On supply side 3 years is about right, new plants won't come online faster. Demand might collapse faster if some AI companies go bankrupt or at very least fail next funding round.
Depends on who goes bankrupt and what happens to their IP when it happens. If OpenAI or Anthropic liquidate, and the IP gets scooped up by MS, Amazon or Google, demand will remain, the public clouds will still want to run them. Maybe some pressure will come off if they lose the appetite to train new models for a while, inference is cheaper, but they'll still finish some of the buildout.
Barring unusual market forces like Taiwan invasion the timeline to ending the acute shortage seems to be mid 2028. The AI still has plenty of money to burn and is the biggest driver, but we’re also shortly before gaming consoles ought to release a new gen (although who knows whether they won’t get delayed for a while). There was even going to be a small upgrade cycle for nerds waiting for 2nm fabbed devices, same as pre-ai datacenters looking for power efficiency. Plenty of pent-up demand, too, as many people simply make do with what they have but will upgrade once the silliness stops.
If you’re looking for ssd/ram prices to go back to the low of 2024/early 2025 it probably won’t happen before China catches up, which will be a while yet. There is some build up of new capcity happening from current manufacturers but it’s significantly less than what the demand increased by.
That was before Trump blew up half the US' munitions inventory in Iran. If Xi is serious about taking back Taiwan, and I don't see why we should doubt his sincerity, the temptation to invade must be irresistible. The only thing that is holding him back is that he purged nearly the entire top brass of the PLA for gross corruption (like "fueling" its ICBMs with water), and he has no idea if his Potemkin army is actually capable of anything.
Eventually supply and demand will get back in a better balance and we will probably see prices rise slower than inflation until adjusted for inflation prices are close to to where they were before but the actual dollar price isn’t likely to go down.
With new fabs built and AI demand shrinking, they will have to. If they don't, considering the last lost price fixing case, they will be absolutely crushed by the EU and probably other governments as well.
Look at the AWS Prime ssds available, and it's a massive list of strange drives you've never heard of, with very few reviews available, almost all using YMTC. The prices aren't fantastic, but given that five sixths of the drive market is straight up gone, I think we need to start reviewing and hoping for the best here, fast. I really hope endurance is indeed as rated, that we aren't about to all get burned incredibly badly for using YMTC chips.
CXMT is indeed starting to get some ram out there. But I am pretty skeptical it's going to make much of a dent. They're currently single digits % of the world ram production. They need to scale a lot to make any dent, and as soon as they do, it feels like there's plenty of people ready to snatch up those supplies.
We need massive massive massive growth in availability and there's no sign that current scale up plans will be at all adequate.
right, but that seems to be the only viable path for any reduction in prices unless the bubble suddenly pops which these ultra qualified people (sam, dario, elon, oracle and so many more) won't let happen.
They won't have choice the bad blood they are creating now won't buy them customers in the future there will be western companies and one country that will design around this AI fiasco.
He's being facetious referring to Amazon Prime. Though AWS and Amazon Ads together account for the entire enterprise value of Amazon, i.e. Wall Street considers Amazon e-commerce to be worthless other than as a loss-leader source of inventory for ads.
The problem with these drives is that you can't ensure the grade of the NAND chips. They could be A+ grade with great endurance but they could also be bottom of the barrel.
I had two KingSpec SSDs sourced from AliExpress that failed too soon: one used YMTC and the other used Intel chips. Both failed within 1 to 2 years.
Initially no but the Chinese will play long and take over much of the existing market worldwide and some companies that are capable will move design and engineering of memory in house and fab with a third party. With in house SOC's now out there memory will be next out of necessity/survival.
"By the end of 2026, we expect CXMT to reach roughly 350 kwspm, which is only modestly below Micron’s estimated ~385 kwspm. This would position CXMT close to becoming the industry’s third-largest memory supplier"
Apple is trying to get a waiver to use CXMT RAM, incidentally. The Chinese government has ordered them to prioritize deliveries to Chinese manufacturers, however.
The new Xbox CEO said recently they are expecting storage prices to be 5x what they were late 2025 by late 2027. And that RAM should be similar.
Anyone making hardware is having a rough time. Like Valve who had to release their new PC at around 40% more expensive than what they originally wanted.
I think they removed the "cheaper" configurations. In essence, the barrier to entry to mac mini was increased without actually changing the original price tag. I suspect the new mac mini (if one is coming) will sport a higher price tag.
It is. They previously got rid of the 256 GB, $599 configuration, and the cheapest option was the 512 GB, $799 config. Now they brought back the 256 GB base model but at $799, and the 512 GB model is $999.
That’s terrible. I purchased my M4 Mac Mini (base 16/256 model) two months ago because I wanted an ARM Mac for a software project. I feared that the M5 Mac Mini would have a price bump, but I would’ve never guessed that Apple would dramatically hike prices for existing models.
I have some choice words for Sam Altman for destroying the personal computing marketplace by cornering the memory market…
The price increases are unsurprising considering Tim Cook said it was "unsustainable" for Apple to keep absorbing the increases. Glad I ordered a new machine a couple days ago.
I suspect that these price increases will stick around permanently (or at least for a long while).
No kidding, I was considering one to replace my 8g air m1. Which was questionable to begin with performance wise, but it's so worn after all these years. Certainly won't do it now.
Yes and no. Relatively speaking, MacBook Neo is still quite cheap, especially since iPad and MacBook Air received even greater price increases. And Apple's competitors are surely experiencing the same component shortages.
Until they raise the price on that too. Dell has explicitly stated it's a "limited time" price, so don't be shocked if it becomes the $699 XPS13 almost immediately.
Just yesterday I saw people saying that Apple wouldn't increase prices until the next refresh.
And I agreed! So… holy shit. I think we're going to see even further price increases across the industry. There already were a ton, but it can always get worse, of course.
Thank you, OpenAI. What would have we done without your attempts at monopolizing destroying the memory market.
The fact that a dozen companies are allowed to buy up the entire global supply of core components, and increase the cost of living for every human on Earth, is full blown dystopian.
> What’s the proposed regulation that would help here? Price controls? They don’t work.
The proposed regulation would be that if a single company/industry buying up supply to the point it starts driving significant inflation for such and such goods, they would be severely restricted from doing so going forward.
It’s a global phenomenon. The latency concerns for data centers are minimal, so they could be built anywhere.
If your country restricted a company from buying too much of a product they need, 10 other competitor companies in other countries would be formed the very next day offering to do the work in their country for a minimal fee.
This is a global market. Supply and demand isn’t going to be cancelled out by politicians in one country trying to squeeze the market.
If you did restrict companies from buying things they need, you would see all future companies in that space incorporated in other countries.
Yeah, imagine doing that for oil. American and EU companies that “hoard” oil get punished. The net effect would be everyone else gets to buy more and prices remain exactly the same.
So, in practice, if, say, the agriculture industry buys up the supply of seeds (they already effectively do) and we see it start driving significant inflation for food (a common concern), the agriculture industry would be restricted from buying seeds?
Yes, because we can't apply specific regulation for specific industries where it makes sense, we have to write them as if we were LLMs so they can be proven to 'not work'.
Good news, you get the DDR2 that has been languishing in a salt cave for the last 20 years.
Reality check: a strategic reserve of modern technology components in volumes needed to impact consumer prices is completely infeasible and illogical.
I’d be fine with the idea of the government maintaining supplies of defense industrial inputs, critical minerals, etc; but as we see with our efforts for rare earths (and even petroleum) you can never stockpile consumer supply levels.
A strategic reserve of a commodity that (historically) depreciates at ~50% per year is a terrible trade for occasionally avoiding demand-driven price spikes.
Because extreme corporate use, that is, what is happening now where a majority of supply is locked up ahead of time via B2B back-room deals, is anti-consumer. Unregulated, it is easy to see how this could lead to a perpetual "rent everything" dystopian environment for consumers.
Every use of DRAM is a corporate use, with the best consumer-friendly examples like Apple’s efforts to hold down prices (until today) being thanks to “back room deals”. Nobody’s buying some DRAM to build a memory stick in their garage.
Apple, Raspberry Pi, Supermicro, and OpenAI all have the same claim to supply you do: you can buy it with money, with the seller being allowed to charge what they want. In fact, high prices are going to be the only way to stimulate supply and encourage the billion dollar investment in additional memory fabs. Price controls or other supply-killing mechanisms are known not to work - it’s Econ 101.
You ignored the part where I mentioned "extreme" and "locked up." To be fair I wasn't necessarily clear what those meant. I'm specifically referring to the deal(s) that OpenAI signed which reserved an outsized chunk of the memory supply, for what is apparently speculative future hardware that hasn't been built yet, or at least to build hardware that no consumer or business will ever be able to physically purchase.
Hopefully you'll agree that there's a difference between even a large buyer like Apple reserving a large chunk of DRAM supply to put in their products that they sell to consumers, and the anti-competitive behavior by OpenAI that I describe above.
Barring any single company from negotiating to buy more than a certain percentage of a given existing market of goods would be a start. Companies would still be free to build their own factories/fabs if they didn't like it.
That, and putting Sam Altman in jail for being a lying fraudster.
One or two companies will come out of this, designing and engineering memory and partnering with someone else to do the fab of that memory no different than making processor chips in Arizona.
The market will take care of itself. The Chinese are going use this to ramp up and build more memory, and some companies out there will take it in-house, In short, they won’t be caught with their pants down again.
> That's why some regulation is not the enemy of the people that some want to make it out to be.
The question is always: What specific regulation?
Regulation is not the magic silver bullet that some want to make it out to be.
You’re not going to solve a global supply and demand change by regulating companies to not buy too many things. The supply would go to other countries. Companies would open international subsidiaries that built the data centers in other countries. Companies would move to other countries which didn’t try to stop them from buying components on the free market.
You can’t regulate companies into keeping prices down. This is an international market. If you passed a law that said RAM had to be sold for no more than 30% higher than last year’s price, the international memory companies would laugh and stop sending RAM to that country.
> Unfortunately, I think regulatory capture is so deep now in most places, one can hardly expect anyone to do anything about it.
I think you need to broaden your understanding of how the DRAM supply chain works and which countries are involved. You can’t mandate low prices for a global commodity. You can try, but the supply will just disappear for that country.
Yes, it's better to not do anything right? After all 'the market' is working for some.
No regulation would catch 100% of this, nor is it meant to. But it can definitely deal with companies opening international subsidiaries etc. Sanctions can be worked around too, but that's a hassle and so countries/companies/individuals generally try to avoid them at all costs.
> But it can definitely deal with companies opening international subsidiaries etc.
You’re still imagining this as a purely single-country issue.
The demand for AI data centers is global. If OpenAI, Anthropic, and xAI weren’t building them, other companies would step in to provide data center services for a fee. Now you have the same buildout, just less efficient and more expensive for the end consumers because we’re paying a new middleman for the compute.
The regulation maximalists would argue that we could then forbid companies from buying foreign data center capacity, but then that means other companies would appear in those other countries offering the AI inference service.
What you’re missing is that this is a global supply and demand issue and you can’t solve it with domestic regulations.
> but then that means other companies would appear in those other countries offering the AI inference service.
That might actually be the goal. A more fragmented market would mean each participant has less money, so they would try to watch their costs a bit more closely. The innovation rate (in non-cost-cutting areas) would probably decrease, maybe even substantially... which some people happen to consistently advocate for. A lot of lost efficiency would be reclaimed in a few years, but the whole system would be more stable, cheaper, and less centralized as a side effect.
Yeah, it would be suicidal to do that when it's your budget that gets the taxes from those giant corporations; who would want to willingly reduce their income for years? The rest of the world would benefit tremendously, but it could be a net plus (socially, politically, if not purely economically) in 5-7 years down the road - even in the country currently benefiting from the corporations the most. But that would be one to two lost elections too late, even if it turned true. So, while it won't happen, if it did, I don't believe we'd be worse for it.
Not saying there isn’t demand, but it’s definitely artificially inflated by VC-fomo and circular-funding ~~fraud~~ shenanigans.
> If OpenAI, Anthropic, and xAI weren’t building them
One of these companies is responsible for buying up DRAM wafers, in what still appears to be an attempt to deny them to everyone else, and another one of these companies seemingly exists to launder money for a fascist billionaire.
There's solutions to everything you mention and as I said, usually when sanctions are applied to countries, companies and individuals are meant to deal exactly with this.
This could range from quanta mandates on the supply side (the RAM manufacturers themselves in this case) to imposing secondary sanctions on 'other companies [that] would step in to provide data center services for a fee'
If the US and the EU did this, these other companies would be mega careful about to whom and how they provide services to, the same way Chinese private companies today are generally super careful about not violating US sanctions.
> If the US and the EU did this, these other companies would be mega careful about to whom and how they provide services to,
There is currently more demand than supply in the entire world.
If the US and EU got together and told DRAM companies that we're going to sanction them if they don't give us cheap RAM, 10 other countries would roll out the red carpet to come bring that DRAM into their countries instead. The data centers would be built there. Then the US and EU would be compute-starved and have no choice but to go to these other countries for compute.
I suggest you read up on the history of attempts to control prices of oil throughout history. Oil is an order of magnitude bigger market than DRAM. If you think it's realistic to suggest that the EU and US could sanction entire countries into keeping some chip prices down so people can save a couple hundred dollars on their next laptop, this isn't a conversation grounded in reality.
> 10 other countries would roll out the red carpet to come bring that DRAM into their countries instead
These 10 countries need the US/EU market for their exports.
But you keep talking as if I am saying I want to sanction those who build more DRAM. No, I want more DRAM, not less!
> we're going to sanction them if they don't give us cheap RAM
That's not what the proposal was. The proposal was to limit the ability of AI goons to completely buy the DRAM market out so that everyone else is forced to pay substantially more.
If the problem is that it feeds into general inflation then it is suddenly not merely 'so people can save a couple hundred dollars on their next laptop'.
It's like oil, it feeds into everything; manufacturing, delivery of goods to your local supermarket, flights etc. etc. you can't simply say 'hey I don't drive a car so high oil prices don't affect me'.
If enterprises and consumers alike are forced to spend substantially more on DRAM, they won't be able to spend on other things and the whole economy will slow down.
> But you keep talking as if I am saying I want to sanction those who build more DRAM. No, I want more DRAM, not less!
The DRAM companies would be building more if they could.
You can't sanction your way into squeezing blood from a stone.
> If enterprises and consumers alike are forced to spend substantially more on DRAM, they won't be able to spend on other things and the whole economy will slow down.
If a country came along and declared that companies couldn't buy the resources they need from other companies, the second order effect would be every major company relocating their headquarters out of that country as soon as possible, along with a sharp decrease in startups being formed in that country.
The economic impacts of this level of command-and-control government would be devastating to the economy. Much more than having to spend a few hundred dollars more on a laptop every 5-10 years.
> The DRAM companies would be building more if they could.
You keep arguing as if there's only one side to this, the producers/DRAM companies who can't scale production fast enough.
But there's two sides to a market, the producers (DRAM makers) and the consumers, (AI industry). I am arguing for increasing the supply by taking some away from the AI industry. This is BECAUSE on the production side there's no way to address this fast enough.
The DRAM producers have also agreed to work together to raise the prices in the past and are probably rather enjoying it being their turn to get a ton of money again. d Free markets break down when cartels form, because then you wind up with an effective monopoly despite having multiple suppliers
It's not in anybody's best interest to take away supply for the AI industry. For better or worse (and whether you believe it or not), AI technologies are coming that will be transformational. If the United States decides to handicap their AI industry, China will simply say "thank you very much" and develop these technologies first. Because of the nature of recursive self-improvement, the country that develops powerful AI first will most likely have an economic lead for quite a while.
It sucks that DRAM is so expensive, but it is for a good (economically useful) cause.
Then, let's see how quickly I can reinterpret whatever power you've grabbed in the name of "doing something" and pervert it for some other nefarious purpose, or just generally bypass the intent entirely as a motivated actor with limitless funds.
Many regulations, once passed, impact only those incapable of navigating around them - typically, the less-fortunate. Invariably, power taken in the name of some transient issue is never later relinquished.
> > That's why some regulation is not the enemy of the people that some want to make it out to be.
> The question is always: What specific regulation?
> Regulation is not the magic silver bullet that some want to make it out to be.
The fact that you ask the important question and then continue to kneejerk at the mention of "regulations" shows the REAL problem. People have problems DISCUSSING the idea. Everyone in the world knows that regulations can be stupid, but that's not the sole property of government, businesses can be colossally stupid too.
My comment was discussing the idea. If you have ideas to discuss, let’s discuss those too.
What I have a problem with is the demand that we accept that regulation will fix everything, but every discussion about the actual effects of regulation gets dismissed.
When an idea only looks good if you can prevent people from discussing the details, it’s probably not a good idea.
Businesses don't generally have the ability to take freedoms, power, etc. and then never relinquish control - their stupidity (in theory) has limited impact on everyone else.
> Businesses don't generally have the ability to take freedoms, power, etc. and then never relinquish control - their stupidity (in theory) has limited impact on everyone else.
They have those powers when they're allowed to become monopolies. Unregulated capitalism leads to monopolies and slavery as they're the best way to capture all capital. Company towns, company stores, company scrip, etc.
> The question is always: What specific regulation?
You're absolutely right that we can't solve this by regulating DRAM prices.
How we got to a situation where a handful of companies can spike the price of consumer electronics several times what it was only a few years ago and these same companies have become the centralized source for information is a journey decades in the making at this point. Decades of insufficient regulations, insufficient enforcement of existing regulations and the lack of any organized efforts to change it.
Microsoft should have been broken up in 2001. The American government should have taken that threat seriously. Governments around the world should have. The dependence of all levels of governments on one single American company for their desktop operating systems and productivity software as well as the spying opportunities that gave American companies and intelligence entities was a grave threat and regulated better to avoid entrenched foreign monopolies. But they didn't. 25 years later and Microsoft still dominates the home OS market and office environment, they have a sizable portion of the cloud, they recently took a huge chunk of the game industry and now the AI industry with their investment in OpenAI.
Even though there's a direct line between a historical lack of regulation on a monopoly like Microsoft and the rise of OpenAI leading to the spike in ram prices it isn't just about Microsoft. You can paint similar pictures about Google, Oracle, Facebook, or Amazon. But to me it isn't just about these companies and regulations/actions directed specifically them but the broader misregulations that have stifled market health and dysfunction that has allowed these criminal organizations to have so much influence.
There could have been real enforcement with criminal penalties and fines that exceed the profits and costs associated with the high-tech employee antitrust litigation.[0] Not doing so has just allowed wealth to continue to accumulate in the hands of criminal people, who not surprisingly continue to do shitty things in their quest for profit. Why were there no personal consequences to Eric Schmidt[1] for these actions, let alone consequences that would have prevented him from attaining the position of influence that he currently has?
The notion of the right to repair should have superseded the DMCA and laws should have been adopted to punish noteworthy companies that lobbied for it and profited from it. There should be more of a focus on governmental standards mandated open interoperability to prevent walled garden business models. This would have kneecapped wealth accumulation among a few corruption groups and allowed a richer more competitive market to flourish. DMCA and copyright extension, WIPO harmonizing of trade law should all have been swept away.
Where's the fallout from Snowden? Were there any massive institutional reforms there? Any jail time for people in government and industry who were involved? How did the lack of regulations and and lasting reform around that debacle shape American society at large and the tech industry?
Everything that we're experiencing today is the result of decades of choices to not regulate the tech industry in any way that resembles other industries. It is a global collective choice to cede power to private individuals based out of the west coast of the US.
They are reserving future HW productions to meet their hypothetical usage as well. Which is why others (like Apple) can’t reserve it for their future products.
Yet the AI labs are speculating on usage, and spending money from investments without clear revenue path.
That's convenient accounting. The reality is that they can't stop training since they risk losing customers if they do so. So they shouldn't factor it out of profitability analysis.
Sorry, I should have said "profit path", good catch!
They have revenue, but their cost scales with revenue and they're losing more than they are making.
Their costs do not scale linearly with revenue. Inference is expensive, but it's a variable cost. Anthropic's overall costs include massive fixed costs in training, which are the same regardless of usage.
It's easy to falsify the claim with a simple experiment: imagine they had no customer at all, $0 in revenue. Their costs would still be massive. If the claim were true, $0 revenue should mean $0 costs, right?
This is not sustainable forever unless their hypothetical usage is realized, and eventually the bill will come due.
Meanwhile, component makers will surely be spinning up more capacity, some of them in a foolhardy manner, and if the bubble does burst, 3-6 months later we'll be seeing fire sales on components and component makers going bankrupt (or getting bailouts, if considered of national importance)
except if as cost here only the inference cost is considered, and not the capital investment, and maintenance costs (not to mention r&d, marketing, and others).
to put it another way, if they just had the corporate API subs today, would they be profitable?
The cure for high prices is high prices. This increase in demand is encouraging economization. Factories which make components are trying to operate for more hours. Producers who haven’t gotten into RAM may try it out. Large companies like Apple may test alternative suppliers. Consumers who don’t really need an upgrade will wait, allowing others who need it to buy one.
Unfortunately, RAM is more like a taxi than an umbrella.
> Anyone who’s spent any time in New York City knows that when it begins to rain, two things happen immediately: It becomes easier to buy an umbrella and it becomes harder to hail a cab. As soon as the first few drops fall, people appear on the street selling cheap umbrellas, while a lucky few pedestrians occupy all the available cabs.
Speaking of which, what's the timeline of the RAM shortage ending? I have no sense for whether it is going to be (for example) 6 months or 3 years.
Barring unusual market forces like Taiwan invasion the timeline to ending the acute shortage seems to be mid 2028. The AI still has plenty of money to burn and is the biggest driver, but we’re also shortly before gaming consoles ought to release a new gen (although who knows whether they won’t get delayed for a while). There was even going to be a small upgrade cycle for nerds waiting for 2nm fabbed devices, same as pre-ai datacenters looking for power efficiency. Plenty of pent-up demand, too, as many people simply make do with what they have but will upgrade once the silliness stops.
If you’re looking for ssd/ram prices to go back to the low of 2024/early 2025 it probably won’t happen before China catches up, which will be a while yet. There is some build up of new capcity happening from current manufacturers but it’s significantly less than what the demand increased by.
Are many DRAM fabs in Taiwan? Does TSMC manufacture DRAM for SK Hynix, Samsung, Micron, or CXMT?
And now China knows that both US and EU are weak and cant even deal with Russia or Iran.
And they also have their own semiconductors manufacturing and cut off from what TSMC produces anyway.
Eventually supply and demand will get back in a better balance and we will probably see prices rise slower than inflation until adjusted for inflation prices are close to to where they were before but the actual dollar price isn’t likely to go down.
Look at the AWS Prime ssds available, and it's a massive list of strange drives you've never heard of, with very few reviews available, almost all using YMTC. The prices aren't fantastic, but given that five sixths of the drive market is straight up gone, I think we need to start reviewing and hoping for the best here, fast. I really hope endurance is indeed as rated, that we aren't about to all get burned incredibly badly for using YMTC chips.
CXMT is indeed starting to get some ram out there. But I am pretty skeptical it's going to make much of a dent. They're currently single digits % of the world ram production. They need to scale a lot to make any dent, and as soon as they do, it feels like there's plenty of people ready to snatch up those supplies.
We need massive massive massive growth in availability and there's no sign that current scale up plans will be at all adequate.
Where?
I had two KingSpec SSDs sourced from AliExpress that failed too soon: one used YMTC and the other used Intel chips. Both failed within 1 to 2 years.
I have another one which is 5 years old by now.
https://newsletter.semianalysis.com/p/chinas-cxmt-is-set-to-...
At current margins their capacity is going to commodity DRAM since their HBM process is not competitive. Will be interesting to see how it pans out
Anyone making hardware is having a rough time. Like Valve who had to release their new PC at around 40% more expensive than what they originally wanted.
:(
I have some choice words for Sam Altman for destroying the personal computing marketplace by cornering the memory market…
I suspect that these price increases will stick around permanently (or at least for a long while).
I don’t think the market shopping for $600 laptops is the same market that wants something that runs Linux.
And I agreed! So… holy shit. I think we're going to see even further price increases across the industry. There already were a ton, but it can always get worse, of course.
Thank you, OpenAI. What would have we done without your attempts at monopolizing destroying the memory market.
Unfortunately, I think regulatory capture is so deep now in most places, one can hardly expect anyone to do anything about it.
The proposed regulation would be that if a single company/industry buying up supply to the point it starts driving significant inflation for such and such goods, they would be severely restricted from doing so going forward.
If your country restricted a company from buying too much of a product they need, 10 other competitor companies in other countries would be formed the very next day offering to do the work in their country for a minimal fee.
This is a global market. Supply and demand isn’t going to be cancelled out by politicians in one country trying to squeeze the market.
If you did restrict companies from buying things they need, you would see all future companies in that space incorporated in other countries.
It’s weird to read all of the calls for regulation to fix this when the DRAM and chip production is happening in other countries.
Maybe we need the same now for computer parts, that are now so important for everything in our modern digital society ?
So that feverish investor speculation and shady circular financing deals don't cause sudden 30+% inflation on any technological device.
Reality check: a strategic reserve of modern technology components in volumes needed to impact consumer prices is completely infeasible and illogical.
I’d be fine with the idea of the government maintaining supplies of defense industrial inputs, critical minerals, etc; but as we see with our efforts for rare earths (and even petroleum) you can never stockpile consumer supply levels.
Apple, Raspberry Pi, Supermicro, and OpenAI all have the same claim to supply you do: you can buy it with money, with the seller being allowed to charge what they want. In fact, high prices are going to be the only way to stimulate supply and encourage the billion dollar investment in additional memory fabs. Price controls or other supply-killing mechanisms are known not to work - it’s Econ 101.
Hopefully you'll agree that there's a difference between even a large buyer like Apple reserving a large chunk of DRAM supply to put in their products that they sell to consumers, and the anti-competitive behavior by OpenAI that I describe above.
That, and putting Sam Altman in jail for being a lying fraudster.
The question is always: What specific regulation?
Regulation is not the magic silver bullet that some want to make it out to be.
You’re not going to solve a global supply and demand change by regulating companies to not buy too many things. The supply would go to other countries. Companies would open international subsidiaries that built the data centers in other countries. Companies would move to other countries which didn’t try to stop them from buying components on the free market.
You can’t regulate companies into keeping prices down. This is an international market. If you passed a law that said RAM had to be sold for no more than 30% higher than last year’s price, the international memory companies would laugh and stop sending RAM to that country.
> Unfortunately, I think regulatory capture is so deep now in most places, one can hardly expect anyone to do anything about it.
I think you need to broaden your understanding of how the DRAM supply chain works and which countries are involved. You can’t mandate low prices for a global commodity. You can try, but the supply will just disappear for that country.
No regulation would catch 100% of this, nor is it meant to. But it can definitely deal with companies opening international subsidiaries etc. Sanctions can be worked around too, but that's a hassle and so countries/companies/individuals generally try to avoid them at all costs.
You’re still imagining this as a purely single-country issue.
The demand for AI data centers is global. If OpenAI, Anthropic, and xAI weren’t building them, other companies would step in to provide data center services for a fee. Now you have the same buildout, just less efficient and more expensive for the end consumers because we’re paying a new middleman for the compute.
The regulation maximalists would argue that we could then forbid companies from buying foreign data center capacity, but then that means other companies would appear in those other countries offering the AI inference service.
What you’re missing is that this is a global supply and demand issue and you can’t solve it with domestic regulations.
That might actually be the goal. A more fragmented market would mean each participant has less money, so they would try to watch their costs a bit more closely. The innovation rate (in non-cost-cutting areas) would probably decrease, maybe even substantially... which some people happen to consistently advocate for. A lot of lost efficiency would be reclaimed in a few years, but the whole system would be more stable, cheaper, and less centralized as a side effect.
Yeah, it would be suicidal to do that when it's your budget that gets the taxes from those giant corporations; who would want to willingly reduce their income for years? The rest of the world would benefit tremendously, but it could be a net plus (socially, politically, if not purely economically) in 5-7 years down the road - even in the country currently benefiting from the corporations the most. But that would be one to two lost elections too late, even if it turned true. So, while it won't happen, if it did, I don't believe we'd be worse for it.
Not saying there isn’t demand, but it’s definitely artificially inflated by VC-fomo and circular-funding ~~fraud~~ shenanigans.
> If OpenAI, Anthropic, and xAI weren’t building them
One of these companies is responsible for buying up DRAM wafers, in what still appears to be an attempt to deny them to everyone else, and another one of these companies seemingly exists to launder money for a fascist billionaire.
Fascist trillionaire
This could range from quanta mandates on the supply side (the RAM manufacturers themselves in this case) to imposing secondary sanctions on 'other companies [that] would step in to provide data center services for a fee'
If the US and the EU did this, these other companies would be mega careful about to whom and how they provide services to, the same way Chinese private companies today are generally super careful about not violating US sanctions.
There is currently more demand than supply in the entire world.
If the US and EU got together and told DRAM companies that we're going to sanction them if they don't give us cheap RAM, 10 other countries would roll out the red carpet to come bring that DRAM into their countries instead. The data centers would be built there. Then the US and EU would be compute-starved and have no choice but to go to these other countries for compute.
I suggest you read up on the history of attempts to control prices of oil throughout history. Oil is an order of magnitude bigger market than DRAM. If you think it's realistic to suggest that the EU and US could sanction entire countries into keeping some chip prices down so people can save a couple hundred dollars on their next laptop, this isn't a conversation grounded in reality.
These 10 countries need the US/EU market for their exports.
But you keep talking as if I am saying I want to sanction those who build more DRAM. No, I want more DRAM, not less!
> we're going to sanction them if they don't give us cheap RAM
That's not what the proposal was. The proposal was to limit the ability of AI goons to completely buy the DRAM market out so that everyone else is forced to pay substantially more.
If the problem is that it feeds into general inflation then it is suddenly not merely 'so people can save a couple hundred dollars on their next laptop'.
It's like oil, it feeds into everything; manufacturing, delivery of goods to your local supermarket, flights etc. etc. you can't simply say 'hey I don't drive a car so high oil prices don't affect me'.
If enterprises and consumers alike are forced to spend substantially more on DRAM, they won't be able to spend on other things and the whole economy will slow down.
I'd argue that's incentive enough.
The DRAM companies would be building more if they could.
You can't sanction your way into squeezing blood from a stone.
> If enterprises and consumers alike are forced to spend substantially more on DRAM, they won't be able to spend on other things and the whole economy will slow down.
If a country came along and declared that companies couldn't buy the resources they need from other companies, the second order effect would be every major company relocating their headquarters out of that country as soon as possible, along with a sharp decrease in startups being formed in that country.
The economic impacts of this level of command-and-control government would be devastating to the economy. Much more than having to spend a few hundred dollars more on a laptop every 5-10 years.
You keep arguing as if there's only one side to this, the producers/DRAM companies who can't scale production fast enough.
But there's two sides to a market, the producers (DRAM makers) and the consumers, (AI industry). I am arguing for increasing the supply by taking some away from the AI industry. This is BECAUSE on the production side there's no way to address this fast enough.
https://en.wikipedia.org/wiki/DRAM_price_fixing_scandal
https://ec.europa.eu/commission/presscorner/detail/en/ip_10_...
It sucks that DRAM is so expensive, but it is for a good (economically useful) cause.
Ah yes, "We have to do something! Something must be better than nothing!"
Famous last words before freedoms of all varieties are eroded.
Then, let's see how quickly I can reinterpret whatever power you've grabbed in the name of "doing something" and pervert it for some other nefarious purpose, or just generally bypass the intent entirely as a motivated actor with limitless funds.
Many regulations, once passed, impact only those incapable of navigating around them - typically, the less-fortunate. Invariably, power taken in the name of some transient issue is never later relinquished.
The fact that you ask the important question and then continue to kneejerk at the mention of "regulations" shows the REAL problem. People have problems DISCUSSING the idea. Everyone in the world knows that regulations can be stupid, but that's not the sole property of government, businesses can be colossally stupid too.
My comment was discussing the idea. If you have ideas to discuss, let’s discuss those too.
What I have a problem with is the demand that we accept that regulation will fix everything, but every discussion about the actual effects of regulation gets dismissed.
When an idea only looks good if you can prevent people from discussing the details, it’s probably not a good idea.
Businesses don't generally have the ability to take freedoms, power, etc. and then never relinquish control - their stupidity (in theory) has limited impact on everyone else.
They have those powers when they're allowed to become monopolies. Unregulated capitalism leads to monopolies and slavery as they're the best way to capture all capital. Company towns, company stores, company scrip, etc.
You're absolutely right that we can't solve this by regulating DRAM prices. How we got to a situation where a handful of companies can spike the price of consumer electronics several times what it was only a few years ago and these same companies have become the centralized source for information is a journey decades in the making at this point. Decades of insufficient regulations, insufficient enforcement of existing regulations and the lack of any organized efforts to change it.
Microsoft should have been broken up in 2001. The American government should have taken that threat seriously. Governments around the world should have. The dependence of all levels of governments on one single American company for their desktop operating systems and productivity software as well as the spying opportunities that gave American companies and intelligence entities was a grave threat and regulated better to avoid entrenched foreign monopolies. But they didn't. 25 years later and Microsoft still dominates the home OS market and office environment, they have a sizable portion of the cloud, they recently took a huge chunk of the game industry and now the AI industry with their investment in OpenAI.
Even though there's a direct line between a historical lack of regulation on a monopoly like Microsoft and the rise of OpenAI leading to the spike in ram prices it isn't just about Microsoft. You can paint similar pictures about Google, Oracle, Facebook, or Amazon. But to me it isn't just about these companies and regulations/actions directed specifically them but the broader misregulations that have stifled market health and dysfunction that has allowed these criminal organizations to have so much influence.
There could have been real enforcement with criminal penalties and fines that exceed the profits and costs associated with the high-tech employee antitrust litigation.[0] Not doing so has just allowed wealth to continue to accumulate in the hands of criminal people, who not surprisingly continue to do shitty things in their quest for profit. Why were there no personal consequences to Eric Schmidt[1] for these actions, let alone consequences that would have prevented him from attaining the position of influence that he currently has?
The notion of the right to repair should have superseded the DMCA and laws should have been adopted to punish noteworthy companies that lobbied for it and profited from it. There should be more of a focus on governmental standards mandated open interoperability to prevent walled garden business models. This would have kneecapped wealth accumulation among a few corruption groups and allowed a richer more competitive market to flourish. DMCA and copyright extension, WIPO harmonizing of trade law should all have been swept away.
Where's the fallout from Snowden? Were there any massive institutional reforms there? Any jail time for people in government and industry who were involved? How did the lack of regulations and and lasting reform around that debacle shape American society at large and the tech industry?
Everything that we're experiencing today is the result of decades of choices to not regulate the tech industry in any way that resembles other industries. It is a global collective choice to cede power to private individuals based out of the west coast of the US.
[0] https://en.wikipedia.org/wiki/High-Tech_Employee_Antitrust_L...
Yet the AI labs are speculating on usage, and spending money from investments without clear revenue path.
See: https://www.wheresyoured.at/brokenomics/ for an interesting write-up on the economics of AI.
It's easy to falsify the claim with a simple experiment: imagine they had no customer at all, $0 in revenue. Their costs would still be massive. If the claim were true, $0 revenue should mean $0 costs, right?
Meanwhile, component makers will surely be spinning up more capacity, some of them in a foolhardy manner, and if the bubble does burst, 3-6 months later we'll be seeing fire sales on components and component makers going bankrupt (or getting bailouts, if considered of national importance)
Doesn't mean people would legitimately use them enough to warrant such infrastracture demand, if they were priced according to actual costs.
So it's a distorted market.
It is more expensive? yeah
Is it "definitely not below cost"? hardly
except if as cost here only the inference cost is considered, and not the capital investment, and maintenance costs (not to mention r&d, marketing, and others).
to put it another way, if they just had the corporate API subs today, would they be profitable?
> Anyone who’s spent any time in New York City knows that when it begins to rain, two things happen immediately: It becomes easier to buy an umbrella and it becomes harder to hail a cab. As soon as the first few drops fall, people appear on the street selling cheap umbrellas, while a lucky few pedestrians occupy all the available cabs.
http://shirky.com/2001/01/