Yesterday I did a price check on the PC I built two years ago. It went from $2300 to $3650. The bulk of that increase was that the ram went from $210 to $940. Its now more expensive than when DDR5 was new.
The value of my desktop pc has almost doubled, my ps5 is worth ~ $150 more than I bought it years ago.
It's gotten to the point where nvidia doesn't even bother to report their gaming revenue anymore. It's a clear sign that we're back to the bad old days of pc gaming being a 'prosumer' hobby, but don't worry I'm sure nvidia and their ilk are salivating at the idea of making pc gaming into a streaming stadia like solution that you pay for monthly
I think I paid a total of around $5,500 for the current components of my PC. Hard to say for sure since my PC has been a Ship of Theseus for over 30 years and started as a 486. The link merely reflects its current state.
At one point, PCPartPicker was showing my PC as worth $11,000. It's now at $7,200 without including the RAM or PSU. That would put it at $9,000.
> It's a clear sign that we're back to the bad old days of pc gaming being a 'prosumer' hobby
Yup.
I think it's especially bad since the gap between budget-grade and mid-grade feels like it's gotten wide. If you wanna play the latest AAA games and not feel like you need to upgrade in 3 years, you can't settle for the budget grade unless you're still gaming at 1080p.
I wouldn't recommend spending under $3,000 for a gaming PC these days, and that's just an absurd price.
I thought I read that Samsung SATA SSDs were discontinued, but apparently that was a rumor and Sansung has denied it. I wonder why they exceed NVMe prices. They're the only SATA drives left with DRAM. I guess they could just be milking that fact.
I bought a 5090 12 months ago, just checked - that’s basically up 50%! I used to joke i’d retire on all the old tech in my loft, everyday now it feels less like a joke!
The memory in the PC I put together early last year is now worth about three times the total cost of all the parts I used to build the thing. It is absolutely crazy.
Memory is still a commodity product, in that there isn't a huge amount of difference between vendors selling products that comply with a certain technical standard. Sometimes the prices of commodities (wheat, silver, crude oil, etc) go way up when supply and demand get out of balance.
I regret not building the PC when I was looking at it. It's not a money thing, at the end of the day, but I can't bring myself to do it.
I had it all priced out, but a bunch of birthdays in my family were coming up and I felt like I shouldn't buy something for myself if it's really their time.
My old laptop will have to cut it for a while. :-)
The squeeze is real even at the SME level. We recently wanted to add another TB of memory to several servers (we do EDA chip design, which eats a lot of memory). Quotes came back to about €200k for 48 x 96GB DDR5-5600 RDIMMs. Mind you, this is for refurbished memory with 1 year warranty. I'm still figuring out if this is FU-pricing or just how things are going to be from now on.
Spec-ing and buying servers has become quite the pain in the past year, at least at the relatively-small scale we operate at. It's "dynamic pricing" with most quotes being valid for 24 hours :(
That's about 50% higher than you'd pay for retail, small quantity, off the shelf alternatives in the US.
Many vendors have limited allocation so they're sending out extremely high quotes to see who will take it. They know everyone is seeing news about high RAM prices and getting new vendors approved can be hard, so they raise prices and see who takes it.
I would recommend getting quotes from multiple vendors. Don't be afraid to push back. They might come back immediately with a lower price in this market. Also check the retail suppliers you can access in your country.
I’m just working with an EDA client to upgrade their almost decade old machines to run the cadence tools…it’s grim. Real real grim. I was pricing out servers this morning with 64gb(!) of RAM for almost 20k. The machines were running now, some of them have almost a terabyte of RAM. I think the designers are just going to have to suck it up and use the slower machines
If you don't need the full memory bandwidth of DDR5 for any particular application area, the best solution is probably just to add flash storage for mostly read-only data and Intel Optane (still sold in the secondary market at around $1/GB) for write-heavy scratchpad space. The latter gives you around DDR3ish performance at a higher power draw than DRAM or modern flash, but its wearout-resistance and lower latency compared to flash still matters.
> Quotes came back to about €200k for 48 x 96GB DDR5-5600 RDIMMs.
That's around EUR 40 per GB? That seems quite high compared to what consumer DDR5 RAM is selling for, though it being RDIMM may account for some of the difference.
That’s about right, sadly. We were buying servers with 1TB DDR5 standard, no longer.
Quotes are 7 days max, lead times fluctuating out months, and often now have language they can choose to not honor the quote for any reason due to price fluctuations.
I got quoted $4K, $6K, and $8K for the same enterprise SSD part in three different quotes within about a week. (I normally expect this part to be $1.5-2.5K.) In one case, a storage array with two SSDs in it cost approximately the same as buying those two SSDs standalone.
When it comes to server RAM, I got quoted double a previous quote within the span of a week.
I still believe that Mark Zuckerberg would have been smarter investing all his AI funding into just making RAM fabs. Would probably have printed easy cash. Instead he's burning cash on making IG / FB less secure because of his in-house AI.
> Quotes came back to about €200k for 48 x 96GB DDR5-5600 RDIMMs
I sell used RAM in trays of 50 sticks (I physically know how much that is, I've held that many). I've had some trays sell for a few thousand dollars. Six figures for just one tray is absurd, but I've never collected enough DDR5 to make a full tray.
GN did a documentary on the situation from the perspective of consumer-facing companies. Seems pretty dire for them, and it's hard to see the long-range consequences, but the idea of consumers being priced out isn't too far out, which to me is a little alarming.
Consumer oriented parts businesses are going to go out of business from this. Motherboards, cases, coolers, power supplies, fans, etc. At this point it's a matter of how many. Selection and competition will be permanently reduced, as I don't see these types of companies coming back any time soon.
Who needs an economy built on consumer spending when a handful of massive tech companies can just keep passing around the same trillion dollar check like a hot potato?
I want a more complete picture of why prices are so high from articles like this.
Is supply actively constrained, or is this mostly in anticipation of future shortages? How much of this is a mix of panic buying and price gouging on bad news?
I care more about the secondhand market. Prices are nuts for old used gear, but that also tracks with patterns I've been seeing since roughly the pandemic where more and more secondhand sellers on the usual platforms setting pricing patterns are small businesses, not hobbyists.
Supply is both already constrained and AI companies have pre-purchased enough HBM at enough of a premium that most of the wafers are allocated to them. All the intermediaries are jacking up their prices so their inventory doesn't empty too quickly, because they may not be able to refill it.
It is hard to overstate the damage that the infinite money being poured into AI is doing to the wider economy. Anything involved in datacenters is going to experience shortages/price rises. A pre-existing problem is power transformers: https://www.reuters.com/business/energy/us-power-transformer...
The impact on domestic electricity prices in a year just after high oil prices is not going to be popular either.
> It is hard to overstate the damage that the infinite money being poured into AI is doing to the wider economy.
I get what you're saying but medium term this is an extremely funny sentiment. This money being poured is likely to end up being a huge boon for a lot of economic sectors, including in the US. Most commodity shortages like this end in a glut, with a medium term win for consumers, even if we have 1-2 (more) years of pricing pain. Meanwhile expensive RAM has so far left stock for people that really need it. Calling this kind of demand economic damage is odd.
> Most commodity shortages like this end in a glut
The 3 RAM manufactures know this too, from painful experience. There won't be a glut this cycle because there are no capacity build-outs. Instead of increasing capacity, some OEMs left the consumer segment to focus on enterprise AI.
> Calling this kind of demand economic damage is odd.
It's not odd at all because the complementary industries are being damaged - possibly permanently: manufacturers of motherboard, cases, fans, and the entire consumer PC supply chain are being negatively impacted. Expect bankruptcies and consolidation if this lasts for 2 more years.
CXMT is a tiny supplier, it will take years before they significantly expand the market. They're pretty much in the same boat as Samsung and SK Hynix, they just have less of an incentive to actively curtail supply.
> I get what you're saying but medium term this is an extremely funny sentiment.
Sentiment is the right word here. None of us really know, and go by feeling. If you perceive the AI boom as approaching tulip craze levels of irrationality, it feels pretty dire.
The RAM is a commodity and may be repurposed afterward. This kind of thing is a bit like a debt jubilee when the dust clears and survivors scavenge the dead. But a lot of other build-out may essentially be waste. To me, apologists for this boom seem to be harboring a variant of the broken windows fallacy. Not all economic activity is productive.
The other kind of damage is opportunity cost. How many players in other industries are being strategically harmed by this situation? We can't all just live on AI token output if these other industries retract too far.
> This money being poured is likely to end up being a huge boon for a lot of economic sectors, including in the US.
No such thing as a free lunch. Whilst a lot of industries are doing extremely well right now (i.e. everything involved in datacenter construction), everyone else has to actually pay for it.
Therein also lies the damage; The economy wasn't doing so great to begin with, and now this massive multi-trillion-dollar expense has been added.
The direct expenses come out of the pockets of Big Tech, but all the indirect stuff like the DRAM crisis affect the wider economy directly.
It's not odd at all to call this economic damage unless you're that disconnected from anything but AI/high-tech SV companies.
There's smaller businesses that I know _need_ data capacity (hard drives) that can't afford them and are facing serious CAPEX challenges.
I work for a start up that helps folks move off of existing on-prem virtualization solutions and every small to medium company across every sector you can think of aren't just able to not afford additional storage and memory, they can't find them either.
There isn't as much stock as you think. I can't wait for this AI bubble to pop. It's been absolutely terrible for 90% of the industry.
Yup. Once production ramps up (and then subsequently overshoots) in the next couple years we're going to have GPUs with 96GB VRAM for the price of a 4090 today.
> It is hard to overstate the damage that the infinite money being poured into AI is doing to the wider economy
I know it's an unpopular opinion on Hacker News but I think most people are overstating the damage to the economy. Increased demand starts to pull more advancements forward and increase spending on production, which benefits everyone.
Transformer demand is real, but what percentage of your electricity bill do you think goes to spending on those transformers? The number is so small that it's a rounding error. Many examples like this where we see headlines about some part going up in price and forget that it's such a negligible piece of our bill that it barely matters.
The cost of fuel inputs for base generation and peak supply are a bigger factor.
> I know it's an unpopular opinion on Hacker News but I think most people are overstating the damage to the economy.
There are two issues... one being that an economy needs people to buy things to work, and people only can buy things if they have money, and people only have money when they have work - but everyone is laying off people left and right due to "AI".
The other issue is that there's trillions of dollars floating around between all the companies. It is likely that a significant chunk of these will fail and give us another 2007.
Memory makers are reluctant to increase fabrication (high cost of capital and boom/bust cycle leads to bankruptcies being common in the industry), and the memory required for AI (HBM) requires more of the production inputs than other types of RAM, thus squeezing regular DDR RAM even more. Long-form article describing this in more detail: https://davidoks.blog/p/ai-is-killing-the-cheap-smartphone
> Is supply actively constrained, or is this mostly in anticipation of future shortages?
Both. AI datacenters create immense demand on DRAM, NAND storage, even HDD storage.
Then Sam Altman went ahead and "bought" 40% of the world's DRAM production, by way of secretly approaching both Samsung and SK Hynix (each unaware of the other getting a similar deal), which sent the entire market into a panic and everyone rushed to buy ahead of the anticipated supply shortage.
The kicker is that it wasn't even a proper purchase agreement. Just a non-binding promise of "By 2030 I will have infinite dollars and buy your DRAM".
I looked at my eBay receipt from 2023 and I paid $84.98 for a "Kingston FURY Beast 64GB (2x32GB) 3200MHz DDR4" listing and now the equivalent on eBay "Buy It Now" is $374.99 for "Kingston FURY Renegade 64GB (2x32GB) DDR4 RAM 3200MHz (KF432C16RBK2/64)". What a timeline it has become for consumer computing three years later!
My anecdata: bought used DDR4 ECC 16gb ram sticks (i.e. serverpulled ram) for $11 off ebay last year (Jan 2025), now as of June 2026 the lowest listing I see is $42, most are around $50.
When I bought the same ECC DDR4 16gb ram stick in 2016 (when DDR4 was rather new) it cost $85 (cant remember if it was new or used)
To be fair, DDR3 is still kinda cheap. It's really only DDR4 and DDR5 that are massively in demand. DDR3 is a bit too old for it to be in high demand by consumers.
Unbelievably, even DDR3 isn't spared. Our suppliers have been quoting us much higher prices for DDR3 for months now, and we're experiencing persistent stock shortages. Fabs are abandoning legacy nodes to prioritize AI capacity, making it rough even for maintaining old gear.
It isn't really unbelievable. Anyone who could sub in DDR4 for their projects would do so which increased the price. And then anyone who was still using DDR4 for projects would respond to the increased price by moving to cheaper DDR3 if they could. It wouldn't take much of an unexpected demand increase to move the needle since I don't believe much DDR3 or DDR4 is even being manufactured anymore. And then you have the supply side: Increased prices of upgrades will cause many of them to be delayed or cancelled denying the refurb/used market supply of the old hardware.
Personally I have a lot of DDR3 RDIMMS laying around that I got for cheap/free and have thought of getting a used workstation to put them in (4 channels of DDR3 will still net you ~60 GB/s of bandwidth) but the only Xeons that support it are Sandy/Ivy. Everything Haswell and newer with AVX2 and rebar/etc. use DDR4.
While DDR3 is cheaper it's still tripled or quadrupled in price over the past two years. I just bought a pair or sticks for an old Mac Pro and it was 4x what i paid just a few years ago.
I work in the refurb department of an e-waste recycling company. I've been collecting some DDR5-based systems for months, but with the prices, they're probably more than what most people are looking to spend on a technically second-hand computer (even if they are like new and still under warranty). I've priced them at about what the CPU + RAM + SSD would sell for separately, and I'm not willing to go much lower than that.
And if you're wondering, who's throwing out DDR5 systems? A local healthcare company. The boxes for some units are crushed and have water stains on them, and I imagine others don't meet their exacting requirements in some minor way (though they look and work OK to me, regardless of scratches and dings on the case).
>they're probably more than what most people are looking to spend on a technically second-hand computer (even if they are like new and still under warranty)
Some people need to buy at these inflated prices anyway, and are ultimately willing to spend. Maybe it differs in your area but there's no risk in listing these on Facebook Marketplace or similar platforms and see the interest.
Edit: they're intended to be retail POS PCs, like cash registers. (The motherboard literally has a header for a cash drawer!) So unless you're a particular kind of retailer, you're not asking your distributor for them. If distributors aren't looking for them, our B2B buyers aren't buying them. If they were Elite/Prodesks, Optiplexes, or Thinkcenters, they'd probably be gone by now.
Why don’t you part them out? The market for the parts will be more liquid. You might sell faster and at a higher price. You could try doing it once and see how it goes.
I'm sure it doesn't help that people continue to buy things at this price.
Steam Deck had a huge price increase (~40-50%) but it still sold out in 24 hours.
All it would take is for everyone not to pull the trigger on buying things for a little bit and prices would fall but instead enough people are buying things at a crazy markup. If anything that's a signal to sell things at higher prices. Of course AI is amplifying this problem but realistically people are still buying consumer hardware at these prices which lets businesses know people will pay this price.
I'm on a machine built from parts in 2014 and it's all very good for me to do every day development so I'm not posting this from a machine I won't have to touch for another 10 years.
I believe Steam Deck is a major exception. Given how fast they sold out I also wonder how many Valve produced this batch. Most vendors of PC parts are seeing steep declines. Also after seeing 2-3 price increases in everything (Switch, Playstation, XBOX, etc.) people are likely getting it now because the prices aren't going down anytime soon and will probably just go up again for a couple of years.
I pulled the trigger on an early Ayn Thor because it was obvious this was going to happen. Something I didn't really want to fit into my budget but knew that if I didn't I would regret it later.
I doubt it helps but this is such a small piece of the pie I'm not sure how much this hurts things either. Steam Deck is kind of a niche in a niche and doesn't sell in huge numbers compared to other players like PC OEMs and phone OEMs that are now all over a barrel as well as OpenAI tries to buy all the RAM so no one else can have it.
> I'm sure it doesn't help that people continue to buy things at this price.
Implicitly, but that's blaming the consumer who has no or few equivalent choices. Purchasing RAM is not like choosing between Coke and Pepsi. A better analogy is that when a hurricane is coming or a natural disaster has already hit, it doesn't help that people will purchase food and fuel at any price.
"The costs are negligible and justified when compared to all the benefits. If you look at the performance gains, the overall cost has in fact been reduced."
- Altman, a.k.a. Dory from Finding Nemo and/or Dario, a.k.a. Carl from Jimmy Neutron.
My bet is that the prices will crash once OpenAI (and/or Antrophic) IPO's have happened.
Right now the biggest threat to their IPO's is that people realize that local models are good enough for whatever they're peddling, what's the most important factor to even running good enough models? RAM since you want the models in memory to not be total slogs.
But remember that markets can stay irrational longer than anyone can hold his breath. If they get more funding there's a good chance they'll invest more in the destruction of the remaining production capacity. Admitting that with normal pricing anyone could have a decent AI-machine for 2K is hard - prices for acceptable AI-machines most likely will go >10K first.
The chance is that they cache out during IPO, and will lose interest to increase capacity, some/many contracts will be canceled, and demand being reduced.
Luckily very few people can configure and are interested in local models. But your nearby datacenter running Chinese open-weight models is also good enough.
My point is that dram demand is mostly orthogonal to whether everyone is using open weight models or secret weight models. Heavy demand for local models (whether secret or open weight) will require even more aggregate DRAM than for shared.
Demand will only go down if people reduce their use of these AI tools. Given how much folks here complain about quotas, I'm very skeptical that will happen willingly.
Open weight models allow for repurposing existing hardware locally, and there's a lot of it around - far more than the amount of new RAM being supplied. So they add some short-term downward pressure to the price. (But not very much, since these datacenter builds are long-term investments that are targeted at eventually running far larger models.)
If regular people can repurpose old hardware, so can shared providers, who can extract more value from the hardware and thus afford to pay more.
In a constrained market, supply and demand favors folks who can most efficiently extract rent. Local models only make sense in a world with abundant compute and energy.
The billionaires locked in a race to spend effectively unlimited funds on AI CapEx will have to be convinced by markets and/or their advisers that there aren't enough profits and that cutting losses (like with Metaverse) in their quixotic quest is necessary.
And taking everyone's 401ks down with them because of the idiotic rules changes to accommodate SpaceX? Its going to be the largest transfer (theft) of wealth since 2008. AI is the only thing driving any kind of growth in the market at all now. That pops, its going to bring the entire economy down with it.
The vast majority of index funds are float adjusted. SpaceX will not have that many shares available relative to the total value of the company. It's a negligible percent of the overall market cap of say, VTI. This commonly repeated trope is misinformation. The change in rules is more of a problem than the actual value.
They are actually quite a bit better than you might think. Qwen3.6 27B is pretty capable at coding.
For non-coding work, they are more than good enough. A lot of the ways my non-technical family members have interacted with AI would be perfectly served by using a local model.
After all, people were more than satisfied with the results from GPT 3. That has long since been surpassed by open weight models.
I'm sure there are things local models are good enough at in non-coding work, but for anything complex I do not find this to be the case.
I'd say local models are fairly capable of even somewhat complex coding execution. For complex non-coding work (research, in-depth analysis, assembly of complex info-dense documents) I'd rather do it by hand than switch from Opus 4.7 to anything I could even theoretically run locally.
I don't know what kind of coding, but for my case it's been useless. Not working code almost every time. It's much quicker to just write it by hand than use that model.
I've been experimenting with Qwen3 Coder Next and Copilot for a little Rust toy project and it's been trucking along. It does require a fair bit of hand-holding (or perhaps I just don't trust it to give it larger tasks), but it works alright.
Honestly, that's the output I get from non-local models, anyway. If I'm going to get plausible nonsense either way, I may as well run it on my own hardware.
I wanted to upgrade my SSD but prices are more than at the end of 2025. I refuse to pay 500 euro for a 4TB SSD. I rather go outside and play with my bike like when I was 5.
I tried to buy new SD cards for my camera. The cards I used to buy at $28 are now $80-120... if you can find them. Another cheaper card I used to buy for $19 is now $46. It's just absolutely insane at this point.
Presumably, this is at 1080p, which looks terrible after you experienced 4K. It's better than nothing, and I probably wouldn't build a 4K rig in this market, but let's not pretend you can get a modern gaming experience out of that old hardware.
Here I'm on a 5800x and a 4070ti super and it's woefully inadequate for some games. Though I suppose if you play at 1080p (which is hard to look at these days for me) you might be OK on low-medium settings in some newer games. I can't see you getting very far with something like Forza Horizon 6 or Death Stranding 2 though.
<FourYorkshiremen>Luxury.</FourYorkshiremen> I'm still using a 1650 Super with 4GB VRAM and it's basically fine. Holding off on a few newer titles, but I'm old and my eyesight is going so 1080 resolution is plenty for me.
I recently booted up an old 4790k system and it was fine on Linux but on Windows it would nag me to update but apparently the CPU is too old for new Windows. I ended up giving it away on the Internet to whomever could pick up but afterwards it ended up with one of those reseller chaps. Ah well, I wish it had made to some kid somewhere.
> Windows it would nag me to update but apparently the CPU is too old for new Windows
When I was trying to upgrade to Windows 11 from Windows 10 I got the same warning. It is the reason why I'm now happily running Xubuntu on the same machine.
I did end up installing a Windows 11 VM (VirtualBox) which works perfectly fine running as a guest under Xubuntu on a 4790k (screw you Microsoft). But I otherwise never want go back to Windows as a daily driver. The enshittification just got too much for me.
I played 100+ hours of RDR2 on a 2060 (non-super), as was the style at the time. When the 30 series came out, I sold that card for more than I paid for it.
AMD just brought the popular 5800X3D back out of retirement to give people maintaining the DDR4 based platforms something to buy. Last I checked used DDR4 was half the price of used DDR5 after the prices of both shot up.
It's not even "old crap" the PC consumer market has basically died in less than year due to your average person getting priced out, I cannot imagine how catastrophic of an impact this is going to have on younger folk and the barrier to entry to desktop computing
It also was the same chip back then and the price difference significant. But maybe you are correct and with the current conditions it doesn't make a difference. I dont know.
Time to migrate all those electron "native" apps to actual native code. I bet with some decent optimizations 4GB will be more than enough for casual user, and still with some free memory to spare.
Pretty insane. I built a Framework Desktop PC back in November. The motherboard (with 128G DDR5 RAM) was $1800. Now it's $2859. Almost 60% increase in 6 months.
Crucial has "good" prices for DDR5 6000mhz memory on Amazon. The downside is you have to wait for shipping -- I had to wait a couple weeks. I bought when these were $300 and equivalent memory prices were like $500 at the time.
All started by openai wafer capacity commitments that aren't being met anymore... The ECC ram systems I've bought are now worth more than the original purchase price of the entire system and I've been debating on just selling them since they're now worth nearly double their value which would outperform any stock trades I've ever made any given year lol.
I am thankful that both my partner and myself are in a pretty good spot when it comes to our gaming PC's. I had hoped to double my RAM at some point but I am still at a comfortable spot.
I am annoyed that the new handhelds are all crazy so sticking with my Legion Go for now.
The one I am annoyed with is storage. I desperately need to get a couple new drives for my NAS (one to replace one that its bad sectors are growing and one to add more storage) and I am not looking forward to spending $600-$700 each for 20TB drives.
Price should send a signal to manufacturers to build more capacity. I wonder if they will though, it takes quite a bit of time, and it's not certain that the demand will continue to exist once built.
Several Chinese manufacturers are doing just that, and have already expanded production: https://techwireasia.com/2026/04/chinese-memory-chips-ymtc-c... But because of tech trade barriers their primary focus is on the domestic market and only secondarily global markets.
Commerce Secretary Howard Lutnick has warned Samsung and SK Hynix they could face 100% tariffs, framing it as a choice between paying a 100% tariff or building memory fabs in America...
Do tariff threats still matter? After being struck down, I thought the only new tariffs which could happen would have to be enacted by Congress? Or world that be a fun game where the new illegal tariffs would be on the books until the courts invalidate them again?
The bottleneck isn't the sticks, it's the chips. The chips are the same for consumer and server applications. What's been happening is that big companies have bought nearly all the wafer capacity for the next year or so, and perhaps some of that capacity has also been redirected from DDR5 to LPDDR5. If a stick manufacturer drops out of the consumer market that kinda doesn't matter, because manufacturing sticks is relatively low tech compared to manufacturing the memory chips. You can compare it to manufacturing video cards vs. manufacturing GPUs (as in the actual processing elements).
Government needs to get out of the way. Micron announced a memory fab in Syracuse in 2023. It took 3 years, 20,000 pages of "environmental review", deals with the government on amount of union contracts during building, etc. for them to break ground in 2026 for a 2030 opening date. In any reasonable world, a 2023 announcement should have broke ground in 2023.
OTOH, a celulose factory near me, built in the 1950's, got their permits fast and with little regard to environment. FF three decades, and their entire surroundings are destoyed for everyone else. Trials go nowhere, because they have all authorizations needed (and a lot of political leverage because they are the main employer in the region). Careful fast-tracking business that have zero incentives to avoid externalization of costs.
The main blocker was that there were bats there so they needed to buy separate land to preserve. 20k pages of environmental review is just make work to spend money and create an unnecessary paper trail. If polluting with x is illegal then its illegal. The review doesnt stop that.
And? The primary goal should be to catch and stop pollution, not make manufacturers spend years promising not to do something they're not allowed to do. If someone wants to build a factory that can't operate without illegal emissions, then so be it. It's their money lost. All that matters is that they don't actually pollute.
Using red tape as some kind of prophylactic is ridiculous. If the state doesn't have the monitoring in place, you have to just trust the company, which is naive if not negligent. If you do have the monitoring, why require the extremely expensive song & dance? To protect corporations from negligently wasting money?
Answer: because the song & dance is primarily about extracting concessions, like union labor or even cash (e.g. promises to pay to fix someone else's pollution, or contributions to various interest groups). The friction and expense involved in today's development review processes are many times more costly to all involved than the social benefit.
The process is there because Industry has proven that it can't be trusted.
The only way to stop it is to verify that it won't happen in the first place by making sure their building plans are up to par. The song-and-dance, well even with the review, they try their damn hardest to cut corners and hood-wig wherever they can.
> If someone wants to build a factory that can't operate without illegal emissions, then so be it. It's their money lost. All that matters is that they don't actually pollute.
That's hopelessly naïve.
If you let them build the facility that can pollute, they're going to pollute.
And if you point to the pollution coming out and tell them "you have to stop," they're going to say "make us."
And if you point to the pollution already in the environment and tell them "you have to clean that up, because you put it there," they're going to say "prove it."
And they're going to tie the government up in court for years or decades, and then oh, whoops, somehow the entity that actually did all the polluting has no more money and can't do anything about it :-( Good thing they were only a subsidiary that all the profit and assets can be moved out of!
And the people who actually live there are suffering from preventable diseases and dying of cancer at rates 5x the national average.
How do I know all this? Because this has been industry's playbook for over a century.
First of all I disagree that it's difficult to get injunctions to stop an activity that was illegal from the start. In fact, sometimes environmental reviews can backfire because they typically require affirmation by the government, which can create a defense to doing something that would otherwise be judged illegal. That type of loophole is why people are so cynical.
But even so, how does the song & dance prevent any of that? It's not like, e.g., a battery manufacturer submits a plan admitting that they're gonna dump stuff.
Sure, but in terms of expenditure and especially time (and time is a huge expense because of the cost of raising and securing financing) that work doesn't make up the bulk of the cost. There's nothing wrong with impact statements themselves. You need independent assessments of hazards beyond the well known ones like chemical storage, where in most cases existing codes and regulations are more than adequate. It's that in many places, California especially, there's no fixed goal post. The review process has effectively become a vehicle for local and even national politics to play out. And a method for detractors to spread FUD (again especially in California where anyone, not just officials or local residents, can challenge and drag out review of an impact statement itself, let alone proposed mitigations). Just look at what has happened with CaHSR (they're still not completely done getting write-off on all the necessary impact statements for the planned route) or the politics involved with approving offshore wind farms, where Trump and, to a lesser extent, the Biden administration showed how much pure politics plays into things and drives costs up. (And federal EIP review process is, at least historically, much simpler and less subject to political games compared to some state analogs.)
The maxim, "a nation of laws, not of men", is applicable here. If you don't have fixed goal posts or rules, governance becomes chaotic, not to mention unjust, inefficient, and ultimately corrupt (so many people like Trump because he promises to substitute his own judgement in place of democratic processes).
It would be nice if we could comprehensively hypothesis and address every possible manner in which things could wrong, but we can't. Most of the time you need to rely on broad rules, at least as a catchall, like don't harm someone or don't pollute, and ensure fast, consistent, and efficient accountability when injuries occur. The current state of development approval has become grossly degenerate in many places. In others it could definitely use some firming up. But let's not pretend that a project in California would get away with most of the stuff they can in Louisiana just because we reformed the review process. For one thing, California is much better about enforcing existing hazard codes at both the planning approval and operational stages.
It's amazing that after all these years, a famous criticism of Neuromancer seems to have been mooted. That is, the bit about Case having that stash of stolen RAM that was his "big score" for the moment, and how Linda Lee stole it from him, yadda, yadda. For years people have read that and said something like "WTF? RAM isn't valuable enough to be a black market commodity".
Well... I guess William Gibson laughs last, after all!
Yeah, the prices are actually down a few pp compared to record highs recorded before the war. The war might even exert some downward pressure. It increased the cost of living, thus reducing demand for comparative luxuries like new devices. Laptop and smartphone markets are contracting heavily right now, even for devices with stable prices enabled by long-term RAM procurement.
Take memory in house Apple (the talent is already there) its time again to kick another hardware supplier out and move on ala Intel, Nvidia, AMD, IBM, and Qualcomm outgoing in 2027-2028 if you want to continue to built devices and actually sell something to the public.
Is it just me or do I have the feeling that we have gone too far with our memory requirements? Why everyone now suddenly need 32Gb, 64Gb or hundreds of additional Gb of RAM?
Same thing with GPUs, is kind of insane having so much processing power and yet requiring more and more. What purpose for? What's the limit? Does it really really pay off such investments?
For me as a non-AI developer (I don't use any models of any kind, nor I train LLMs at all), a system with 16Gb seems to be more than enough for a vast number of applications.
Actually I was using a laptop with 16GB around 2013 already. That's 13 years ago. Memory requirements kind of stagnated after that. Obviously, LLMs are driving up memory needs now.
I have a 48GB development machine which is nice if you are running IDEs, backend stuff via docker, etc. And even just having a bunch of RAM there for file caching helps keep things fast.
I was in the US two weeks ago, looked everywhere for 1 module 32 gb ddr4 sodimm, couldn't find it anywhere. But apparently it was pretty expensive as well (from the price tag on the empty shelf in best buy)
Crazy, the other day I looked in my local store order history and say that I bought G.SKILL RipJaws V F4-3600C18D-32GVK, a DDR4 32 GB 3600 MHz kit of two sticks.
I bought it for 82 EUR, before the whole ongoing situation.
Now the same spec costs upwards of 290 EUR, about 3.5x the original price and even on Amazon the best prices I can find are upwards of like 210 EUR (2.5x).
this is almost certainly a us tarriff/China sanctions thing rather than an AI thing. sticks here outside of the us tarriff system never really changed, I just bought like a month or two ago 128GB DDR5 for $500 at a time when US best buy was listing the same kingston 32GB sticks for $200 each.
That said, getting hold of them was hard and needed a special order.
"back in my time", a dialup and 32 megs(!) of ram was enough for most stuff, including internet browsing.
I have no idea why a weather forecast site needs tens of megabytes of resources, and gig+ of ram for my browser, since i get no more info from it, than i did back then. Same for chat programs (how is discord different than irc? and why does it need so much ram to do so? same for slack), mail clients, etc.
Maybe it's time to kick developers to start optimizing stuff a bit, since neither they nor the users can't afford "unlimited" ram anymore.
edit: i'm not saying we need to get back to literally 32 megs of ram, just to make developers performance test their stuff on a laptop that was on sale 3 years ago in their local supermarket, i.e. stuff their users use at home.
I've taken advantage of this.. Over the last 6 months I actually made money on my old PC parts inventory, but don't plan on buying any new hardware anytime soon.
My strategy of just buying the best quality parts within my budget when I need a new build has never failed me yet. Is pretty insane that I bought 4x 32gig dd4 ram for $400 in 2024; literally the same part I paid $200 now goes for over $900:
Consumers need to start playing legal warfare against the companies for openly distorting the market. The ramifications will only hurt us and there needs to be a true comeuppance.
I bought a server with 768GB RAM a couple of years ago, for cheap. Today just the RAM itself is worth more than I paid for the entire server. DDR4...
Same thing with SSD, two years ago you could buy 4TB SSDs for like 250 USD, and today its more than $700. Madness!
I built a little home server last year, and only put 32 of the total 64 GB I wanted. The rationale was that RAM is cheap, and I could spend that month's fun budget on a 8TB harddrive instead.
Boy do I ever regret that. Every time I compile some code and the VM I use goes OOM, I die a little bit inside knowing that less than 100 bucks or so would have fixed this if I just went for it.
This is insane. Didn’t know how bad it got. I bought a mini PC a few years ago with 64 gigs in it for a home VM server for like $600 total. Looks like I’m keeping it a while.
Usually these bottlenecks lead to a price crash later. Of course that’s also part of what fuels the bottleneck. Companies are afraid of over investing in production and being left with underwater capital later.
Here's another data point for you all. Last year in April 2025 I bought 32gb DDR5 at $123.69. I checked it this most recent April and the exact same product was $679. All prices in Canadian dollars.
It'll calm down once the Antrophic and/or OpenAI IPO's are done, no need to protect themselves from people running local models by buying everything once the bosses have gotten their money.
OpenAI and Anthropic are certainly strong drivers, but there's a large demand from many other players: cloud provider, accelerator vendors, and so on. I think there's no end in sight.
This is textbook negative externalities, of the AI buildout on everyone who isn't using RAM/GPUs for AI, of the use of electricity and water on anyone who isn't using it for AI. The cynic in me thinks this will go down in history alongside asbestos, leaded gasoline/paint, and the opioid crisis.
Like clockwork, people naturally want to have their cake and to eat it too, so there will be the incessant complaining about the externalities. Half the people lack the brainpower to see the good and bad are intrinsically linked, and the other half just like complaining.
But at least for now, both halves aren't pulling back (in fact it's increasing), and money, not complaining, steers the ship.
We can be cynics of AI without ignoring reality, if no one wanted this no one would be chatting with Claude or ChatGPT directly, but people obviously are.
The fact is there are people that do in fact want this, and it isn't just CEO's hoping to cut jobs.
There is certainly a lot of demand at the current price of free or subsidised subscriptions. It remains to be seen what the demand is at profitable prices.
If the vendors decide that free (ad-supported) use is necessary to keep demand, we will be entering a new era of surveillance capitalism instead.
It's very, very questionable if people want the situation we have. I have yet to meet anyone in person who is really excited about AI. Of course it's useful, but at this cost?
Claiming people want this is like claiming that people wanted WW2 because look we're all enjoying the tech that was developed during it!
I unfortunately have met a few. I have one friend that legit scares me... we saw how people reacted to o4 being discontinued.
Though I do agree that most people probably don't want as much AI as is being shoved on us right now, there is a subset that do want at least some of it.
More my point, yeah I think there is an issue of the actual demand being extremely over estimated due to shady practices (like of course Gemini gets a lot of use when every single google search calls it whether you want it or not). But we also should not be so quick to disregard there being real demand just to hope for the outcome we want.
It's like how some people like listening to Ed Sheeran. So yes, there is demand. But nobody is "getting real work done" with these toy AI models.
Real AI is a geo-political threat, and will not be allowed to exist for the average person. So, enjoy your toy AI models, because that's all you're getting.
>The cynic in me thinks this will go down in history alongside asbestos, leaded gasoline/paint, and the opioid crisis.
Can you elaborate? Leaded gasoline is estimated to have contributed to the deaths of like tens (hundreds?) of millions of people. Asbestos probably millions.
Why would high RAM prices be remembered alongside these events?
Honestly that seems slightly down even if it’s still ridiculous. The ram I bought for $100 a year ago was $500 a couple of months ago. Could just be the particular sticks I got though
$375 is the cheapest kit, and the price is using a promo code
"Price tracking courtesy of PCPartPicker now reveals the cheapest 32GB DDR5 RAM you can buy is $375. Specifically, four XPOWER kits from Silicon Power will set you back $374.97 thanks to a promo code."
So I thought I would not be paying for that LLM BS just because you don't haven't been suckered into a subscription... I guess I was just lacking sufficient _intelligence_ to realize this kind of thing would happen.
As an ancient person, and considering my sense of inflation of other things I actually buy, This doesnt sound that high to me. At some point in my life CPU Ghz stopped rising and cores was the only thing to grow, and that seemed to be more slowly than clock rate did. So I guess I fully expected that asymptote to apply to RAM too. If you plot price per GB of ram across last 25 years the memory prices growth is barely a bump in the chart. In (nearly) no other physical product do we expect prices to plummet over time whilst simultaneously getting way better. Maybe we're just spoiled by the first 50 - 75 years of wild innovation that most new things go through before they asymptote ?
Yeah, I'm pretty ancient myself. If we look back ~30 years, we're still getting a good value at even these inflated prices for electronics. Food, housing, transportation has skyrocketed, but TVs and computers are comparatively cheap, even with RAM and storage being up 100+%.
I am also that many years old, and I don’t think your analogy holds for this particular situation.
The exact same sticks/model #’s of RAM cost $200 last year, now costs $950. I don’t remember existing CPU’s ever going up in price by 5x (300-400% increases).
This is the PCPartPicker chart that I monitor: https://pcpartpicker.com/trends/price/memory/#ram.ddr5.5600.... - $900 for 2x32GB, used to be $200 a year ago.
Yesterday I did a price check on the PC I built two years ago. It went from $2300 to $3650. The bulk of that increase was that the ram went from $210 to $940. Its now more expensive than when DDR5 was new.
The value of my desktop pc has almost doubled, my ps5 is worth ~ $150 more than I bought it years ago.
It's gotten to the point where nvidia doesn't even bother to report their gaming revenue anymore. It's a clear sign that we're back to the bad old days of pc gaming being a 'prosumer' hobby, but don't worry I'm sure nvidia and their ilk are salivating at the idea of making pc gaming into a streaming stadia like solution that you pay for monthly
https://pcpartpicker.com/user/Sohcahtoa82/saved/n76zkL
I think I paid a total of around $5,500 for the current components of my PC. Hard to say for sure since my PC has been a Ship of Theseus for over 30 years and started as a 486. The link merely reflects its current state.
At one point, PCPartPicker was showing my PC as worth $11,000. It's now at $7,200 without including the RAM or PSU. That would put it at $9,000.
> It's a clear sign that we're back to the bad old days of pc gaming being a 'prosumer' hobby
Yup.
I think it's especially bad since the gap between budget-grade and mid-grade feels like it's gotten wide. If you wanna play the latest AAA games and not feel like you need to upgrade in 3 years, you can't settle for the budget grade unless you're still gaming at 1080p.
I wouldn't recommend spending under $3,000 for a gaming PC these days, and that's just an absurd price.
Nvidia already has a PC gaming streaming service that integrates with your steam library and a couple other launchers' libraries.
Those are rookie numbers
https://stocks.sjer.red/
I thought I read that Samsung SATA SSDs were discontinued, but apparently that was a rumor and Sansung has denied it. I wonder why they exceed NVMe prices. They're the only SATA drives left with DRAM. I guess they could just be milking that fact.
I bought a 5090 12 months ago, just checked - that’s basically up 50%! I used to joke i’d retire on all the old tech in my loft, everyday now it feels less like a joke!
The memory in the PC I put together early last year is now worth about three times the total cost of all the parts I used to build the thing. It is absolutely crazy.
You could get a 990pro 2 terabyte Samsung SSD for €150 just last year. Memory had become a commodity product.
Memory is still a commodity product, in that there isn't a huge amount of difference between vendors selling products that comply with a certain technical standard. Sometimes the prices of commodities (wheat, silver, crude oil, etc) go way up when supply and demand get out of balance.
I regret not building the PC when I was looking at it. It's not a money thing, at the end of the day, but I can't bring myself to do it.
I had it all priced out, but a bunch of birthdays in my family were coming up and I felt like I shouldn't buy something for myself if it's really their time.
My old laptop will have to cut it for a while. :-)
It shows 3.5" spinning rust are climbing this year too.
We are cooked tbh.
At least it seems to not get worse.
The squeeze is real even at the SME level. We recently wanted to add another TB of memory to several servers (we do EDA chip design, which eats a lot of memory). Quotes came back to about €200k for 48 x 96GB DDR5-5600 RDIMMs. Mind you, this is for refurbished memory with 1 year warranty. I'm still figuring out if this is FU-pricing or just how things are going to be from now on.
Spec-ing and buying servers has become quite the pain in the past year, at least at the relatively-small scale we operate at. It's "dynamic pricing" with most quotes being valid for 24 hours :(
> €200k for 48 x 96GB DDR5-5600 RDIMMs
That's about 50% higher than you'd pay for retail, small quantity, off the shelf alternatives in the US.
Many vendors have limited allocation so they're sending out extremely high quotes to see who will take it. They know everyone is seeing news about high RAM prices and getting new vendors approved can be hard, so they raise prices and see who takes it.
I would recommend getting quotes from multiple vendors. Don't be afraid to push back. They might come back immediately with a lower price in this market. Also check the retail suppliers you can access in your country.
I’m just working with an EDA client to upgrade their almost decade old machines to run the cadence tools…it’s grim. Real real grim. I was pricing out servers this morning with 64gb(!) of RAM for almost 20k. The machines were running now, some of them have almost a terabyte of RAM. I think the designers are just going to have to suck it up and use the slower machines
If you don't need the full memory bandwidth of DDR5 for any particular application area, the best solution is probably just to add flash storage for mostly read-only data and Intel Optane (still sold in the secondary market at around $1/GB) for write-heavy scratchpad space. The latter gives you around DDR3ish performance at a higher power draw than DRAM or modern flash, but its wearout-resistance and lower latency compared to flash still matters.
> Quotes came back to about €200k for 48 x 96GB DDR5-5600 RDIMMs.
That's around EUR 40 per GB? That seems quite high compared to what consumer DDR5 RAM is selling for, though it being RDIMM may account for some of the difference.
Micron is building a factory near me but it's years out.
The only hope here is China floods the market which would be a good or bad thing however you look at it.
That’s about right, sadly. We were buying servers with 1TB DDR5 standard, no longer.
Quotes are 7 days max, lead times fluctuating out months, and often now have language they can choose to not honor the quote for any reason due to price fluctuations.
I got quoted $4K, $6K, and $8K for the same enterprise SSD part in three different quotes within about a week. (I normally expect this part to be $1.5-2.5K.) In one case, a storage array with two SSDs in it cost approximately the same as buying those two SSDs standalone.
When it comes to server RAM, I got quoted double a previous quote within the span of a week.
Buying servers is basically quote roulette now.
I still believe that Mark Zuckerberg would have been smarter investing all his AI funding into just making RAM fabs. Would probably have printed easy cash. Instead he's burning cash on making IG / FB less secure because of his in-house AI.
That would require having actual technological flair beyond 'buying potential competitors'
Indeed... I am starting to be convinced he's only good in terms of running social media apparatuses.
> Quotes came back to about €200k for 48 x 96GB DDR5-5600 RDIMMs
I sell used RAM in trays of 50 sticks (I physically know how much that is, I've held that many). I've had some trays sell for a few thousand dollars. Six figures for just one tray is absurd, but I've never collected enough DDR5 to make a full tray.
GN did a documentary on the situation from the perspective of consumer-facing companies. Seems pretty dire for them, and it's hard to see the long-range consequences, but the idea of consumers being priced out isn't too far out, which to me is a little alarming.
https://m.youtube.com/watch?v=zyQwAhppWj8
[delayed]
Consumer oriented parts businesses are going to go out of business from this. Motherboards, cases, coolers, power supplies, fans, etc. At this point it's a matter of how many. Selection and competition will be permanently reduced, as I don't see these types of companies coming back any time soon.
Who needs an economy built on consumer spending when a handful of massive tech companies can just keep passing around the same trillion dollar check like a hot potato?
I want a more complete picture of why prices are so high from articles like this.
Is supply actively constrained, or is this mostly in anticipation of future shortages? How much of this is a mix of panic buying and price gouging on bad news?
I care more about the secondhand market. Prices are nuts for old used gear, but that also tracks with patterns I've been seeing since roughly the pandemic where more and more secondhand sellers on the usual platforms setting pricing patterns are small businesses, not hobbyists.
Supply is both already constrained and AI companies have pre-purchased enough HBM at enough of a premium that most of the wafers are allocated to them. All the intermediaries are jacking up their prices so their inventory doesn't empty too quickly, because they may not be able to refill it.
It is hard to overstate the damage that the infinite money being poured into AI is doing to the wider economy. Anything involved in datacenters is going to experience shortages/price rises. A pre-existing problem is power transformers: https://www.reuters.com/business/energy/us-power-transformer...
The impact on domestic electricity prices in a year just after high oil prices is not going to be popular either.
> It is hard to overstate the damage that the infinite money being poured into AI is doing to the wider economy.
I get what you're saying but medium term this is an extremely funny sentiment. This money being poured is likely to end up being a huge boon for a lot of economic sectors, including in the US. Most commodity shortages like this end in a glut, with a medium term win for consumers, even if we have 1-2 (more) years of pricing pain. Meanwhile expensive RAM has so far left stock for people that really need it. Calling this kind of demand economic damage is odd.
> Most commodity shortages like this end in a glut
The 3 RAM manufactures know this too, from painful experience. There won't be a glut this cycle because there are no capacity build-outs. Instead of increasing capacity, some OEMs left the consumer segment to focus on enterprise AI.
> Calling this kind of demand economic damage is odd.
It's not odd at all because the complementary industries are being damaged - possibly permanently: manufacturers of motherboard, cases, fans, and the entire consumer PC supply chain are being negatively impacted. Expect bankruptcies and consolidation if this lasts for 2 more years.
Good that there is a _fourth_ one. That's actively ramping up and looking to increase market share. CXMT.
CXMT is a tiny supplier, it will take years before they significantly expand the market. They're pretty much in the same boat as Samsung and SK Hynix, they just have less of an incentive to actively curtail supply.
> I get what you're saying but medium term this is an extremely funny sentiment.
Sentiment is the right word here. None of us really know, and go by feeling. If you perceive the AI boom as approaching tulip craze levels of irrationality, it feels pretty dire.
The RAM is a commodity and may be repurposed afterward. This kind of thing is a bit like a debt jubilee when the dust clears and survivors scavenge the dead. But a lot of other build-out may essentially be waste. To me, apologists for this boom seem to be harboring a variant of the broken windows fallacy. Not all economic activity is productive.
The other kind of damage is opportunity cost. How many players in other industries are being strategically harmed by this situation? We can't all just live on AI token output if these other industries retract too far.
> This money being poured is likely to end up being a huge boon for a lot of economic sectors, including in the US.
No such thing as a free lunch. Whilst a lot of industries are doing extremely well right now (i.e. everything involved in datacenter construction), everyone else has to actually pay for it.
Therein also lies the damage; The economy wasn't doing so great to begin with, and now this massive multi-trillion-dollar expense has been added.
The direct expenses come out of the pockets of Big Tech, but all the indirect stuff like the DRAM crisis affect the wider economy directly.
It's not odd at all to call this economic damage unless you're that disconnected from anything but AI/high-tech SV companies.
There's smaller businesses that I know _need_ data capacity (hard drives) that can't afford them and are facing serious CAPEX challenges.
I work for a start up that helps folks move off of existing on-prem virtualization solutions and every small to medium company across every sector you can think of aren't just able to not afford additional storage and memory, they can't find them either.
There isn't as much stock as you think. I can't wait for this AI bubble to pop. It's been absolutely terrible for 90% of the industry.
Yup. Once production ramps up (and then subsequently overshoots) in the next couple years we're going to have GPUs with 96GB VRAM for the price of a 4090 today.
They're not ramping up capacity though.
https://www.techzine.eu/news/devices/141775/sk-hynix-to-doub...
CXMT will.
> This money being poured is likely to end up being a huge boon for a lot of economic sectors
The evidence coming out is saying otherwise.
> It is hard to overstate the damage that the infinite money being poured into AI is doing to the wider economy
I know it's an unpopular opinion on Hacker News but I think most people are overstating the damage to the economy. Increased demand starts to pull more advancements forward and increase spending on production, which benefits everyone.
Transformer demand is real, but what percentage of your electricity bill do you think goes to spending on those transformers? The number is so small that it's a rounding error. Many examples like this where we see headlines about some part going up in price and forget that it's such a negligible piece of our bill that it barely matters.
The cost of fuel inputs for base generation and peak supply are a bigger factor.
> I know it's an unpopular opinion on Hacker News but I think most people are overstating the damage to the economy.
There are two issues... one being that an economy needs people to buy things to work, and people only can buy things if they have money, and people only have money when they have work - but everyone is laying off people left and right due to "AI".
The other issue is that there's trillions of dollars floating around between all the companies. It is likely that a significant chunk of these will fail and give us another 2007.
Memory makers are reluctant to increase fabrication (high cost of capital and boom/bust cycle leads to bankruptcies being common in the industry), and the memory required for AI (HBM) requires more of the production inputs than other types of RAM, thus squeezing regular DDR RAM even more. Long-form article describing this in more detail: https://davidoks.blog/p/ai-is-killing-the-cheap-smartphone
> Is supply actively constrained, or is this mostly in anticipation of future shortages?
Both. AI datacenters create immense demand on DRAM, NAND storage, even HDD storage.
Then Sam Altman went ahead and "bought" 40% of the world's DRAM production, by way of secretly approaching both Samsung and SK Hynix (each unaware of the other getting a similar deal), which sent the entire market into a panic and everyone rushed to buy ahead of the anticipated supply shortage.
The kicker is that it wasn't even a proper purchase agreement. Just a non-binding promise of "By 2030 I will have infinite dollars and buy your DRAM".
Supply is constrained. AI bought up so much that we're seeing wafer shortages. Absurd US tariffs messed with stuff. And we're just getting started.
I looked at my eBay receipt from 2023 and I paid $84.98 for a "Kingston FURY Beast 64GB (2x32GB) 3200MHz DDR4" listing and now the equivalent on eBay "Buy It Now" is $374.99 for "Kingston FURY Renegade 64GB (2x32GB) DDR4 RAM 3200MHz (KF432C16RBK2/64)". What a timeline it has become for consumer computing three years later!
I was picking up DDR3 16GB sticks for $5/piece on eBay last year. The world has gone mad.
My anecdata: bought used DDR4 ECC 16gb ram sticks (i.e. serverpulled ram) for $11 off ebay last year (Jan 2025), now as of June 2026 the lowest listing I see is $42, most are around $50.
When I bought the same ECC DDR4 16gb ram stick in 2016 (when DDR4 was rather new) it cost $85 (cant remember if it was new or used)
To be fair, DDR3 is still kinda cheap. It's really only DDR4 and DDR5 that are massively in demand. DDR3 is a bit too old for it to be in high demand by consumers.
Unbelievably, even DDR3 isn't spared. Our suppliers have been quoting us much higher prices for DDR3 for months now, and we're experiencing persistent stock shortages. Fabs are abandoning legacy nodes to prioritize AI capacity, making it rough even for maintaining old gear.
It isn't really unbelievable. Anyone who could sub in DDR4 for their projects would do so which increased the price. And then anyone who was still using DDR4 for projects would respond to the increased price by moving to cheaper DDR3 if they could. It wouldn't take much of an unexpected demand increase to move the needle since I don't believe much DDR3 or DDR4 is even being manufactured anymore. And then you have the supply side: Increased prices of upgrades will cause many of them to be delayed or cancelled denying the refurb/used market supply of the old hardware.
Personally I have a lot of DDR3 RDIMMS laying around that I got for cheap/free and have thought of getting a used workstation to put them in (4 channels of DDR3 will still net you ~60 GB/s of bandwidth) but the only Xeons that support it are Sandy/Ivy. Everything Haswell and newer with AVX2 and rebar/etc. use DDR4.
I wanted to buy another one and they are like $90 now :(
While DDR3 is cheaper it's still tripled or quadrupled in price over the past two years. I just bought a pair or sticks for an old Mac Pro and it was 4x what i paid just a few years ago.
I work in the refurb department of an e-waste recycling company. I've been collecting some DDR5-based systems for months, but with the prices, they're probably more than what most people are looking to spend on a technically second-hand computer (even if they are like new and still under warranty). I've priced them at about what the CPU + RAM + SSD would sell for separately, and I'm not willing to go much lower than that.
And if you're wondering, who's throwing out DDR5 systems? A local healthcare company. The boxes for some units are crushed and have water stains on them, and I imagine others don't meet their exacting requirements in some minor way (though they look and work OK to me, regardless of scratches and dings on the case).
>they're probably more than what most people are looking to spend on a technically second-hand computer (even if they are like new and still under warranty)
Some people need to buy at these inflated prices anyway, and are ultimately willing to spend. Maybe it differs in your area but there's no risk in listing these on Facebook Marketplace or similar platforms and see the interest.
I've had them listed on FB Marketplace for months. In that time, I've sold four (3 to one guy, one to myself), and still have 20-ish more.
i5-13500E, 32GB DDR5, 512 GB NVMe, scratched and dinged a bit for $500 is practically a steal! Hardly anyone's biting or clicking, I don't get it...
New open box models (not scratched and dinged), if anyone's interested: https://www.ebay.com/itm/188355326179
Edit: they're intended to be retail POS PCs, like cash registers. (The motherboard literally has a header for a cash drawer!) So unless you're a particular kind of retailer, you're not asking your distributor for them. If distributors aren't looking for them, our B2B buyers aren't buying them. If they were Elite/Prodesks, Optiplexes, or Thinkcenters, they'd probably be gone by now.
Why don’t you part them out? The market for the parts will be more liquid. You might sell faster and at a higher price. You could try doing it once and see how it goes.
I've seriously considered doing that. Since they're still under warranty for another year or two, I've decided not to... for the time being.
I assume you're not in the Seattle area?
Edit: Pittsburgh
That assumption is correct.
I'm sure it doesn't help that people continue to buy things at this price.
Steam Deck had a huge price increase (~40-50%) but it still sold out in 24 hours.
All it would take is for everyone not to pull the trigger on buying things for a little bit and prices would fall but instead enough people are buying things at a crazy markup. If anything that's a signal to sell things at higher prices. Of course AI is amplifying this problem but realistically people are still buying consumer hardware at these prices which lets businesses know people will pay this price.
I'm on a machine built from parts in 2014 and it's all very good for me to do every day development so I'm not posting this from a machine I won't have to touch for another 10 years.
I believe Steam Deck is a major exception. Given how fast they sold out I also wonder how many Valve produced this batch. Most vendors of PC parts are seeing steep declines. Also after seeing 2-3 price increases in everything (Switch, Playstation, XBOX, etc.) people are likely getting it now because the prices aren't going down anytime soon and will probably just go up again for a couple of years.
I pulled the trigger on an early Ayn Thor because it was obvious this was going to happen. Something I didn't really want to fit into my budget but knew that if I didn't I would regret it later.
I doubt it helps but this is such a small piece of the pie I'm not sure how much this hurts things either. Steam Deck is kind of a niche in a niche and doesn't sell in huge numbers compared to other players like PC OEMs and phone OEMs that are now all over a barrel as well as OpenAI tries to buy all the RAM so no one else can have it.
> I'm sure it doesn't help that people continue to buy things at this price.
Implicitly, but that's blaming the consumer who has no or few equivalent choices. Purchasing RAM is not like choosing between Coke and Pepsi. A better analogy is that when a hurricane is coming or a natural disaster has already hit, it doesn't help that people will purchase food and fuel at any price.
It's unbelievable and it's only getting worse.
A 2x32GB DDR5 kit I paid $150 for 11 months ago costs $910 today from the same retailer. A 2x16GB DRR4 kit that was $105 last year is now $230.
The RAM alone in my newest machine would sell for over double what I paid for the entire machine one year ago.
"The costs are negligible and justified when compared to all the benefits. If you look at the performance gains, the overall cost has in fact been reduced."
- Altman, a.k.a. Dory from Finding Nemo and/or Dario, a.k.a. Carl from Jimmy Neutron.
I haven't watched those movies in a while. Can I get an explanation on those a.k.a.s?
I remember both, the former is positive but suffers from short-term memory loss. the latter is always afraid of things, and also seems to never learn.
My bet is that the prices will crash once OpenAI (and/or Antrophic) IPO's have happened.
Right now the biggest threat to their IPO's is that people realize that local models are good enough for whatever they're peddling, what's the most important factor to even running good enough models? RAM since you want the models in memory to not be total slogs.
But remember that markets can stay irrational longer than anyone can hold his breath. If they get more funding there's a good chance they'll invest more in the destruction of the remaining production capacity. Admitting that with normal pricing anyone could have a decent AI-machine for 2K is hard - prices for acceptable AI-machines most likely will go >10K first.
You are saying there will be even more demand for RAM and that will cause the prices to crash?
The chance is that they cache out during IPO, and will lose interest to increase capacity, some/many contracts will be canceled, and demand being reduced.
Perhaps all hobbyst developers interested in local AI combined is a smaller demand than AI companies hoarding parts. That would make demand decrease.
If local models are good enough, doesn't that increase demand for DRAM as everyone buys DRAM for their poorly utilized local machines?
Surely it is a more efficient use of DRAM to run inference on shared hardware with large batch sizes and more utilization.
Luckily very few people can configure and are interested in local models. But your nearby datacenter running Chinese open-weight models is also good enough.
My point is that dram demand is mostly orthogonal to whether everyone is using open weight models or secret weight models. Heavy demand for local models (whether secret or open weight) will require even more aggregate DRAM than for shared.
Demand will only go down if people reduce their use of these AI tools. Given how much folks here complain about quotas, I'm very skeptical that will happen willingly.
Open weight models allow for repurposing existing hardware locally, and there's a lot of it around - far more than the amount of new RAM being supplied. So they add some short-term downward pressure to the price. (But not very much, since these datacenter builds are long-term investments that are targeted at eventually running far larger models.)
If regular people can repurpose old hardware, so can shared providers, who can extract more value from the hardware and thus afford to pay more.
In a constrained market, supply and demand favors folks who can most efficiently extract rent. Local models only make sense in a world with abundant compute and energy.
My bet is that we're not gonna see any adjustments in RAM pricing until one of the planned data center projects collapses in a spectacular way.
The billionaires locked in a race to spend effectively unlimited funds on AI CapEx will have to be convinced by markets and/or their advisers that there aren't enough profits and that cutting losses (like with Metaverse) in their quixotic quest is necessary.
One theory: they will need to throw away all these Nvidia cards in the trash at some point right ?
Because what to do with power-consuming outdated hardware ? let's say 5 years from now ?
They will need new RAM.
I wonder.
I’d gladly take a few of these self-contained rack-clusters off their hands when they do.
I’d even get a house with a garage or something just for that.
And honestly, we will have much bigger problems if that bubble pops in a spectacular fashion.
Which problems would that be? Nasdaq crashing by a few percent and a major player to go under. Seems almost inevitable at some point.
And taking everyone's 401ks down with them because of the idiotic rules changes to accommodate SpaceX? Its going to be the largest transfer (theft) of wealth since 2008. AI is the only thing driving any kind of growth in the market at all now. That pops, its going to bring the entire economy down with it.
The vast majority of index funds are float adjusted. SpaceX will not have that many shares available relative to the total value of the company. It's a negligible percent of the overall market cap of say, VTI. This commonly repeated trope is misinformation. The change in rules is more of a problem than the actual value.
> that local models are good enough for whatever they're peddling
they are not. Unless you are satisfied with plausible, but mostly garbage output.
They are actually quite a bit better than you might think. Qwen3.6 27B is pretty capable at coding.
For non-coding work, they are more than good enough. A lot of the ways my non-technical family members have interacted with AI would be perfectly served by using a local model.
After all, people were more than satisfied with the results from GPT 3. That has long since been surpassed by open weight models.
I'm sure there are things local models are good enough at in non-coding work, but for anything complex I do not find this to be the case.
I'd say local models are fairly capable of even somewhat complex coding execution. For complex non-coding work (research, in-depth analysis, assembly of complex info-dense documents) I'd rather do it by hand than switch from Opus 4.7 to anything I could even theoretically run locally.
I don't know what kind of coding, but for my case it's been useless. Not working code almost every time. It's much quicker to just write it by hand than use that model.
I've been experimenting with Qwen3 Coder Next and Copilot for a little Rust toy project and it's been trucking along. It does require a fair bit of hand-holding (or perhaps I just don't trust it to give it larger tasks), but it works alright.
I get results comparable to the saas. Maybe Anthropic sold you too much crack tokens.
isn't that literally all output an LLM generates?
Honestly, that's the output I get from non-local models, anyway. If I'm going to get plausible nonsense either way, I may as well run it on my own hardware.
Same thing with storage.
I wanted to upgrade my SSD but prices are more than at the end of 2025. I refuse to pay 500 euro for a 4TB SSD. I rather go outside and play with my bike like when I was 5.
Even spinning metal has gone through the roof even on refurb and server pulls.
Settle for a modest 2TB and spend the rest on ice cream during your rides.
getting a road bike is understated, at least where i live it is so nice to roam around outisde cities with just a few cars around
I tried to buy new SD cards for my camera. The cards I used to buy at $28 are now $80-120... if you can find them. Another cheaper card I used to buy for $19 is now $46. It's just absolutely insane at this point.
I'm playing the newest games on ddr3 with a 2080 and a 4790k. It's a simple life.
Presumably, this is at 1080p, which looks terrible after you experienced 4K. It's better than nothing, and I probably wouldn't build a 4K rig in this market, but let's not pretend you can get a modern gaming experience out of that old hardware.
Here I'm on a 5800x and a 4070ti super and it's woefully inadequate for some games. Though I suppose if you play at 1080p (which is hard to look at these days for me) you might be OK on low-medium settings in some newer games. I can't see you getting very far with something like Forza Horizon 6 or Death Stranding 2 though.
2080? I'm still using a 1070. That thing rocks
<FourYorkshiremen>Luxury.</FourYorkshiremen> I'm still using a 1650 Super with 4GB VRAM and it's basically fine. Holding off on a few newer titles, but I'm old and my eyesight is going so 1080 resolution is plenty for me.
1070 as well. Probably one of the best ROIs from a single component
Some might call this a boomer build but the 4790k is an absolute beast of a CPU and still holds up.
I recently booted up an old 4790k system and it was fine on Linux but on Windows it would nag me to update but apparently the CPU is too old for new Windows. I ended up giving it away on the Internet to whomever could pick up but afterwards it ended up with one of those reseller chaps. Ah well, I wish it had made to some kid somewhere.
That makes sense. I'm on Windows 10 LTSC so I don't get nags and scare screens from Microsoft as much as a regular machine might.
> Windows it would nag me to update but apparently the CPU is too old for new Windows
When I was trying to upgrade to Windows 11 from Windows 10 I got the same warning. It is the reason why I'm now happily running Xubuntu on the same machine.
I did end up installing a Windows 11 VM (VirtualBox) which works perfectly fine running as a guest under Xubuntu on a 4790k (screw you Microsoft). But I otherwise never want go back to Windows as a daily driver. The enshittification just got too much for me.
I got a 4700k in 2013 for 200 bucks from Microcenter. The thing has been continuously in service ever since. It's my "homelab" box now.
it's also a capable local inference stack!
I've got a 9 year old Xeon W and 64GB of DDR4. Its not as fast as some modern DDR5 stuff, but boy does it work
I played 100+ hours of RDR2 on a 2060 (non-super), as was the style at the time. When the 30 series came out, I sold that card for more than I paid for it.
AMD just brought the popular 5800X3D back out of retirement to give people maintaining the DDR4 based platforms something to buy. Last I checked used DDR4 was half the price of used DDR5 after the prices of both shot up.
Shame I have 2TiB of ECC DDR4 lying around :(
Would be nice to be able to own property.
Better to get rid of that stuff now unless it’s fast. You can get used 2666 DDR4 for about $200/64 GiB stick.
You can probably get a mansion in France for that.
People used to mock "you'll own nothing" as conspiracy theory.
Everyone re-releasing their old crap. Nvidia is remaking the 3060 too I believe because they can use cheaper older RAM on it.
I'm just waiting for Zhaoxin to break into the international market with something
I know what you're meaning but the 5800X3D is still a beast of a gaming CPU. It remains a great option for AM4 gaming.
It's not even "old crap" the PC consumer market has basically died in less than year due to your average person getting priced out, I cannot imagine how catastrophic of an impact this is going to have on younger folk and the barrier to entry to desktop computing
A laptop with a 226v/256v cost around the same than the "cheap" Mac Neo and can play most games (old games native/new games with AI scaling)
Nice, I missed that.
The 5700X3D has been the smarter pick back then, it fits to the current latent user hostility of AMD to focus on the more expensive processor.
That's literally the same chip. With pricing being whatever, who cares which one they revive.
It also was the same chip back then and the price difference significant. But maybe you are correct and with the current conditions it doesn't make a difference. I dont know.
They need to sell something. I bet sales of DDR5 related stuff are down.
Time to migrate all those electron "native" apps to actual native code. I bet with some decent optimizations 4GB will be more than enough for casual user, and still with some free memory to spare.
Pretty insane. I built a Framework Desktop PC back in November. The motherboard (with 128G DDR5 RAM) was $1800. Now it's $2859. Almost 60% increase in 6 months.
Crucial has "good" prices for DDR5 6000mhz memory on Amazon. The downside is you have to wait for shipping -- I had to wait a couple weeks. I bought when these were $300 and equivalent memory prices were like $500 at the time.
Been solid so far for 6-ish months.
https://www.amazon.com/Crucial-6000MHz-Overclocking-Desktop-...
The demise of personal computing. All according to plan
To me, this is just one more thing that makes the current times we live in "not fun".
All started by openai wafer capacity commitments that aren't being met anymore... The ECC ram systems I've bought are now worth more than the original purchase price of the entire system and I've been debating on just selling them since they're now worth nearly double their value which would outperform any stock trades I've ever made any given year lol.
I am thankful that both my partner and myself are in a pretty good spot when it comes to our gaming PC's. I had hoped to double my RAM at some point but I am still at a comfortable spot.
I am annoyed that the new handhelds are all crazy so sticking with my Legion Go for now.
The one I am annoyed with is storage. I desperately need to get a couple new drives for my NAS (one to replace one that its bad sectors are growing and one to add more storage) and I am not looking forward to spending $600-$700 each for 20TB drives.
GPU prices have gone 5-10% down in the last month.
They are still too high. You can't buy any desirable GPU like a 5090 at MSRP.
Price should send a signal to manufacturers to build more capacity. I wonder if they will though, it takes quite a bit of time, and it's not certain that the demand will continue to exist once built.
Several Chinese manufacturers are doing just that, and have already expanded production: https://techwireasia.com/2026/04/chinese-memory-chips-ymtc-c... But because of tech trade barriers their primary focus is on the domestic market and only secondarily global markets.
Fortunately RAM is largely a commodity and more for sale only in China means more for sale elsewhere.
We are down to 3 manufacturers thanks to RAM boom and bust cycles. The remaining ones are the one that know how not to over invest.
With China’s CXMT and YMTC it’s now going to be five again.
A fab for high end memory costs $20B and 5 years to build.
It will happen, but yeah it takes time and money
$20B is like pocket money at the scale the whole industry is moving.
It will only happen if the bottom doesn't fall out of the market for AI datacenters in the next few months.
It's coming up on a year since the crisis started and at every single point it's been "next few months"
Sometime I wish China had capacity to manufacture RAM. They would build fab within 1 year..
Am I missing something?
https://techwireasia.com/2026/04/chinese-memory-chips-ymtc-c...
https://wccftech.com/another-chinese-dram-maker-breaks-into-...
They do, it’s called CXMT and they are making RAM under contract for Corsair
Commerce Secretary Howard Lutnick has warned Samsung and SK Hynix they could face 100% tariffs, framing it as a choice between paying a 100% tariff or building memory fabs in America...
Punishing instead of supporting.
Do tariff threats still matter? After being struck down, I thought the only new tariffs which could happen would have to be enacted by Congress? Or world that be a fun game where the new illegal tariffs would be on the books until the courts invalidate them again?
Who cares about America. American tech bros created the problem.
Also YMTC.
So far haven't we seen the opposite? Consumer focused ram production shutting down to make more volume for server dimms or etc?
The bottleneck isn't the sticks, it's the chips. The chips are the same for consumer and server applications. What's been happening is that big companies have bought nearly all the wafer capacity for the next year or so, and perhaps some of that capacity has also been redirected from DDR5 to LPDDR5. If a stick manufacturer drops out of the consumer market that kinda doesn't matter, because manufacturing sticks is relatively low tech compared to manufacturing the memory chips. You can compare it to manufacturing video cards vs. manufacturing GPUs (as in the actual processing elements).
The fabs are the same and are the actual bottleneck. The chips are different for CPUs and GPUs.
What do you mean? RAM manufacturers shifting their capacities to server RAM and HBM is exactly what’s happening.
Government needs to get out of the way. Micron announced a memory fab in Syracuse in 2023. It took 3 years, 20,000 pages of "environmental review", deals with the government on amount of union contracts during building, etc. for them to break ground in 2026 for a 2030 opening date. In any reasonable world, a 2023 announcement should have broke ground in 2023.
OTOH, a celulose factory near me, built in the 1950's, got their permits fast and with little regard to environment. FF three decades, and their entire surroundings are destoyed for everyone else. Trials go nowhere, because they have all authorizations needed (and a lot of political leverage because they are the main employer in the region). Careful fast-tracking business that have zero incentives to avoid externalization of costs.
Oh, yeah, we are in this calamity because of government interference, not unbridled capitalism. Sure.
Buddy they ain't building an ice-cream parlor. 200 miles of the Hudson river is a Superfund site. The biggest polluters, PCB's, lead and mercury.
The main blocker was that there were bats there so they needed to buy separate land to preserve. 20k pages of environmental review is just make work to spend money and create an unnecessary paper trail. If polluting with x is illegal then its illegal. The review doesnt stop that.
And? The primary goal should be to catch and stop pollution, not make manufacturers spend years promising not to do something they're not allowed to do. If someone wants to build a factory that can't operate without illegal emissions, then so be it. It's their money lost. All that matters is that they don't actually pollute.
Using red tape as some kind of prophylactic is ridiculous. If the state doesn't have the monitoring in place, you have to just trust the company, which is naive if not negligent. If you do have the monitoring, why require the extremely expensive song & dance? To protect corporations from negligently wasting money?
Answer: because the song & dance is primarily about extracting concessions, like union labor or even cash (e.g. promises to pay to fix someone else's pollution, or contributions to various interest groups). The friction and expense involved in today's development review processes are many times more costly to all involved than the social benefit.
The process is there because Industry has proven that it can't be trusted. The only way to stop it is to verify that it won't happen in the first place by making sure their building plans are up to par. The song-and-dance, well even with the review, they try their damn hardest to cut corners and hood-wig wherever they can.
> If someone wants to build a factory that can't operate without illegal emissions, then so be it. It's their money lost. All that matters is that they don't actually pollute.
That's hopelessly naïve.
If you let them build the facility that can pollute, they're going to pollute.
And if you point to the pollution coming out and tell them "you have to stop," they're going to say "make us."
And if you point to the pollution already in the environment and tell them "you have to clean that up, because you put it there," they're going to say "prove it."
And they're going to tie the government up in court for years or decades, and then oh, whoops, somehow the entity that actually did all the polluting has no more money and can't do anything about it :-( Good thing they were only a subsidiary that all the profit and assets can be moved out of!
And the people who actually live there are suffering from preventable diseases and dying of cancer at rates 5x the national average.
How do I know all this? Because this has been industry's playbook for over a century.
First of all I disagree that it's difficult to get injunctions to stop an activity that was illegal from the start. In fact, sometimes environmental reviews can backfire because they typically require affirmation by the government, which can create a defense to doing something that would otherwise be judged illegal. That type of loophole is why people are so cynical.
But even so, how does the song & dance prevent any of that? It's not like, e.g., a battery manufacturer submits a plan admitting that they're gonna dump stuff.
The plan details how those chemical storage tanks are designed and constructed. The monitoring, in place, the contingency plans.
It raises the stakes, Obviously they can still cheat but now it's a matter of criminal negligence not civil law.
Sure, but in terms of expenditure and especially time (and time is a huge expense because of the cost of raising and securing financing) that work doesn't make up the bulk of the cost. There's nothing wrong with impact statements themselves. You need independent assessments of hazards beyond the well known ones like chemical storage, where in most cases existing codes and regulations are more than adequate. It's that in many places, California especially, there's no fixed goal post. The review process has effectively become a vehicle for local and even national politics to play out. And a method for detractors to spread FUD (again especially in California where anyone, not just officials or local residents, can challenge and drag out review of an impact statement itself, let alone proposed mitigations). Just look at what has happened with CaHSR (they're still not completely done getting write-off on all the necessary impact statements for the planned route) or the politics involved with approving offshore wind farms, where Trump and, to a lesser extent, the Biden administration showed how much pure politics plays into things and drives costs up. (And federal EIP review process is, at least historically, much simpler and less subject to political games compared to some state analogs.)
The maxim, "a nation of laws, not of men", is applicable here. If you don't have fixed goal posts or rules, governance becomes chaotic, not to mention unjust, inefficient, and ultimately corrupt (so many people like Trump because he promises to substitute his own judgement in place of democratic processes).
It would be nice if we could comprehensively hypothesis and address every possible manner in which things could wrong, but we can't. Most of the time you need to rely on broad rules, at least as a catchall, like don't harm someone or don't pollute, and ensure fast, consistent, and efficient accountability when injuries occur. The current state of development approval has become grossly degenerate in many places. In others it could definitely use some firming up. But let's not pretend that a project in California would get away with most of the stuff they can in Louisiana just because we reformed the review process. For one thing, California is much better about enforcing existing hazard codes at both the planning approval and operational stages.
Why manufacturers would build more capacity to decrease the price (and profits)?
This is similar situation to housing market. Prices are going up and supply is being restricted by whatever means.
It will be a bit of Catch 22.
It's amazing that after all these years, a famous criticism of Neuromancer seems to have been mooted. That is, the bit about Case having that stash of stolen RAM that was his "big score" for the moment, and how Linda Lee stole it from him, yadda, yadda. For years people have read that and said something like "WTF? RAM isn't valuable enough to be a black market commodity".
Well... I guess William Gibson laughs last, after all!
Crazy... I hope this is temporary. If this is the new normal, we're all going to be priced out of computers eventually.
If the industry had a say, we will be priced out. Look at the nvidia dgx spark. This is going to be the new norm if they have their way.
What's the theory of the conspiracy here?
How much of is this is actually due to the AI hype cycle, and what's the impact of the global energy clusterfuck that is the Strait of Hormuz?
DRAM prices were in the stratosphere late last year, pre-dating the Iran war by 4+ months.
Yeah, the prices are actually down a few pp compared to record highs recorded before the war. The war might even exert some downward pressure. It increased the cost of living, thus reducing demand for comparative luxuries like new devices. Laptop and smartphone markets are contracting heavily right now, even for devices with stable prices enabled by long-term RAM procurement.
Take memory in house Apple (the talent is already there) its time again to kick another hardware supplier out and move on ala Intel, Nvidia, AMD, IBM, and Qualcomm outgoing in 2027-2028 if you want to continue to built devices and actually sell something to the public.
They would need their own FAB. Apple just designs, it doesn't really make anything, they sell devices where all the parts are manufactured by others.
250 EUR (that is with VAT for 2x16 GB DDR4 [1] seems like a fair price.
[1] https://tweakers.net/pricewatch/1419292/corsair-vengeance-lp...
Maybe we can start to become a little less profligate with our memory usage.
There certainly is lots and lots of potential.
Is it just me or do I have the feeling that we have gone too far with our memory requirements? Why everyone now suddenly need 32Gb, 64Gb or hundreds of additional Gb of RAM?
Same thing with GPUs, is kind of insane having so much processing power and yet requiring more and more. What purpose for? What's the limit? Does it really really pay off such investments?
For me as a non-AI developer (I don't use any models of any kind, nor I train LLMs at all), a system with 16Gb seems to be more than enough for a vast number of applications.
Actually I was using a laptop with 16GB around 2013 already. That's 13 years ago. Memory requirements kind of stagnated after that. Obviously, LLMs are driving up memory needs now.
I have a 48GB development machine which is nice if you are running IDEs, backend stuff via docker, etc. And even just having a bunch of RAM there for file caching helps keep things fast.
Just looked for my order receipt out of curiosity, this was in Jan 23: £160 for Kingston FURY Renegade 32GB (16GB x 2) 6000MT/s DDR5 CL32 DIMM Silver
I was in the US two weeks ago, looked everywhere for 1 module 32 gb ddr4 sodimm, couldn't find it anywhere. But apparently it was pretty expensive as well (from the price tag on the empty shelf in best buy)
Crazy, the other day I looked in my local store order history and say that I bought G.SKILL RipJaws V F4-3600C18D-32GVK, a DDR4 32 GB 3600 MHz kit of two sticks.
I bought it for 82 EUR, before the whole ongoing situation.
Now the same spec costs upwards of 290 EUR, about 3.5x the original price and even on Amazon the best prices I can find are upwards of like 210 EUR (2.5x).
this is almost certainly a us tarriff/China sanctions thing rather than an AI thing. sticks here outside of the us tarriff system never really changed, I just bought like a month or two ago 128GB DDR5 for $500 at a time when US best buy was listing the same kingston 32GB sticks for $200 each.
That said, getting hold of them was hard and needed a special order.
Not just AI, tariffs also affect price.
RAM is used outside of the US as well.
Really? Wow, thanks for that!
At one point I remember DDR2 ECC coating like $150-$180
Looking at it from that frame, it seems reasonable.
Everything is relative.
I thought 128MB of SDRAM was a good deal at $100.
I also thought $479 for 32GB of DDR4 was nuts back in 2016/2017.
"back in my time", a dialup and 32 megs(!) of ram was enough for most stuff, including internet browsing.
I have no idea why a weather forecast site needs tens of megabytes of resources, and gig+ of ram for my browser, since i get no more info from it, than i did back then. Same for chat programs (how is discord different than irc? and why does it need so much ram to do so? same for slack), mail clients, etc.
Maybe it's time to kick developers to start optimizing stuff a bit, since neither they nor the users can't afford "unlimited" ram anymore.
edit: i'm not saying we need to get back to literally 32 megs of ram, just to make developers performance test their stuff on a laptop that was on sale 3 years ago in their local supermarket, i.e. stuff their users use at home.
I've taken advantage of this.. Over the last 6 months I actually made money on my old PC parts inventory, but don't plan on buying any new hardware anytime soon.
I bought 64GB of DDR4 in December of '24. Best timing-the-market of my life.
Man, I remember my first 8MB cost over $400.
It’s time to bring back all the old software hacks that were so common in the 90s
Like SoftRAM? ;)
https://en.wikipedia.org/wiki/SoftRAM
They’d have killed today if they were AI first
8G ought to be enough for anybody.
This but unironically. As long as "anybody" means "any developer dogfooding their own program"
I am counting my blessings after updating my and my wife's gaming PCs right before all of this happened.
My strategy of just buying the best quality parts within my budget when I need a new build has never failed me yet. Is pretty insane that I bought 4x 32gig dd4 ram for $400 in 2024; literally the same part I paid $200 now goes for over $900:
https://www.amazon.com/dp/B0BJ7X9P1W
Consumers need to start playing legal warfare against the companies for openly distorting the market. The ramifications will only hurt us and there needs to be a true comeuppance.
I bought a server with 768GB RAM a couple of years ago, for cheap. Today just the RAM itself is worth more than I paid for the entire server. DDR4... Same thing with SSD, two years ago you could buy 4TB SSDs for like 250 USD, and today its more than $700. Madness!
Hobbyist computing is dead for the foreseeable future. These prices are untenable.
Take your ram budget, buy micron stock, wait a few days, sell it and buy ram lol
I guess I'm old and haven't paid much attention recently, but $375 for 32Gb doesn't sound that bad to me.
But then again I remember spending hundreds of dollars as a high school student to upgrade my family's 8Mb desktop to 40 Mb.
I built a little home server last year, and only put 32 of the total 64 GB I wanted. The rationale was that RAM is cheap, and I could spend that month's fun budget on a 8TB harddrive instead.
Boy do I ever regret that. Every time I compile some code and the VM I use goes OOM, I die a little bit inside knowing that less than 100 bucks or so would have fixed this if I just went for it.
While back I recalled I had 16GB of DDR4 somewhere. I went and found it in an old bin box. It's now in my safe in case I need it for a machine.
This is the stupidest freaking timeline...
This is insane. Didn’t know how bad it got. I bought a mini PC a few years ago with 64 gigs in it for a home VM server for like $600 total. Looks like I’m keeping it a while.
Usually these bottlenecks lead to a price crash later. Of course that’s also part of what fuels the bottleneck. Companies are afraid of over investing in production and being left with underwater capital later.
It was just yesterday that SK Hynix announced to double their capacity over the next five years.
https://www.bloomberg.com/news/articles/2026-06-02/sk-hynix-...
Here's another data point for you all. Last year in April 2025 I bought 32gb DDR5 at $123.69. I checked it this most recent April and the exact same product was $679. All prices in Canadian dollars.
Regret not buying more RAM last summer. And SSD:s as well. Prices are silly.
Can we just go back to pre-AI world?
It'll calm down once the Antrophic and/or OpenAI IPO's are done, no need to protect themselves from people running local models by buying everything once the bosses have gotten their money.
OpenAI and Anthropic are certainly strong drivers, but there's a large demand from many other players: cloud provider, accelerator vendors, and so on. I think there's no end in sight.
Can we just go back to pre-{anything-here} world?
gigabit internet and the death of flash for web video has been wonderful to be honest.
There was a period in 2012-2016 when things were pretty nice.
good way to get us a in a singularity
This is textbook negative externalities, of the AI buildout on everyone who isn't using RAM/GPUs for AI, of the use of electricity and water on anyone who isn't using it for AI. The cynic in me thinks this will go down in history alongside asbestos, leaded gasoline/paint, and the opioid crisis.
People want this, the demand is there.
Like clockwork, people naturally want to have their cake and to eat it too, so there will be the incessant complaining about the externalities. Half the people lack the brainpower to see the good and bad are intrinsically linked, and the other half just like complaining.
But at least for now, both halves aren't pulling back (in fact it's increasing), and money, not complaining, steers the ship.
Do they?
> People want this, the demand is there.
It's impossible to avoid using AI multiple times a day, just because it's forced into every product under the sun.
That is NOT demand. None of those users WANT this.
Walk around a university campus at peek at students' laptops and you will find 90%+ of them in the middle of a conversation with a chatbot.
We can be cynics of AI without ignoring reality, if no one wanted this no one would be chatting with Claude or ChatGPT directly, but people obviously are.
The fact is there are people that do in fact want this, and it isn't just CEO's hoping to cut jobs.
There is certainly a lot of demand at the current price of free or subsidised subscriptions. It remains to be seen what the demand is at profitable prices.
If the vendors decide that free (ad-supported) use is necessary to keep demand, we will be entering a new era of surveillance capitalism instead.
It's very, very questionable if people want the situation we have. I have yet to meet anyone in person who is really excited about AI. Of course it's useful, but at this cost?
Claiming people want this is like claiming that people wanted WW2 because look we're all enjoying the tech that was developed during it!
I unfortunately have met a few. I have one friend that legit scares me... we saw how people reacted to o4 being discontinued.
Though I do agree that most people probably don't want as much AI as is being shoved on us right now, there is a subset that do want at least some of it.
More my point, yeah I think there is an issue of the actual demand being extremely over estimated due to shady practices (like of course Gemini gets a lot of use when every single google search calls it whether you want it or not). But we also should not be so quick to disregard there being real demand just to hope for the outcome we want.
Talk to more people?
I talk to my Uber driver whenever I'm visiting somewhere, and yes, some of them are actually excited about AI.
You're wrong. There is demand. More and more people are exploring AI and getting real work done with it.
Like using it finally be able to do warcrimes without any human pulling the trigger, along the lines of https://en.wikipedia.org/wiki/AI-assisted_targeting_in_the_G...
Perhaps, then, Hamas will finally disarm?
It's like how some people like listening to Ed Sheeran. So yes, there is demand. But nobody is "getting real work done" with these toy AI models.
Real AI is a geo-political threat, and will not be allowed to exist for the average person. So, enjoy your toy AI models, because that's all you're getting.
I'm sorry friend, but you're sounding out of touch here. People are getting real work done all day every day.
>The cynic in me thinks this will go down in history alongside asbestos, leaded gasoline/paint, and the opioid crisis.
Can you elaborate? Leaded gasoline is estimated to have contributed to the deaths of like tens (hundreds?) of millions of people. Asbestos probably millions.
Why would high RAM prices be remembered alongside these events?
Honestly that seems slightly down even if it’s still ridiculous. The ram I bought for $100 a year ago was $500 a couple of months ago. Could just be the particular sticks I got though
$375 is the cheapest kit, and the price is using a promo code
"Price tracking courtesy of PCPartPicker now reveals the cheapest 32GB DDR5 RAM you can buy is $375. Specifically, four XPOWER kits from Silicon Power will set you back $374.97 thanks to a promo code."
that’s what I get for not clicking through TFA
i have 128gb ddr4 from a few years ago. i think i paid like 300-400 for it.
its paired to a 5950x so im sure it will be fine for a few more years
One would think the CPU prices would drop as RAM crunches PC sales, but apparently not
So I thought I would not be paying for that LLM BS just because you don't haven't been suckered into a subscription... I guess I was just lacking sufficient _intelligence_ to realize this kind of thing would happen.
We're all grownups, why should we care about RAM prices? /s
This is insane. We've built our apps and websites to require ungodly amounts of memory and now AI scrapes away said websites while pricing us out.
Fast apps and websites need to make a comeback.
As an ancient person, and considering my sense of inflation of other things I actually buy, This doesnt sound that high to me. At some point in my life CPU Ghz stopped rising and cores was the only thing to grow, and that seemed to be more slowly than clock rate did. So I guess I fully expected that asymptote to apply to RAM too. If you plot price per GB of ram across last 25 years the memory prices growth is barely a bump in the chart. In (nearly) no other physical product do we expect prices to plummet over time whilst simultaneously getting way better. Maybe we're just spoiled by the first 50 - 75 years of wild innovation that most new things go through before they asymptote ?
Yeah, I'm pretty ancient myself. If we look back ~30 years, we're still getting a good value at even these inflated prices for electronics. Food, housing, transportation has skyrocketed, but TVs and computers are comparatively cheap, even with RAM and storage being up 100+%.
I am also that many years old, and I don’t think your analogy holds for this particular situation.
The exact same sticks/model #’s of RAM cost $200 last year, now costs $950. I don’t remember existing CPU’s ever going up in price by 5x (300-400% increases).