurbished or used models, I've never had a bad experience. Especially now, when computers are not getting much better whereas prices are increasing faster than performance.
What's next is no custom built PCs. They want us running dumb thin clients and subscribing to compute. Or it will be like phones. We'll get pre-built PCs that we aren't allowed to repair and they'll be forced to be obsolete every few years.
"they"? i see companies jacking their prices up, plain and simple. and us idiots still pay. ask yourself does intel no longer wish to sell CPUs to consumers? doesnt sound reasonable that intel would want to decimate their main market so AI companies can rule the world for some reason
If the custom PC market forcibly gets converted to pre-built PCs only, I don't see why intel is selling less CPUs. So I don't understand your point here.
he said thinpcs and dumb clients with the compute power in the cloud, not pre-built gaming pcs. intel sell both desktop and server, if you cut out desktop for all but the low powered i3 it takes to run a thinpc, they lose a massive chunk of revenue.
You act like companies do long term planning. They don't. Its always next quarter. And if you don't believe me, just look back to the pandemic, look at peleton, or car manufacturers or any of the dozens of examples
Do the guys that buy out the market have real use for all the hardware - or is it just hype? A solution against investors trying to corner the market would be to sell virtual hardware. Let them buy as much options on virtual "to be delivered" hardware" as they want. We also need an option market for virtual LLM-tokens, where the investors can put all their money without affecting real people.
It'll be fine. The supply chain for these components is inelastic, but that means once manufacturing capacity increases, it'll stay there. We'll see lower prices, especially if there is an AI crash and a mass hardware selloff like some people are predicting.
True if production capacity increases but it's an oligopoly and manufacturers are being very cautious because they don't want to cut into their margins. That's the problem with concentration. The market becomes ineffective for customers.
It's not about cutting in to their margins, if they end up scaling up production it will take several years and cost an untold amount of billions. When the AI bubble pops, if there's not replacement deman there's a very real chance of them going bankrupt.
> According to Mosley, Seagate is not expanding its production capacities for now. Growth is to come only from higher-capacity hard drives, not from additional unit numbers.
If it takes 2 years to increase, after 2 years everything will be thin clients already. Completely locked in, fully under control and everybody used to it. Very dystopian TBH.
No but you can switch and get comfortable with fully cloud based solutions when you your computer ages and the prices for the now one are through the roof.
Server grade hardware (rack blades) is already poor fit for consumer needs and AI dedicated hardware straight up requires external liquid cooling systems. It will be expensive to adopt them.
Strip it for usable parts (DRAM/VRAM/NANAD chips, maybe also some of the controllers), then recover any valuable metals (copper heat sinks, gold on contact pins), simple as that. :)
The number of HDDs sold has been in decline for over a decade. I doubt there is massive appetite for expanding production capacity
On the other hand the total storage capacity shipped each year has risen, as a combination of HDDs getting larger and larger, and demand shifting from smaller consumer HDDs to larger data center, enterprise and NAS HDDs. I'm not sure how flexible those production lines are, but maybe the reaction will be shifting even more capacity to higher-capacity drives with cutting-edge technology
I already see the ‘save us China’ meme comments, but I want to go a step further and ask: Did they ever develop domestic HDD production (like of the tooling needed to make HDD’s)?
As I understand it, the technology holders for this specific line of technology are Korea (Samsung), Japan (Toshiba), and US (WD/Seagate).
I don’t think China ever developed the tooling internally? Am I wrong?
This feels like the ballpoint pen situation. Where China could build everything but the ballpoint since it took specialized machinery. But HDD gets to be a hard sell with how far NAND has come, and China could just lean on economies of scale here and skip the hard work of what at some point could become overcomplicated abandonware technology. Similar to approach between internal combustion cars and EV’s…
EDIT: And I am very aware production happens in places like Thailand, I don’t forget the flood pricing, I also don’t forget clamcoin shortages. Specifically mean where the core expertise is held.
That's the electronics industry in general though. The shortages are real and a normal part of growing pains for any industry that's so capital-intensive and capacity constrained.
I'll go against the grain and claim this might be a good thing long term. Yes, it sucks also, I was planning to expand my NAS but guess I'll figure out how to compress stuff instead.
Which goes into why I think this might be good. Developers have kind of treated disks as "oh well" with binaries ballooning in size, even when it can easily solved, and there is little care to make things lightweight. Just like I now figure out a different solution to recover space, I'm hoping with a shortage this kind of thing will be more widespread, and we'll end up with smaller things until the shortage is over. "Necessity is the mother of all invention" or however it goes.
If anything the ecosystem is moving away from that, not into that. People seem more and more averse of relying on SaaS when local applications can do, but this might also be somewhat of a bubble I'm perceiving, rather than a global trend I suppose.
It just goes to show how totally corporations have captured western aligned governments. Our governments are powerless to do anything (aside from some baby steps from the EU).
China is now the only solution to fix broken western controlled markets.
Cynicism in unhelpful, and it's not correct. This has nothing to do with governments and everything to do with market economics. This sort of thing happens every few years in computing.
Then what's your solution? It's a capital intensive business to get into, without some regulatory changes either on the supply side or the demand side, there's no way for it to be naturally prevented without consumers bearing the brunt of the downside. Yes, the market will eventually correct itself but until then consumers suffer.
The only legal check for monopoly corporations is regulation/taxation. That doesn’t work cross border. Especially when the other side has nationalised and artificially props the monopoly.
The solution then is removing the product from market till local competition takes its place.
Looks like we need a computer hardware reserves the same way there are regional reserves for food, fuels and other critical commodities?
And for the same reason - to avoid the dominant players going "oh shiny" on short term lucrative adventures or outright trying to manipulate the market - causing people to starve and making society grind to a halt.
The real "computer hardware reserves" is the used market. Average folks and smaller businesses will realize that their old gear now has resale value and a lot more of it will be entering the resale/refurbishment market instead of being thrown away as e-waste.
Still, at least in the short to medium term, there are many companies, institutions and individuals that are used to getting new hardware that is under warranty - developer hardware refreshes, store hardware updates and deployments, some schools and academic institutions that provide hardware to students or even just people getting a laptop for schools use.
I think people would be used to getting new hardware for all of these & might be pretty much left out cold without a machine or having to opt for a reused or resold machine without the necessary technical knowledge to support it themselves.
Repairability, upgradability and standards compliance needs to be minimum in consumer products. No to proprietary connectors. No soldered SSD or RAM. For home use, allow relaxed licensing options for discarded enterprise products like switches, Wifi Access Points etc. (Juniper Mist APs are fantastic, but are a brick without their cloud).
Currently, I cannot put in a market bought SSD in my Macbook. I cannot put in a SSD in my Unifi router without buying their $20 SSD tray. I cannot put third party ECC-RAM and SSDs in my Synology NAS because the policy has only been lifted on HDDs but nothing else.
I fear opposite will happen. Only leveraged companies have access to DRAM and NAND and hence will use it to lock us into their ecosystem as consumers won't even get access to storage in the open market otherwise.
I picked up a few hundred TB from a chia farm sale. Glad for it. I think I'm set for a while. Honestly, the second they started buying this stuff I started buying hardware. The only problem for me is that they're even ruining the market for RTX 6000 Pro Blackwells.
Honestly looks highly suspicious to me. Because ok they might need some big storage like petabits. But how can this be a match in proportion with the capacity that is currently usually needed for everything that is hard drive hungry. Any cloud service, any storage service, all the storage needed for private photo/video/media storage for everything that is produced everyday, for all consumer hardwares like computers...
Gpu I understand but hard drive looks excessive. It's like if tomorrow there is a shortage of computer cabling because ai datacenter needs some.
If you're building for future training needs and not just present, it makes more sense. Scaling laws say the more data you have, the smarter and more knowledgeable your AI model gets in the end. So that extra storage can be quite valuable.
There are many new data-centers they are being filled with servers. Most servers have at least 2 HDD (mirror) for the OS. I would not be surprised if on a huge scale even 2 HDD per server could cause HDD shortage.
There are likely models which are trained on 4k video and it should be stored somewhere too.
Even things like logs and metrics can consume petabytes for a large (and complex) cluster. And the less mature the software the more logs you need to debug it in production.
In the AI race investments if not unlimited at least abundant. In such conditions optimization of hardware usage is the waste of time and velocity is the only things which matters.
So at least for HDDs it is unlikely they will have super custom "AI HDDs" like thay have with custom High Bandwidth Memory displacing regular RAM in fab output & being harder to reuse later.
So hopefully once they all go bust, there should be a lot of cheap enterprise HDDs on the marked as creditors pick through the wreckage. :)
If component prices keep going up and the respective monopoly/duopoly/triopoly for each component colludes to keep prices high/supply constrained, then eventually devices will become too expensive for the average consumer. So what’s the game plan here? Are companies planning to let users lease a device from them? Worth noting that Sony already lets you do this with a ps5. Sounds like we’re headed towards a “you will own nothing and be happy” type situation
"You will use AI, because that will be the only way you will have a relaxed life. You will pay for it, own nothing and be content. Nobody cares if you are happy or not."
We could also vote the policians protecting these uncompetitive markets out of power and let regulators do their job. There has been too many mergers in the component market.
You also have to look at the current status of the market. The level of investment in data centers spurred by AI are unlikely to last unless massive gains materialize. It's pretty clear some manufacturers are betting things will cool down and don't want to overcommit.
> We could also vote the policians protecting these uncompetitive markets out of power and let regulators do their job.
Could we though? Even if gerrymandering and voter suppression weren't already out of control, and getting worse, there are very few politicians who could or would do anything about all this.
I feel traditional "rust" hard disks would be inefficient for AI use. Unless they include SSDs (which I feel these data centers are more likely to be using) in the definition as well...
In a large rack-mount SAN with hundreds or thousands of disks connected to your computing equipment via fiberchannel, the performance difference is pretty negligible.
The real advantage of SSDs in this use case is storage density and power efficiency. On the other hand, your compute resources might be packed in so tight with power-intensive stuff that you appreciate the spinning rust “wasting” space.
The TV is best device for unleashing your creativity by upvoting your favourite Sora creators! Become an expert at any field by activating premium prompts from our partners! By connecting camera you can have meaningful motivating discussions with your deceased loved ones (camera required for fraud prevention. Remember, not looking at the screen during ads is punishable by law)
You have 31/999 credits remaining. What activity would you like to productively participate in today?
Source: I've been using Chrome's built in translation heavily since around 2018-19,
and not before that only because that's the time I've realized it'd became god damn good.
GPT-2 was in 2019. That was really the start of the LLM bandwagon. You could argue it was ChatGPT that brought it to the masses, but that was still only in 2022. 3 years is not a long time.
Yes I argue exactly that. Notice how I've avoided saying LLM or GPT and said "AI hype". Everyone here on HN has had heard of GPT back then, yet we've been through like what, 3, 4, 5? different hypes since 2018, AI hype is only being the latest.
Maybe to clarify my argument a little bit more, "AI" was already good and was being used for useful/productive purposes before it became the hype. It'd be better had it stay the same way instead of it being the latest money generation machine and sucking the air out of the room.
Everyone: things suck, better move my stuff on a small home server. The hyper-scaler mafia: NOT ON MY WATCH!
The only silver lining is that newer devices will have to scale down memory, so developers will have to ditch memory-sucking frameworks and start to optimize things again.
I’m just afraid that prices of $everything will go up soon and will not come down anymore, like they did after Covid.
If it’s temporary I can live with it.
I guess this was inevitable with the absolute insane money being poured into AI.
>If it’s temporary I can live with it.
Given this has been going on for years at this point, the high prices of graphics cards through crypto and now AI, it feels like this is the new normal, forever propped up by the next grift.
I don't think this ideology and investment strategy will survive this grift. There's too much geopolitical instability and investment restructuring for it to work again. Everyone is looking at isolationist policies. I mean mastercard/visa is even seen as a risk outside US now.
Yup, when you can’t trust partners (or even nominal allies), what else is there but isolationism?
cooperation.
Sure you have to isolate certain rogue states - North Korea, Russia, USA. Always the way.
It's not really isolation but exclusion. Push all risks as far away from you as possible.
When everything ‘outside’ is a risk, what would you call a summary of that policy?
Well a risk has an abstract level and is either increasing or decreasing. You can look at your risk profile over time and work out how to define policy going forwards. It takes a long time to make changes at country level.
US is medium risk and increasing rapidly. Run away quickly.
> I don't think this ideology and investment strategy will survive this grift
Big tech will be deemed "too big to fail" and will get a bail out. The tax payers will suffer.
Big tech has already failed. Which is why it got into politics.
Traps tend to only go one way.
> I’m just afraid that prices of $everything will go up soon and will not come down anymore, like they did after Covid.
That's how inflation works. In this case it seems more narrow though, there's hope the prices will go down. Especially if the AI hype finds a reason to flounder.
> I’m just afraid that prices of $everything will go up soon and will not come down anymore, like they did after Covid.
Just like the price of labour. Your salary went up and doesn't come down
In the UK weekly earnings increased 34% from December 2019 to December 2025.
CPI went up 30% in the same period.
Obviously that CPI covers things which went up more, and things which went up less, and your personal inflation will be different to everyone elses. Petrol prices end of Jan 2020 were 128p a litre, end of Jan 2025 they are 132p a litre [0]. Indeed petrol prices were 132p in January 2013. If you drive 40,000 miles a year you will thus see far lower inflation than someone who doesn't drive.
[0] https://www.rac.co.uk/drive/advice/fuel-watch/
Amazing how many people on the internet refuse to accept basic facts
We useless eaters are to be priced out of life soon enough.
Traditionally that hasn't gone well for the rich folk.
It also hasn't gone well for the non-rich folk either.
Truer than even you dare to admit.
How many useless living humans do you know? They go somewhere. Something happens to them. Whatever it is it’s about to happen to 30% of the population.
What’s the opposite of survivor bias?
It takes a profound lack of empathy to refer to your neighbors as "useless living humans"
Furthermore, it's usually just plain dangerous.
It is meant in the capitalist sense. Your horror at the statement is the point.
Capitalism relies on consumers. It's pretty much central to the idea. That's why the world on average (median) has gotten wealthier and better off over the decades.
Capitalism relies on the circulation of capital. Consumers are an anomaly.
Citation: nvidia/openai circular loans network.
> How many useless living humans do you know?
Oh, I can think of about 77 million right off the top of my head.
I am afraid my NAS will have a hard drive failure, and I won’t be able to order replacements. I should’ve bought a back up
Exactly, me too and at 10 years it's gonna happen :(
> If it’s temporary I can live with it. I guess this was inevitable with the absolute insane money being poured into AI.
Hyperscalars refresh hardware and firesale old stock.
~2028 is going to see a lot of high power refurb supply hit the market.
Do they? When I was at AWS in 2019-2021 there were some servers in the fleet— not many, but they definitely existed— which dated to before 2010. Maybe it depends on the type:class, but I think hyperscalers run hardware until it dies, then junk it. Where have you seen hyperscaler hardware for sale?
Non-standard racks (Meta?) started showing up on the aftermarket around the time that Meta and a few other folks were the only ones using them.
Forget the details, as it's been a decade+, but everything gets junked at some point.
They're almost always stripped down, then sold as board+chassis only.
I've seen plenty of stuff from google and meta datacenters on the market.
Had some weirdo AMD GPUs that were GCP-only that were offered to folks if you bought in bulk quantity. A whole ton of whitebox Ethernet switches from Facebook were on the market a number of years ago.
Lots of weird stuff like OCP 25G NICs that only were bought in quantity by hyperscalers as well come out in waves. You don't see a lot of these things hit eBay and traditional "used marketplaces" but they are out there if you know where to look and have connections with liquidators. You're unlikely to get deals on a few here and there though unless the market becomes absolutely saturated - usually hundreds are MoQ for a lot of these things.
Storage will be unlikely to see as the risk:reward for a giant company usually is too great. If data was ever written to them, typically they will be shredded vs. secure wiped. Very few large corporations want to take the risk of some liquidator either not running a proper wipe process or something slipping through the cracks. Even with encrypted drives I haven't seen much in quantity hit unless they were quite literally never utilized and straight pulls from large JBOD orders or the like. However, ServerDirect and other places seem to have cracked this nut a little bit considering the number of refurb drives they run through - not sure what their sources are though. Certainly not hyperscaler levels in any case.
Many prices didn't really even come down much after COVID.
GPU prices never went back to normal. Harddrives neither, I bought 14TB drives 10 years ago for €200, they've never been that low again.
The main thing that the powers that be have always underestimated is the insane creativity the common people have when it comes to wanting things, but being forced to use alternative ways. Not going to say it won't suck, but interesting ways will indeed be found.
You’re going to find what, ways to make hand crafted survival RAM and drives in your backyard chip foundry?
Call me cynical if you like, but I don’t see this optimism that assumes the banal idea that somehow good always wins, when that’s simply not possible and in fact bad-guys have won many times before, it’s just that “dead men tell no tales” and the winners control what you think is reality.
One way of putting it, is the winners are ‘the good guys’.
People will find a way to not need as much RAM, and thus the devices that require it.
Same way the price of groceries going up means people buy only what they need and ditch the superfluous.
The Chinese have end-to-end production capacity for lower capacity, lower performance/reliability consumer HDDs, so these are quite safe. Maybe we'll even see enterprise architectures where that cheap bottom-of-the-barrel stuff is used as opportunistic nearline storage, and then you have a far lower volume of traditional enterprise drives providing a "single source of truth" where needed.
In the same way that China is stepping into RAM production, I suspect they will step into the gap for high capacity drives as well. The market abhors a vacuum and China is eager to fill it, even at minimal levels of profit. Chinese manufacturers have become very good at providing an acceptable level of quality at a good price, maybe not the highest quality, but acceptable for consumer use.
I can understand that incumbents may not want to overinvest in capacity, which could be financially precarious, but they are also putting themselves in danger by opening up avenues for competition. One more thing ruined by AI mania, I suppose.
I come from a time when people had to use 1 to a maximum of 48 kilobytes for their entire computer. Later on i once went to Helsinki to watch with my own eyes what people can program in a restriction of 4 to 64 kilobytes. Computers have amazed me, but people who used them, have amazed me a lot of more times.
I wasn't saying it'll be good and that the good guys win, but a lot of insane creativity to circumvent restrictions will pop up.
> Later on i once went to Helsinki to watch with my own eyes what people can program in a restriction of 4 to 64 kilobytes.
The Demoscene will save us all
> The only silver lining is that newer devices will have to scale down memory, so developers will have to ditch memory-sucking frameworks and start to optimize things again.
No. Prices will just go up, less innovation in general.
A few places will have no choice - low price elasticity, combined with things that need to actually work.
Saw this one coming and got my personal stuff out. It's running on an old Lenovo crate chucked in my hallway.
Work is fucked. 23TB of RAM online. Microservices FTW. Not. Each node has OS overhead. Each pod has language VM overhead. And the architecture can only cost more over time. On top of that "storage is cheap so we won't bother to delete anything". Stupid mentality across the board.
It is one tiny sliver of silver lining that “storage/memory/compute is cheap” nonsense that has produced all kinds of outsourced human slop code. That mentality is clearly going to have to die.
It could even become a kind of renaissance of efficient code… if there is any need for code at all.
The five guys left online might even get efficient and fast loading websites.
Honorable mention of the NO-TECH and LOW-TECH magazine site because I liked the effort at exploring efficient use of technology, e.g., their ~500KB solar powered site.
https://solar.lowtechmagazine.com/about/
I think your ideological perspective is spot on.
We went from using technology to solve problems to the diametric opposite of creating new problems to solve with technology. The latter will have to contract considerably. As you say, many problems can be solved without code. If they even need to be solved in the first place.
On the efficiency front, most of what we built is for developer efficiency rather than runtime efficiency. Also needs to stop.
I'm a big fan of low tech. I still write notes on paper and use a film camera. Thanks for the link - right up my street!
Unless people notice that they just built lots of useless datacenters and push back towards a mainframe + terminal setup, because ah sorry, modern software just runs much better that way, and you can save money on our inexpensive laptop with subscription model
At least we can add "use the least amount of RAM and drive space" to our AI prompts.
/s
Well you can do that, but then the AI won't be nearly as smart as it was before...
> so developers will have to ditch memory-sucking frameworks
Since when have developers ever lowered hardware requirements? Prosumers will just cough up the extra money while casual users will continue to be left in the dust, like they have been for practically the last decade (or longer).
>so developers will have to ditch memory-sucking frameworks and start to optimize things again.
Can you DM me your contact details? I have a nice shiny new bridge that I can get you a great deal on.
After, RAM, SSD, GPUs, now HDDs what else is there left to sell out? Power supplies, fans?
In a way this feels a bit absurd for these AI centers to hog HDDs.
As pointed by others neither training nor inference require HDDs and storing raw data should not require that much.
So my hypothesis is that it is a double whammy of overall declining consumer sided HDD demand, leaving data centers as main source of demand and additional demand from the new AI centers.
I feel like the AI centers are just buying HDDs because why not throw a HDD in each server blade even if there is no need? The money is there to be spent and it must be spent.
As someone who has been building computers since 1989 it feels like end of personal hobby casual building.
I will end with an imperfect analogy with multiplayer gaming. It is quite common in multiplayer games for higher level players to wish to acquire some tradeskill they neglected to acquire earlier. maybe a new quest appears, or new "must have" item that requires such skill.
They (past me included) have too much game money and no wish to acquire tradeskill items slowly. So the "rich" will overpay by 2x or 10x or even 100x the usual price.
That is free market at work right?
In the process whole low level economy is destroyed due to 2nd order effects. Meaning a new player starting out can only be a farmer.
So if a student comes to me wishing to start building computers what advice do I give them? Farm something?
> As someone who has been building computers since 1989 it feels like end of personal hobby casual building.
We have a long way to go before the average PC costs even half as much as it did in 1989 (adjusted for inflation). And of course the performance for typical consumer use is orders of magnitude better than it was back then.
My parents love to tell me how in either late 1997 or early 1998 they bought the first PC for our family, a Compaq with a Pentium 3, 12 GB hard drive, 128MB of RAM, and no graphics acceleration at all beyond whatever was integrated on the Pentium 3. It cost $2000 back then so probably almost $4000 today. My high end 4090 rig cost a bit more than that to build for comparison and that machine is better than 98% of machines out there today.
CPUs: https://www.tomshardware.com/pc-components/cpus/intel-amd-se...
> Power supplies, fans?
https://www.reddit.com/r/hardware/comments/1q9zd0f/power_sup...
Gas turbines
>So if a student comes to me wishing to start building computers what advice do I give them? Farm something?
Buy used stuff? 99.9% of consumers have no need for anywhere near the cutting edge tech. I do far more than most people and get by just fine with a workstation I bought used in 2014. My newest Laptop is ~2018 and that was only because I wanted something with 4K that I could flip to tablet.
Raspberry PI's, SOCS, Microcontrollers, there's a million things today that are awesome. Are hobbyist students needing to build datacenters!?
Most of the computers I buy are refurbished or used models, I've never had a bad experience. Especially now, when computers are not getting much better whereas prices are increasing faster than performance.
Damn. First GPUs, then RAM, now hard drives?
What's next, the great CPU shortage of 2026?
yes https://www.reuters.com/world/china/intel-amd-notify-custome...
Oh no, looks like my 8700k will have to hold out a little longer.
Better start hoarding Silica.
I think hard drives was before RAM but it kind of all happened contemporaneously.
What's next is no custom built PCs. They want us running dumb thin clients and subscribing to compute. Or it will be like phones. We'll get pre-built PCs that we aren't allowed to repair and they'll be forced to be obsolete every few years.
"they"? i see companies jacking their prices up, plain and simple. and us idiots still pay. ask yourself does intel no longer wish to sell CPUs to consumers? doesnt sound reasonable that intel would want to decimate their main market so AI companies can rule the world for some reason
Why doesn't that sound reasonable, when nvidia did exactly that?
because nvidia owns half of its own suppliers AND customers. intel completely missed out on the AI gravy train
If the custom PC market forcibly gets converted to pre-built PCs only, I don't see why intel is selling less CPUs. So I don't understand your point here.
he said thinpcs and dumb clients with the compute power in the cloud, not pre-built gaming pcs. intel sell both desktop and server, if you cut out desktop for all but the low powered i3 it takes to run a thinpc, they lose a massive chunk of revenue.
You act like companies do long term planning. They don't. Its always next quarter. And if you don't believe me, just look back to the pandemic, look at peleton, or car manufacturers or any of the dozens of examples
a lot of effort goes into each cpu gen, often takes more than a quarter
Do the guys that buy out the market have real use for all the hardware - or is it just hype? A solution against investors trying to corner the market would be to sell virtual hardware. Let them buy as much options on virtual "to be delivered" hardware" as they want. We also need an option market for virtual LLM-tokens, where the investors can put all their money without affecting real people.
Sam Altman (just yesterday): Codex weekly users have more than tripled since the beginning of the year!
https://twitter.com/sama/status/2023233085509410833
Funny how they all meet at the "You'll own nothing" Swiss club.
Next on the data centers' menu: jet engines https://www.wsj.com/business/energy-oil/how-jet-engines-are-...
It'll be fine. The supply chain for these components is inelastic, but that means once manufacturing capacity increases, it'll stay there. We'll see lower prices, especially if there is an AI crash and a mass hardware selloff like some people are predicting.
True if production capacity increases but it's an oligopoly and manufacturers are being very cautious because they don't want to cut into their margins. That's the problem with concentration. The market becomes ineffective for customers.
It's not about cutting in to their margins, if they end up scaling up production it will take several years and cost an untold amount of billions. When the AI bubble pops, if there's not replacement deman there's a very real chance of them going bankrupt.
> According to Mosley, Seagate is not expanding its production capacities for now. Growth is to come only from higher-capacity hard drives, not from additional unit numbers.
If it takes 2 years to increase, after 2 years everything will be thin clients already. Completely locked in, fully under control and everybody used to it. Very dystopian TBH.
You're going to throw out all your computers in two years?
No but you can switch and get comfortable with fully cloud based solutions when you your computer ages and the prices for the now one are through the roof.
Server grade hardware (rack blades) is already poor fit for consumer needs and AI dedicated hardware straight up requires external liquid cooling systems. It will be expensive to adopt them.
Strip it for usable parts (DRAM/VRAM/NANAD chips, maybe also some of the controllers), then recover any valuable metals (copper heat sinks, gold on contact pins), simple as that. :)
The number of HDDs sold has been in decline for over a decade. I doubt there is massive appetite for expanding production capacity
On the other hand the total storage capacity shipped each year has risen, as a combination of HDDs getting larger and larger, and demand shifting from smaller consumer HDDs to larger data center, enterprise and NAS HDDs. I'm not sure how flexible those production lines are, but maybe the reaction will be shifting even more capacity to higher-capacity drives with cutting-edge technology
Future show HN: how I managed to parallelize 100 tape drives to load windows and play video games
VHS cassettes: maybe not so obsolete, after all?
Also, the Return of PKZIP.
Stacker and Doublespace!
memmaker and qemm
Video Backup System for Amiga!
You could do worse... https://youtu.be/1hc52_PWeU8
Somebody once commented on social media that cryptocurrency feels like a dropshipped alien technology designed to deplete a civilization's resources.
Now that crypto has failed to paperclip maximize the world, maybe LLMs are attempt number two.
I already see the ‘save us China’ meme comments, but I want to go a step further and ask: Did they ever develop domestic HDD production (like of the tooling needed to make HDD’s)?
As I understand it, the technology holders for this specific line of technology are Korea (Samsung), Japan (Toshiba), and US (WD/Seagate).
I don’t think China ever developed the tooling internally? Am I wrong?
This feels like the ballpoint pen situation. Where China could build everything but the ballpoint since it took specialized machinery. But HDD gets to be a hard sell with how far NAND has come, and China could just lean on economies of scale here and skip the hard work of what at some point could become overcomplicated abandonware technology. Similar to approach between internal combustion cars and EV’s…
EDIT: And I am very aware production happens in places like Thailand, I don’t forget the flood pricing, I also don’t forget clamcoin shortages. Specifically mean where the core expertise is held.
There hasn't been a better time in the past 15 years to push for a new video or image codec. Saving storage Space is important again.
This is assuming most of what we stored are either images or video.
New video codecs needs hardware support for good performance means you need to buy new hardware. Not a good time for that.
No one can be surprised to see that all of these artificial "shortages" are impacting components with monopoly or few actors producers...
That's the electronics industry in general though. The shortages are real and a normal part of growing pains for any industry that's so capital-intensive and capacity constrained.
I'll go against the grain and claim this might be a good thing long term. Yes, it sucks also, I was planning to expand my NAS but guess I'll figure out how to compress stuff instead.
Which goes into why I think this might be good. Developers have kind of treated disks as "oh well" with binaries ballooning in size, even when it can easily solved, and there is little care to make things lightweight. Just like I now figure out a different solution to recover space, I'm hoping with a shortage this kind of thing will be more widespread, and we'll end up with smaller things until the shortage is over. "Necessity is the mother of all invention" or however it goes.
Think that's more of a "silver lining" instead of the overall trend being a "good thing long term." It's still pretty terrible.
Yeah, true, that's more aligned with what I was thinking and more accurate. Thanks :)
There is an increasing chance the "invention" will be that nobody owns personal computers, and now has to rent from the cloud.
If anything the ecosystem is moving away from that, not into that. People seem more and more averse of relying on SaaS when local applications can do, but this might also be somewhat of a bubble I'm perceiving, rather than a global trend I suppose.
Save us, China.
It just goes to show how totally corporations have captured western aligned governments. Our governments are powerless to do anything (aside from some baby steps from the EU).
China is now the only solution to fix broken western controlled markets.
Cynicism in unhelpful, and it's not correct. This has nothing to do with governments and everything to do with market economics. This sort of thing happens every few years in computing.
Then what's your solution? It's a capital intensive business to get into, without some regulatory changes either on the supply side or the demand side, there's no way for it to be naturally prevented without consumers bearing the brunt of the downside. Yes, the market will eventually correct itself but until then consumers suffer.
The only legal check for monopoly corporations is regulation/taxation. That doesn’t work cross border. Especially when the other side has nationalised and artificially props the monopoly.
The solution then is removing the product from market till local competition takes its place.
Looks like we need a computer hardware reserves the same way there are regional reserves for food, fuels and other critical commodities?
And for the same reason - to avoid the dominant players going "oh shiny" on short term lucrative adventures or outright trying to manipulate the market - causing people to starve and making society grind to a halt.
The real "computer hardware reserves" is the used market. Average folks and smaller businesses will realize that their old gear now has resale value and a lot more of it will be entering the resale/refurbishment market instead of being thrown away as e-waste.
That is a very good point!
Still, at least in the short to medium term, there are many companies, institutions and individuals that are used to getting new hardware that is under warranty - developer hardware refreshes, store hardware updates and deployments, some schools and academic institutions that provide hardware to students or even just people getting a laptop for schools use.
I think people would be used to getting new hardware for all of these & might be pretty much left out cold without a machine or having to opt for a reused or resold machine without the necessary technical knowledge to support it themselves.
Repairability, upgradability and standards compliance needs to be minimum in consumer products. No to proprietary connectors. No soldered SSD or RAM. For home use, allow relaxed licensing options for discarded enterprise products like switches, Wifi Access Points etc. (Juniper Mist APs are fantastic, but are a brick without their cloud). Currently, I cannot put in a market bought SSD in my Macbook. I cannot put in a SSD in my Unifi router without buying their $20 SSD tray. I cannot put third party ECC-RAM and SSDs in my Synology NAS because the policy has only been lifted on HDDs but nothing else. I fear opposite will happen. Only leveraged companies have access to DRAM and NAND and hence will use it to lock us into their ecosystem as consumers won't even get access to storage in the open market otherwise.
I picked up a few hundred TB from a chia farm sale. Glad for it. I think I'm set for a while. Honestly, the second they started buying this stuff I started buying hardware. The only problem for me is that they're even ruining the market for RTX 6000 Pro Blackwells.
I'm confused, that doesn't make sense to me:
> They largely come from hyperscalers who want hard drives for their AI data centers, for example to store training data on them.
What type of training data? LLMs need relatively little of that. For example, DeepSeek-V3 [1], still a relatively large model:
> We pre-train DeepSeek-V3 on 14.8 trillion diverse and high-quality tokens
At 2 bytes per token, that's 29.6 terabytes. That's basically nothing compared to the amount of 4K content that is uploaded to YouTube every day.
1: https://arxiv.org/html/2412.19437v1
You may have answered your own question if they're wanting to train models on video and other media.
So Veo (Google), Sora (OpenAI) or Seedance (ByteDance)?
Honestly looks highly suspicious to me. Because ok they might need some big storage like petabits. But how can this be a match in proportion with the capacity that is currently usually needed for everything that is hard drive hungry. Any cloud service, any storage service, all the storage needed for private photo/video/media storage for everything that is produced everyday, for all consumer hardwares like computers...
Gpu I understand but hard drive looks excessive. It's like if tomorrow there is a shortage of computer cabling because ai datacenter needs some.
If you're building for future training needs and not just present, it makes more sense. Scaling laws say the more data you have, the smarter and more knowledgeable your AI model gets in the end. So that extra storage can be quite valuable.
If you’re building a text-only model then the storage is limited but once you get to things like video then it’ll explode exponentially
A few random thoughts:
There are many new data-centers they are being filled with servers. Most servers have at least 2 HDD (mirror) for the OS. I would not be surprised if on a huge scale even 2 HDD per server could cause HDD shortage.
There are likely models which are trained on 4k video and it should be stored somewhere too.
Even things like logs and metrics can consume petabytes for a large (and complex) cluster. And the less mature the software the more logs you need to debug it in production.
In the AI race investments if not unlimited at least abundant. In such conditions optimization of hardware usage is the waste of time and velocity is the only things which matters.
So at least for HDDs it is unlikely they will have super custom "AI HDDs" like thay have with custom High Bandwidth Memory displacing regular RAM in fab output & being harder to reuse later.
So hopefully once they all go bust, there should be a lot of cheap enterprise HDDs on the marked as creditors pick through the wreckage. :)
Previously:
Thanks a lot, AI: Hard drives are sold out for the year, says WD
https://news.ycombinator.com/item?id=47034192
It's not clear if they are going to redistribute production lines in favor of hyper-scalers - are we going to see fewer SATA drives on the market?
Time for some heavy regulation
That's not going to happen when AI is already propping up a significant chunk of the economy.
There is appetite in some circles for a consumer boycott but not much coordination on targets.
> There is appetite in some circles for a consumer boycott
its not being used anywhere lol where are they meant to boycott?
/dev/null as a service, mooning
Big tech buying up HDDs in order to force the consumer base into cloud computing and storage.
I guess this is the real way that AI will destroy humanity. By sucking up everything and making it expensive.
[SHOW HN]: I built a self-replicating LLM that runs off of common grocery store items and alcohol.
Supply of 2nd hand enterprise stuff is also showing a slowdown. Seeing less of it show up in eBay
If component prices keep going up and the respective monopoly/duopoly/triopoly for each component colludes to keep prices high/supply constrained, then eventually devices will become too expensive for the average consumer. So what’s the game plan here? Are companies planning to let users lease a device from them? Worth noting that Sony already lets you do this with a ps5. Sounds like we’re headed towards a “you will own nothing and be happy” type situation
> Sounds like we’re headed towards a “you will own nothing and be happy” type situation
That's when I sell of my current hardware and house, buy a cow and some land somewhere in the boondocks and become a hermit.
It could be a level up from that.
"You will use AI, because that will be the only way you will have a relaxed life. You will pay for it, own nothing and be content. Nobody cares if you are happy or not."
We could also vote the policians protecting these uncompetitive markets out of power and let regulators do their job. There has been too many mergers in the component market.
You also have to look at the current status of the market. The level of investment in data centers spurred by AI are unlikely to last unless massive gains materialize. It's pretty clear some manufacturers are betting things will cool down and don't want to overcommit.
> We could also vote the policians protecting these uncompetitive markets out of power and let regulators do their job.
Could we though? Even if gerrymandering and voter suppression weren't already out of control, and getting worse, there are very few politicians who could or would do anything about all this.
Elections aren't going to save us.
First they came for the GPUs, but I did not speak out, for I was not a gamer.
Then they came for the RAM, but I did not speak out, for I had already closed Firefox.
Then they came for the hard drives, but I did not speak out, for I had the cloud.
Then my NAS died, and there was no drive left to restore from backup.
I hope the data centres burn
Then prices will definitely go up more.
Then more burning data centres will occur
Doesn't sound like that will help anything.
they’re pushing for AI, but nobody will have a device to use it?
I feel traditional "rust" hard disks would be inefficient for AI use. Unless they include SSDs (which I feel these data centers are more likely to be using) in the definition as well...
They need it hoard datasets.
In a large rack-mount SAN with hundreds or thousands of disks connected to your computing equipment via fiberchannel, the performance difference is pretty negligible.
The real advantage of SSDs in this use case is storage density and power efficiency. On the other hand, your compute resources might be packed in so tight with power-intensive stuff that you appreciate the spinning rust “wasting” space.
Chromebook with a 64gig shitty eMMC is what Google and friends would love you to use. Pay that cloud drive subscription!
The TV is best device for unleashing your creativity by upvoting your favourite Sora creators! Become an expert at any field by activating premium prompts from our partners! By connecting camera you can have meaningful motivating discussions with your deceased loved ones (camera required for fraud prevention. Remember, not looking at the screen during ads is punishable by law)
You have 31/999 credits remaining. What activity would you like to productively participate in today?
But hey, we get slop videos of the pope doing something funny, that's just as cool as being able to purchase computer hardware, right?
To be fair, heise is a german news site, and the very article is auto-translated by AI from its german counterpart.
AI use for translation is a good fit for the tech. The problem is generative AI.
Uhm, kind of, if you don't mind random stuff getting inserted into your translation sometimes.
(Source: this has actually happened to me.)
Machine translation have become pretty good long before AI hype started.
Not really.
Really.
Source: I've been using Chrome's built in translation heavily since around 2018-19, and not before that only because that's the time I've realized it'd became god damn good.
GPT-2 was in 2019. That was really the start of the LLM bandwagon. You could argue it was ChatGPT that brought it to the masses, but that was still only in 2022. 3 years is not a long time.
Yes I argue exactly that. Notice how I've avoided saying LLM or GPT and said "AI hype". Everyone here on HN has had heard of GPT back then, yet we've been through like what, 3, 4, 5? different hypes since 2018, AI hype is only being the latest.
Maybe to clarify my argument a little bit more, "AI" was already good and was being used for useful/productive purposes before it became the hype. It'd be better had it stay the same way instead of it being the latest money generation machine and sucking the air out of the room.
We can also lose our jobs in a few years so there’s that