[HN Gopher] Hard drives on backorder for two years as AI data ce...
___________________________________________________________________
Hard drives on backorder for two years as AI data centers trigger
HDD shortage
Author : pabs3
Score : 132 points
Date : 2025-11-12 05:36 UTC (17 hours ago)
(HTM) web link (www.tomshardware.com)
(TXT) w3m dump (www.tomshardware.com)
| mock-possum wrote:
| > delivery times for enterprise-grade HDDs delayed by two years.
|
| I sleep
|
| > so hyperscalers are now switching to QLC NAND-based SSDs to
| avoid these backorders ... This could lead to SSD prices rising
| worldwide
|
| Real shit
| BoredPositron wrote:
| The funniest thing about this is that, with high GPU prices,
| rising RAM costs, and now increasing SSD prices, Apple will end
| up producing the most affordable PCs.
| ipsum2 wrote:
| Apple uses the same RAM, SSD, etc as everyone else does. They
| don't have a magic supply chain that is unaffected by the
| broader world.
| BoredPositron wrote:
| They don't use the same SSDs? They don't use the same RAM?
| They have their own supply chain in place? Whatcha talking
| about bud?
| Incipient wrote:
| They use the same suppliers. The problem is the base
| chip, and also the wafer itself, all of which will impact
| apple.
|
| (apple doesn't use hdds so not talking about that here).
| BoredPositron wrote:
| The problem is not chip supply it's manufacturing. Apple
| has their own manufacturing suppply. This is not the chip
| crisis of the last years. Hyper scalers are switching to
| consumer hardware because there is nothing in storage for
| Prosumer anymore and the manufacturing pipelines for
| these are smaller and harder to scale than consumer ones.
| appointment wrote:
| Appple uses custom DRAM and NAND chips?
| jeroenhd wrote:
| The NAND itself seems to come from companies like SK
| Hynix, which are currently struggling to meet demand.
|
| Apple doesn't make their own NAND, just like they don't
| make their own screens. They did write their own NAND
| controller and stuck it into their CPUs, but the flash
| memory that actually stores data doesn't come directly
| from their factories.
| BoredPositron wrote:
| It's not a fucking nand problem. The producers reduced
| capacity because of the slow consumer PC market in 2024.
| With hyperscalers shifting to consumer products the
| market is strained but everyone is scaling up their
| production now. This is not the same as the chip shortage
| we had in the last few years. The production capabilities
| are there they are just adjusting to new market
| realities. We will see a whiplash for a few months but we
| can and will produce enough chips for everyone which we
| were not able to do in the 20-22 crisis. Stop getting
| your "insights" from influencers and stop parroting them.
| Incipient wrote:
| It's ABSOLUTELY a chip shortage. TSMC is booked out years
| in advance and jacking prices _because they can_.
|
| On the dram side, because that's what I saw recently, sk
| hynix is down to TWO WEEKS of inventory:
| https://wccftech.com/sk-hynix-ddr5-inventory-down-to-
| just-2-...
|
| I'd be curious what evidence there is to support its
| purely an end-product manufacturing shortage?
| BoredPositron wrote:
| I don't see any evidence of a general chip shortage in
| your link either. That they are down in stock because of
| unprecedented demand after a downturn in the last 3 years
| in consumer demands of consumer products is something
| completely different than being down because of
| production capacity. They reduced production in 2023 and
| 2024. It's exactly what I wrote in my previous comment.
| microtherion wrote:
| Yes, but Apple tends to lock in their supply with long
| term contracts and prepayments, so they often are
| protected from supply shocks.
| siva7 wrote:
| They have a magic supply chain that is unaffected by the
| broader world which is one of the reasons why Tim Apple was
| chosen by Steve Jobs as his successor.
| jrvarela56 wrote:
| If every other PC is more expensive, they will just increase
| prices.
| HackerNewt-doms wrote:
| No, Apple has effectively promoted iCloud as the alternative
| to local storage as part of its product differentiation
| strategy in the lower price segment.
|
| Apple will almost certainly introduce the same approach for
| the budget MacBook as well.
| margalabargala wrote:
| This is a long term good thing.
|
| It sucks right now and will probably suck through 2027.
|
| By 2028 or so we'll have a 50% drop in price-per-storage for
| these components.
| andy_ppp wrote:
| Will the AI bubble last until 2028? I'm still unclear how AI
| will return even 10% of this investment in profit.
| citrin_ru wrote:
| Depending on the future you predict 10% may be a good ROI -
| if AI will replace humans and traditional economy will
| collapse all other investments will loos value even more. In
| such scenario you cannot save the money you only can loose
| less if you will make a right investment.
| adrianN wrote:
| I'm not sure I understand why the economy would collapse if
| AI replaced humans. Wouldn't the companies just make more
| profit because they save labor costs and stonks go up?
| netdevphoenix wrote:
| The economy is a macrosystem. For a company to make
| profit, someone needs to buy. For someone to buy a
| service or a product, they need to have the money needed
| to buy. If AI replaces humans and UBI doesn't happen,
| costs might drop to near zero but so will revenue as
| virtually no one will be able to buy anything. So the
| economy will certainly collapse.
| adrianN wrote:
| It depends on how many jobs AI replaces and which. It is
| unlikely that LLMs will replaces plumbers any time soon
| for example. Many technological inventions have replaced
| humans in lots of jobs, the economy has only grown since
| then.
| netdevphoenix wrote:
| >Many technological inventions have replaced humans in
| lots of jobs, the economy has only grown since then.
|
| Indeed it has, but many of those it replaced did not
| necessarily get moved to better positions if they were
| moved at all. The industrial revolution was great if you
| were young and had little to no responsibilities. Your
| average middle aged farmer with a large family to feed
| faced poverty or near poverty and those who were able to
| move to the cities faced worse labour conditions. People
| on here love talking about the winners while conveniently
| ignoring the loser side.
|
| And that's not even mentioning the skilling up issue to
| which older workers are more vulnerable. Who will fund
| the nation wide skill up costs for them?
| rwyinuse wrote:
| Industrial revolution was all right as long as you owned
| the farm, as most farmers did. The losers were those farm
| workers who weren't needed in as great numbers anymore,
| and they weren't paid well in the first place. I agree on
| labour conditions though, there industrialisation was a
| clear downgrade for many.
|
| What worries me about AI is how it's not obvious which
| new jobs it may create. Younger farm workers and children
| of farmers could just move and work in factories, where
| the employers mostly took care of their training. I can't
| see such opportunities here. I believe whole world needs
| either UBI, or governments paying for jobs that
| previously weren't seen worth paying for. Otherwise the
| economy will collapse due to mass unemployment and
| resulting lack of demand.
| andy_ppp wrote:
| No I'm predicting a 90%+ loss.
| chrismorgan wrote:
| > _By 2028 or so we 'll have a 50% drop in price-per-storage
| for these components._
|
| Do you mean relative to six months ago, or now? Because a lot
| of the prices have already more than doubled.
|
| (I'm upset because the computer I've been planning to build,
| which three months ago would have come to around [?]90,000, is
| now up to [?]1,20,000 and climbing week by week, half due to
| price increases on the same part, half due to forced
| substitutions on RAM since the cheaper 32GB 6400MT/s DDR5
| sticks are completely unavailable. And looking into laptops,
| for the first time _ever_ I'm seeing manufacturet SODIMM or SSD
| upgrades being cheaper than aftermarket.)
| walterbell wrote:
| _> manufacturer SODIMM or SSD upgrades being cheaper than
| aftermarket._
|
| Temporarily thanks to old stock.
| palmotea wrote:
| > This is a long term good thing. ... By 2028 or so we'll have
| a 50% drop in price-per-storage for these components.
|
| Per the op:
|
| > and the ongoing DRAM shortage is proof of this, with memory
| kits costing more than double what they did just a few months
| ago.
|
| > While enterprise-grade QLC SSDs would entirely power this
| pivot, Sandisk has already raised NAND prices by 50%, according
| to another DigiTimes report, after initially warning of a 10%
| increase two months ago.
|
| So you're basically saying prices may return to normal in two
| years, and that's somehow a good thing compared to them not
| being inflated in the first place?
| karlkloss wrote:
| And when the AI bubble bursts, "refurbished" HDDs and GPUs will
| flood the market. Save your money now and be prepared.
| esseph wrote:
| If this bubble pops you might need that money for food when
| bananas go from $1.50 to $150.00
| Nux wrote:
| GPUs yes, but there'll be no HDDs making it alive, they'll get
| destroyed to protect whatever rubbish they had on.
| HugoTea wrote:
| Depends how it goes down, if a company goes into insolvency
| all security policies are off the table and random hardware
| can get shifted into lot bidding.
| pmontra wrote:
| HDD can be written multiple times with random data if data
| centers really have to protect what their former customers
| wrote on them. I never looked at those details in standard
| contracts.
|
| There is also encryption at rest.
| teeray wrote:
| > HDD can be written multiple times with random data
|
| Which costs more in compute than simply throwing the drive
| in a shredder
| walterbell wrote:
| What's the disposal method for shredded drives?
| faeyanpiraat wrote:
| Mount Doom
| rwyinuse wrote:
| Not really, if we give the HDD some resale value. There's
| a market for used but functional hard drives.
| tencentshill wrote:
| They have enough investor money they don't need to recoup
| it selling used drives. Straight to the shredder.
| jimwalsh wrote:
| All the large datacenter/cloud companies do not let hard
| drives leave the building.
| SoftTalker wrote:
| All you really need to do is write one pass of zeros on
| them. That will prevent anyone but a very dedicated
| adversary with expensive equipment from recovering any
| data, especially on TB scale drives.
|
| Can still take hours per drive though, which is why a lot
| of people skip it.
| edoceo wrote:
| I make a random 1MB chunk, then write that all over the
| drive, at overlapping offsets. I've been told that really
| clears it. On IDE-spinning-rust disks I trusted it, not
| sure if I should trust these modern SSD
| p1mrx wrote:
| Why rewrite the same 1MB chunk, instead of making new
| random chunks?
|
| Redundant data at least opens the possibility that the
| drive could deduplicate.
| krior wrote:
| Or encrypt it and just trash the encryption header.
| archagon wrote:
| I believe many enterprise drives have instant-erase
| functionality (presumably deleting an encryption key).
| SoftTalker wrote:
| If they were encrypted to begin with, yes. Many are not.
| archagon wrote:
| I was under the impression that these drives may be
| transparently encrypted by default. (Rollable encryption
| key in hardware, invisible to end-user.)
| bigbuppo wrote:
| Nah, the liquidators aren't going to care about that. Those
| hard drives are going to be shipped out with all your wildest
| porn chat bot fantasies.
| maerF0x0 wrote:
| "Shredded onsite" means by the next user when they format
| the drive and write contents to it /s
| microtherion wrote:
| My understanding is that refurbished GPUs from e.g. crypto
| mining are something people are warned away from, because
| they've often been run into the ground. Are AI usage practices
| different?
| trenchpilgrim wrote:
| > My understanding is that refurbished GPUs from e.g. crypto
| mining are something people are warned away from, because
| they've often been run into the ground.
|
| No, this wasn't the case. While there were never
| comprehensive studies various tech media purchased these
| cards to run testing and found that, other than scammers,
| they all performed to expectation.
| lm28469 wrote:
| You can buy used car tires with 1% of thread left and
| they'll perform amazingly during your one time test too
| rwyinuse wrote:
| Car tire is a moving part that wears in use. Only moving
| parts a GPU has are its fans, which can be replaced. The
| rest may last really long, or fail brand new.
| SoftTalker wrote:
| Thermal cycling will eventually cause problems with
| continuity but I'm not sure that's really predictable. We
| have some really old GPUs that eventually had problems
| under load, we attributed it to thermal degradation as we
| had ruled out pretty much everything else.
| lm28469 wrote:
| Believe it or not but the "really long time" you're
| mentioning is much less "long" when you run the card 24/7
| at 100% in enclosed spaces
|
| Electricity is a moving part btw, electronic components
| do wear and tear
| mschuster91 wrote:
| > And when the AI bubble bursts, "refurbished" HDDs and GPUs
| will flood the market.
|
| GPUs? No way. The datacenter cards don't even have video output
| ports, and I think the chips destined for AI / ML training also
| have everything video/render related removed from the silicon,
| makes for more yield.
|
| And the other way around, using (cheap) consumer GPUs in
| servers, I think at least NVDA tries to prevent that with
| driver-based DRM, so there won't be any flooding coming from
| there either.
| krackers wrote:
| >Picking QLC over TLC allows them to maintain costs while
| achieving sufficient endurance for cold storage.
|
| How does that work, doesn't QLC have less write endurance?
| ycombinete wrote:
| Yes, but QLC has much higher density.
|
| I think it's the higher density that makes it better for cold
| storage, which generally has infrequent access, and more reads
| than writes.
|
| Hence the QLC's endurance being " _sufficient_ for cold storage
| ".
| esseph wrote:
| Cold storage normally doesn't have frequent writes or frequent
| reads.
| Havoc wrote:
| Ssd for cold storage seems like an odd choice in itself. If
| that's genuinely done due to availability then we really are in
| for a wild ride
| bayindirh wrote:
| I believe "cold storage" in this parlance is more like "read-
| oriented" rather than being accessed once in three years.
| bayindirh wrote:
| In short: Aggressive overprovisioning.
|
| Enterprise SSDs are not expensive only because they have better
| flash chips, but they have much more of them.
|
| A top of the line write oriented SSD comes with 4-7x more
| capacity than what it says on the tin, but that extra capacity
| is used for cell replacement rather than capacity itself.
|
| Mixed use comes with 2-4x overprovisioning, and read oriented
| is around 2x IIRC.
| aaronax wrote:
| I find these numbers to be way outside of what I have heard
| of. I would be surprised if you could give an example that
| comes with even 1.5x capacity. (4TB capacity, 6TB actual
| flash on chips, for example.)
| mrandish wrote:
| I _really_ hope the AI bubble bursts sooner rather than later.
| Sooner will impact the broad economy less severely (although it
| 'll still be pretty bad) and curtail these supply chain
| shortages. If the bubble keeps inflating, the storage makers will
| have already mostly built out excess capacity and the crash will
| lead to even longer-term supply distortion.
| walterbell wrote:
| "OMEC" (Organization of Memory Exporting Countries) NAND
| production quotas?
|
| https://x.com/jukanlosreve/status/1988505115339436423
| Samsung Electronics has lowered its target for NAND wafer output
| this year to around 4.72 million sheets, about 7% down from the
| previous year's 5.07 million. Kioxia also adjusted its output
| from 4.80 million last year to 4.69 million this year.. SK hynix
| and Micron are likewise keeping output conservatively constrained
| in a bid to benefit from higher prices. SK hynix's NAND output
| fell about 10%, from 2.01 million sheets last year to around 1.80
| million this year. Micron's situation is similar: it is
| maintaining production at Fab 7 in Singapore--its largest NAND
| production base--in the low 300,000-sheet range, keeping a
| conservative supply posture.
|
| Micron has a new US fab coming online in 2027, which should
| improve supply.
|
| DRAM price fixing scandal: 1998-2002, 2016-2018,
| https://en.wikipedia.org/wiki/DRAM_price_fixing_scandal
| acd10j wrote:
| It's a cartel just like the OPEC oil cartel.
| Eddy_Viscosity2 wrote:
| Is there anything that isn't functionally a cartel at this
| point?
| Pet_Ant wrote:
| Yes, anywhere there is a race to the bottom. That is why
| you see cartels rise up, like unions are cartels. Things
| that are not cartels generally have ever diminishing
| margins.
| Eddy_Viscosity2 wrote:
| I see the rising of cartels more just the inevitable
| consequence of market systems where some players just
| getting bigger and bigger. Then when only a few large
| players are left, they collude (either openly, or just by
| unspoken agreement). So less of a race to the bottom, but
| a race to the top.
| Pet_Ant wrote:
| I think we agree. To risk being trampled by a race to the
| bottom, they choose to rise above by forming a cartel.
| mchannon wrote:
| During 1998-2000, AOL was ordering so many free trial CDs that it
| locked up world production, and music CDs faced 8-12 week delays.
| It was rumored that certain weeks there were no albums getting
| fabricated at all, worldwide.
|
| I wonder if history isn't repeating itself. AOL CDs had pretty
| much jumped the shark by 2000.
| PeaceTed wrote:
| All that manufacturing demand led to enough free machinery to
| get those free Ubunutu CD's! ;)
| rsync wrote:
| It's rough out there and has become increasingly difficult to
| maintain our pace of storage deployment.
|
| Further - and most concerning - is the pollution of the supply
| chain with refurbished/recertified stock being sold and marketed
| as "new".
|
| One example:
|
| https://kozubik.com/items/MaestroTechnology/
|
| I strongly advise buyers to stick with trusted suppliers, avoid
| Amazon/ebay channels, and _carefully_ vet your incoming stock
| with SMART tools to ensure you receive what you think you are ...
| especially for SSD parts.
| kkylin wrote:
| Question for all of you more knowledgeble than I: can SMART
| data be tampered with? When I get, say, a refurbished Mac from
| Apple, I'm trusting Apple won't stoop to that. But a SSD vendor
| I've never heard of?
| fny wrote:
| Yes, it can be tampered with. Drives can even lie about the
| amount of storage they support. I once bought a 1TB pen drive
| that was only 32MB for $10. (Yes, I knew it was a scam
| beforehand.)
| rsync wrote:
| Yes. There are vendor-specific utilities that have escaped
| into the wild that allow bad actors to reset various SMART
| counters, etc.
|
| A lot of abuse came to light during the launch and initial
| mining of the (ridiculous) Chiacoin[1] during which Chia
| miners would burn through SSDs to within a hair of their
| usable life, reset their SMART stats, and sell them as new on
| Amazon or ebay.
|
| As can be seen in my above comment, larger distributors like
| "Maestro Technologies" have their stock polluted with parts
| like this and I find it _very unlikely_ that they are not
| aware of the status of these parts they are selling as new.
|
| [1] https://en.wikipedia.org/wiki/Chia_Network
| estimator7292 wrote:
| DO NOT assume SMART is reliable. You can wipe SMART stats or
| write any values you want.
|
| You have to actually examine the real bits on the drive.
| Resellers don't want to take the time to actually zero a drive,
| they usually just nuke the partition table.
|
| You also need to physically examine the drive. Corroded
| fingerprints on the PCB, wear on the port contacts, scratches
| from mounting rails, etc.
|
| That's how it found out that the last "new" drive I bought on
| Amazon was actually a used Backblaze drive. It contained
| terabytes of customer data, and a shit ton of cleartext files.
| SMART, of course, reported it was a brand new drive with zero
| hours. Cleartext logs on the drive showed many thousands of
| hours of runtime.
|
| Physical examination is the only reliable method.
| neilv wrote:
| > _That 's how it found out that the last "new" drive I
| bought on Amazon was actually a used Backblaze drive. It
| contained terabytes of customer data, and a shit ton of
| cleartext files. SMART, of course, reported it was a brand
| new drive with zero hours. Cleartext logs on the drive showed
| many thousands of hours of runtime._
|
| This sounds like it could be a big problem for Backblaze
| customers, and consequently for Backblaze.
|
| Can you alert the Backblaze CEO about their insufficiently-
| decommissioned drives leaking out like this?
|
| Backblaze customers also need to know, but I would give
| Backblaze the first shot at figuring out how to notify, whom,
| of what.
| SoftTalker wrote:
| > drive I bought on Amazon was actually a used Backblaze
| drive
|
| Assuming this is true, I find it weird/surprising that
| Backblaze doesn't at least zero their drives before disposing
| of them? I have to do that at my work, and at least by policy
| I could lose my job if I skipped doing it.
| loloquwowndueo wrote:
| But you don't work at backblaze :)
| catigula wrote:
| Nearly any product you can buy from Amazon, even when it says
| shipped from Amazon, is suspect.
|
| I wouldn't shop there at all. It's a literal scam market.
| Allegedly.
| evanjrowley wrote:
| The future of AI is everything we liked getting ruined by AI.
| criemen wrote:
| So far, due to the AI boom, we're lacking:
|
| * HDDs
|
| * SSDs
|
| * DRAM
|
| * GPUs, obviously
|
| * Power to hook up to our datacenters
|
| Anything I'm missing? What a crazy world we're living in.
| magicmicah85 wrote:
| Water? Clean Air? ROI?
| captainkrtek wrote:
| Nearly nothing left unexploited in the pursuit of profits.
| SlavikCA wrote:
| And there is no profit, too.
| captainkrtek wrote:
| Its very depressing that we're almost quite literally
| converting money into BTUs via GPUs and not even making a
| buck, meanwhile the ocean gets hotter
| stuff4ben wrote:
| jobs...
| maerF0x0 wrote:
| Makes me wonder about the things that go into datacenters and
| power plants:
|
| - Air conditioners
|
| - generators
|
| - Racks/Cages
|
| - fire supression
|
| - Concrete, Steel, power lines
|
| etc
| wmf wrote:
| All those have multi-year backlogs.
| PeaceTed wrote:
| For all the issues with power use, I do grant it that a lot
| of materials and labor in needed at the same time. It isn't
| JUST burning up cash to make Sora videos.
|
| But by the same measure, the Great Egyptian pyramids would
| have been a huge boom for work even if the final product
| didn't achieve much for most people.
| derefr wrote:
| I have a hypothesis that it was AI, not COVID/sanctions/etc,
| that was mainly responsible for the 2020-to-ongoing "chip
| shortage." Ignoring companies with their own fabs (Intel) and
| companies with pre-existing reserved-capacity contracts with
| fabs (Apple), everyone else is stuck waiting in line behind
| batch after batch of fab orders from Nvidia.
|
| Downstream of that, AI is effectively also responsible for the
| current generation of game consoles never declining in price.
|
| Because game consoles are fixed platforms that continue to be
| manufactured over 5+ years, _normally_ the most expensive parts
| in the system (the CPU and GPU) would gradually get cheaper to
| manufacture [and in turn, cheaper to buy] over the course of
| the console 's lifetime--which was often passed onto the
| consumer in the form of the console's MSRP gradually
| decreasing. Either the process node for the console's silicon
| design would stay fixed, and demand for this process node would
| gradually decrease as larger fab customers move on to newer
| nodes, decreasing the (effectively auction-based) pricing for
| fab time on the older node; or the console manufacturer['s
| silicon vendor] would put their silicon design through a
| process-shrink, and so, while still paying top dollar for use
| of the fab's newest node, would be getting more chips per wafer
| out of that, and again could charge less.
|
| But instead, what we've seen since the start of the AI boom is
| that there's no longer any price-reduced timeslots to be sold
| to manufacture the low-BOM parts for these price-sensitive
| console-maker customers. Instead, both Nvidia and AMD are now
| getting such high value out of even the older nodes, that 1.
| the fabs know they can squeeze them, charging full price for
| those slots as well; and 2. both Nvidia and AMD, in their roles
| as silicon vendors to the console makers, haven't been able to
| justify using (very much of) the time they're paying so much
| for to fab low-BOM-cost parts to fulfill their pre-existing
| outstanding purchase orders, when they could instead be
| fulfilling much-higher-margin _new_ POs from the hyperscalers.
|
| Thus every console of the ninth generation (PS5, Xbox Series,
| Switch) still selling for their launch price (with no help from
| process shrinks); thus none of these consoles having been able
| to be produced in excess of demand to the point that supply-
| and-demand ever drove retail prices below MSRP; and thus the
| only tenth-gen console so far, the Switch 2, taking ~4 years
| longer than anticipated to release+.
|
| ---
|
| + Nintendo were very likely waiting for Nvidia to run off
| enough of the Tegra T239 _with a sufficiently low passed-on fab
| cost_ , for Nintendo to both 1. be able to build a backstock
| and non-run-dry pipeline of Switch 2s, and 2. be able to be
| positive-margin on charging the same price as the Switch 1 for
| them. They waited four years, and neither thing ever happened;
| so they eventually just gave up and priced the Switch 2 higher,
| and also built out an entirely novel D2C + online-marketplace-
| partner distribution pipeline so they could ration the tiny
| initial supply of units they had been able to build with the
| chips Nvidia had supplied them so far.
|
| Though, that being said, Nintendo actually got a double whammy
| here. They were _also_ waiting for fast NAND to come down in
| price, so that they could have physical game cards manufactured
| for a trivial BOM price while still enabling the "direct GPU
| disk-streamed assets" pipeline that games of the last
| generation had begun relying on. Obviously, as today's article
| points out, that hasn't been happening either! Thus game key
| cards; thus SD Express cards only beginning to trickle out,
| with no sizes above 128GiB available at the Switch 2's launch
| time; and thus those SD Express cards being ridiculously priced
| for their capacity compared to equivalent transfer-speed + die-
| size NAND (as seen in e.g. low-profile/flush-mount flash
| drives.)
| PeaceTed wrote:
| That sounds reasonable but I suspect it was a little bit of
| both. Initially Covid slow down followed by large demand. A
| lot of older Nodes were shutdown (300-400nm) with the Covid
| slow down and a lot of chips ended up having to be moved over
| to newer nodes once demand picked up again. It led to a big
| swing in the flow of production and would have roll on
| effects.
|
| Combine that with a huge GPU boom and you have the setup for
| production issues.
| 1970-01-01 wrote:
| * True innovation
|
| Every tech startup circa 2023 doesn't do anything without
| smearing AI all over their front page.
| domador wrote:
| * Time to wade through all the AI slop
|
| * Trust that what we are watching or reading is human-sourced
| and real
|
| * Time for real creators' unique voices to remain unique
|
| * Hope for society's future
| penguin_booze wrote:
| Intelligence. We're lacking intelligence.
| ChrisMarshallNY wrote:
| AI: "Astounding Idiocy"
| ls612 wrote:
| Thank god I built my NAS in early October it seems like I got the
| proverbial last train out of town.
| rwyinuse wrote:
| Same here, though I did it a bit earlier. Data center SSD's I
| bought a while back for 50 euros each from Ebay now sell for
| over 200 euros. Similar increase for HDD's too.
|
| Maybe in 3-5 years there will be a fantastic time to upgrade
| again, after AI bubble has burst and slightly used stuff gets
| dumped from data centers to online marketplaces.
| doublerabbit wrote:
| Same. Wish I bought 2x 12TB drives now rather than 2x8TB.
| ls612 wrote:
| I bought 3x22TB WD Red Pros to put into a RaidZ1 vdev so I'll
| be good for a while. I have a clear upgrade path all the way
| up to 8 drives with 2x vdevs in a pool but that is something
| for far future me to consider.
| zkmon wrote:
| This is getting out of control. Like a monster eating off all the
| things that are needed for normal people. Worse than nukes race,
| which at least did not affect common man. No one really gains
| anything. All gains are temporary competitive edge, that vanishes
| quickly while locking everyone into the vicious cycle.
| fajitaforce5 wrote:
| Maybe when the bubble bursts there will be lots of good used
| hardware at massive discounts?
| criemen wrote:
| But why? Training hardware maybe will appear, but everything
| that's running inference has real, paying customers on it,
| with current capabilities level. Even if the bubble bursts,
| why would that demand evaporate? People still would want to
| use chatgpt, claude code etc., even if they stop getting any
| better tomorrow.
| krior wrote:
| But maybe not for the price it actually takes to run these
| models. Afaik atm they are all subsidized.
| criemen wrote:
| I'd expect at least API pricing for all major players to
| be margin-positive. Positive ROI if you include model
| training cost? Maybe not, but in a hypothetical scenario
| where all capital for new model training goes away, the
| latest frontier model weights still continue to exist.
|
| Subscription prices for claude code et. al? No idea.
| PeaceTed wrote:
| Yes, but how much do people want to pay to use this stuff?
| That is the real question.
|
| It might be that more than $30 a month is too much and
| people stop using it. However, I suspect it would end up a
| fair bit higher than many would think. If you are a 'whale'
| (to use the gambling term) then there is a good chance that
| they could charge in the hundreds of dollars.
|
| Maybe it become a wealth divide between those that can
| afford it and those who cannot. In equality yet again. Then
| the echos of Dune, the 'Butlerian jihad' and 'Thou shalt
| not make a machine in the likeness of a man's mind' will
| start to come up.
| pinkmuffinere wrote:
| Maybe the real paperclip maximizer was the irrational
| exuberance we made along the way :')
| whynotmaybe wrote:
| Hunger sharpens ingenuity.
|
| Soon we'll see many "AI" product that analyses network storage to
| identify files that could be deleted because they're the nth copy
| of some random report that was never published.
| walterbell wrote:
| Cloud storage and on-prem storage vendors have long provided
| deterministic block-level dedupe via hashing, without expensive
| LLM content analysis.
| mschuster91 wrote:
| So, it's Chia all over again? [1]
|
| Sometimes, I think government supply control isn't that much of a
| bad thing. That way, the government could force the availability
| of goods for the common market as well and not just for the
| really big dogs flush with ample VC money to burn who can pay any
| price.
|
| [1] For the uninitiated: a cryptocurrency where the limiting
| factor wasn't CPU, RAM or GPU compute resource, but storage - in
| 2021, there was so much craze around it that HDD and SSD prices
| exploded, and after the bubble collapsed _a lot_ of heavily
| abused drives flooded the markets.
| Neywiny wrote:
| I just got 2 16TB drives for $90 each off Craigslist. Perfect for
| video game storage or downloading huge AI models or whatever. Or
| legit Linux ISOs. Wish I got more but I was too late. Can't find
| anything close to that price online
| jdprgm wrote:
| $5 a TB is insanely cheap and someone just getting rid of stuff
| and not really caring about the price. Nothing has ever been
| that cheap at retail or manufacturer recertified. Lowest was
| like $9-10 last year and now we are sitting at more like $15.
___________________________________________________________________
(page generated 2025-11-12 23:01 UTC)