Potentially unpopular take: memory manufacturers have been operating on the margins of profitability for quite a while now. Their products are essentially an indistinguishable commodity. Memory from Samsung or Micron or another manufacturer may have slight differences in overclockability, but that matters little to folks who just want a stable system. Hopefully the shortage leads large purchasers to engage in long-term contracts with the memory manufacturers which give them the confidence to invest in new fabs and increased capacity. That would be great for everyone. Additionally, we're likely to see Chinese fab'd DRAM now, which they've been attempting since the '70s but never been competitive at. With these margins, any new manufacturer could gain a foothold.
If LLMs' utility continues to scale with size (which seems likely as we begin training embodied AI on a massive influx of robotic sensor data) then it will continue to gobble up memory for the near future. We may need both increased production capacity _and_ a period of more efficient software development techniques as was the case when a new 512kb upgrade cost $1,000.
> Hopefully the shortage leads large purchasers to engage in long-term contracts with the memory manufacturers which give them the confidence to invest in new fabs and increased capacity.
Most DRAM is already purchased through contracts with manufacturers.
Manufacturers don't actually want too many extremely long term contracts because it would limit their ability to respond to market price changes.
Like most commodities, the price you see on places like Newegg follows the "spot price", meaning the price to purchase DRAM for shipment immediately. The big players don't buy their RAM through these channels, they arrange contracts with manufacturers.
The contracts with manufacturers will see higher prices in the future, but they're playing the long game and will try to delay or smooth out purchasing to minimize exposure to this spike.
> Additionally, we're likely to see Chinese fab'd DRAM now, which they've been attempting since the '70s but never been competitive at.
Companies like Samsung and SK Hynix have DRAM fabs in China already. This has been true for decades. You may have Chinese fab'd DRAM in the computer you're using right now.
Are you referring to complete home-grown DRAM designs? That, too, was already in the works.
> Manufacturers don't actually want too many extremely long term contracts because it would limit their ability to respond to market price changes.
I don't agree with this sentence. Why would not the same apply advice to oil and gas contracts? If you look at the size and duration of oil and gas contracts for major energy importers, they often run 10 years or more. Some of the contracts in Japan and Korea are so large, that a heavy industrial / chemical customers will take an equity stake in the extraction site.
Except silicon, power, and water (and a tiny amount of plastic/paper for packaging), what else does a fab need that only produces DRAM? If true, then power is far and away the most variable input cost.
> Why would not the same apply advice to oil and gas contracts?
Because oil & gas suppliers only ever sell one product, and memory fabs can dynamically switch product mix in response to supply & demand to optimize profits. The same sand, power and water can make DDR4, HBM or DDR5
Neither do chips, even if they all start as silicon from the ground. What the earlier comment was saying is that the actual composition of crude oil varies by location so you aren't necessarily getting the same ratio of finished products at the process. With silicon you have a bit more control over what goes into the fab. But you're still at the mercy of demand from the market.
The crude composition defines a range of possible products, not exactly ratios. Longer chain hydrocarbons are also cracked to yield more light products.
> defines a range of possible products, not exactly ratios
I'm not sure I follow, varying range necessarily implies varying ratios (e.g. a product missing from the range means its ratio is zero).
Even when in theory you can obtain some higher quality products, the composition of the crude can make it too complex and expensive to practically obtain them.
You don't want to refine gasoline from heavy crude, especially in winter when demand is lower. For gasoline or kerosene you want to start from lighter crude. Same with many undesired components (either from the crude or resulting from the refining methods), the more you have, the more complex the refining, and the resulting ratio of products you obtain varies.
So in practice what you get out of the refining process absolutely depends on the characteristics of the crude, and many other things like market demand or the capability of your refinery.
Same as with silicon. The process to make the wafer results in different quality if you want to make low tech or cutting edge semiconductor products.
That way? I was trying to say that the mix of hydrocarbon molecules is different for each and every oil field due to local geological variation. Even within the field, since eg lighter molecules presumably come out first.
Are you seriously trying to compare raw commodity inputs traded on the futures market to finished semiconductor products that are expected to become deprecated, uncompetitive and/or EOL'd in a few years?
Yes it looks like he does. And I don't see why not.
The fact that their products become deprecated, gives even more incentive to manufacturers to want long term contracts.
>So many a time LNG companies break contracts and pay hefty penalties if the spot rate is high enough.
What do you mean "Break contracts"? I thought the conversation was about Futures contracts, you don't break them. You sell your contract or you take/give delivery (or cash settle).
There's no specific mention of futures upthread of this comment.
Not all gas is sold by futures, you can have a contract for, say, delivery of 20 million cubic metres of gas a year and a penalty if that isn't met. Some people actually want the gas for gas-related purposes rather then as a financial phantom.
Same for DRAM - Dell actually wants the chips to put in computers, an economic abstraction doesn't help much when you need to ship real computers to get paid, and many customers aren't in the market for a laptop future (Framework pre-orders notwithstanding).
As I understand hydrocarbon trading (oil and gas), futures is a tiny portion of the settled market. The vast majority is traded through long-term, privately negotiated contracts. As I said previously, many of those contracts are so large that the end buyer takes an equity stake in the extraction site.
> Except silicon, power, and water (and a tiny amount of plastic/paper for packaging), what else does a fab need that only produces DRAM? If true, then power is far and away the most variable input cost.
Borrowing costs can be wildly variable and are the main cost of making silicon. All the "inputs" over the lifecycle of a fab are so completely dwarfed by the initial capital costs that you can pretty much ignore them in any economic analysis. The cost of making chips is the cost of borrowing money to pay for capital costs, and the depreciation of the value of that capital.
This theory sounds nice, but do you have any sources to share? For example, I assume a fab runs for about 20-30 years. The labor inputs must be very high over this period. Basically, there are no poorly paid people inside a fab. And wouldn't "wildly variable borrowing costs" also affect oil and gas who need to finance the research phase and construction of the plant?
> For example, I assume a fab runs for about 20-30 years.
If only.
20 years ago, fabs were being built to use 90nm class technology. Chips made on such an old node are so cheap today it can't pay even fraction of a percent of the capital costs of the plant per year. So all of it's capital has to have been depreciated a long time ago.
The oldest process node in high-volume production for memory is currently 1α, which started production in January 2021. It is no longer capable of making high-end products and is definitely legacy, and also has to have essentially depreciated all of the capital costs. The time a high-end fab stays high-end and can command premium prices, and during which it has to depreciate all the capital is ~3-5 years. After that either you push the plant to produce legacy/low price and low margin items, or you rebuild it with new tools with costs >$10B.
Also, even if fabs did last 20-30 years, the capital costs would dominate.
> And wouldn't "wildly variable borrowing costs" also affect oil and gas who need to finance the research phase and construction of the plant?
I don't understand? Nothing else costs anywhere near as much capital to produce than silicon chips. Thanks to the inexorable force of Moore's second law, fabs are machines that turn capital investment into salable product, nothing like it has ever existed before.
Micron Fab 6 is about 2300 staff (including 1000 contractors).
Even if you pay them all 500k per year, that's "only" about a billion a year in payroll.
The New York fab plan costs something like 20 billion more or less now to build, with 100 billion over 20 years.
Also, maybe the calculus is different right now in the US, but it used to be the semiconductor workers were expected to have PhDs coming out of their ears but were not actually paid very well, with salaries in Taiwanese fabs being around the $50-60k mark, and lower paid workers being more like $20k or less. Presumably US fabs will be automated to an even greater extent due to labour costs.
So it's very possible that servicing debt on the capital outlay is substantially more expensive than the payroll.
As I mentioned, various groups within China have been working on China-native DRAM since the '70s. What's new are the margins and market demand to allow them to be profitable with DRAM which is still several years behind the competition.
Well, what really prompted this crisis is AI, as well as Samsung shutting down some production (and I have to say I don't think they mind that the pricing has skyrocketed as a result!)
If the shortage of RAM is because of AI (so servers/data centers I presume?), wouldn't that mean the shortage should be localized to RDIMM rather than the much more common UDIMM that most gaming PCs use? But it seems to me like the pricing is going up more for UDIMM than RDIMM.
UDIMM and RDIMM use the same DRAM chips. And my understanding is that the fabs can switch between DDR5, LPDDR5, and maybe HBM as needed. This means high demand for one type can create a shortage of the others.
> This means high demand for one type can create a shortage of the others.
Wouldn't that mean that a shortage of DRAM chips should cause price difference in all of them? Not sure that'd explain why RDIMM prices aren't raising as sharply as UDIMM. That the fab and assembly lines have transitioned into making other stuff makes sense why'd there be a difference though, as bradfa mentioned in their reply.
It's a valid question if you're not familiar with the RAM market. Sorry you're getting downvoted for it.
The manufacturers make the individual chips, not the modules (DIMMs). (EDIT: Some companies that make chips may also have business units that sell DIMMS, to be pedantic.)
The R in RDIMM means register, aka buffer. It's a separate chip that buffers the signals between the memory chips and the controller.
Even ECC modules use regular memory chips, but with extra chips added for the ECC capacity.
It can be confusing. The key thing to remember is that the price is driven by the price of the chips. The companies that make DIMMs are buying chips in bulk and integrating them on to PCBs.
> Even ECC modules use regular memory chips, but with extra chips added for the ECC capacity.
Quite a few unbuffered designs in the past had a "missing chip". If you ever wondered why a chip was missing on your stick, it's missing ECC. Don't know if it's still the case with DDR5 though.
I have not seen that yet DDR5, I think the signal integrity requirements are too high now to even have unused pads open. Most sticks don’t appear to have many traces at all on the top/bottom sides, just big power/ground planes.
Also with DDR5 each stick is actually 2 channels so you get 2 extra dies.
Presumably this is being marketed indistinguishably from regular ECC RAM?
If so, that's terrible news. It was already difficult enough to find ECC RAM for "workstation" class machines (i.e.: High end, non-server CPUs that support ECC such as AMD Threadripper).
It's not - ECC RAM still means real, between the RAM and CPU ECC. It's literally an extra 8 bits per channels, for an extra 16 bits per dim. 40 bits vs 32.
Because manufacturers transitioned fab and assembly lines from low margin dram to higher margin products like hbm, hence reducing dram supply. But the demand for consumer grade dram hasn’t changed much so prices for it go up.
The “regular” and “premium” label at the pump is misleading. The premium gas isn’t better. It’s just different. Unless your car specifically requires higher octane fuel, there is no benefit to paying for it. https://www.kbb.com/car-advice/gasoline-guide/
Only if it has the physical ability to use more octane.
In theory, your average Camry running on 87 is pulling spark timing to ride the edge of knock for best fuel efficiency by being lean, but how much? It was designed to be safe on even kinda shitty gas, that has lower than 87 octane at points, and the ECU is going to err on the side of caution.
That naturally aspirated 2AR-FE in a Camry does not have the ability to compress harder, so if you put 93 in it, it may only be able to "utilize" the extra knock resistance up to say 89 by advancing spark timing.
Meanwhile your average Golf TSI probably can, and the VW GTI I have demonstrably gets better gas mileage on 93 octane, even though it is "rated" for 87 octane (and therefore has a lower mpg claim than it is capable of), but this was an engine that previously was rated at 91 octane and nerfs itself so hard on 87 that it is dramatically easier to stall, and the power figures are rated on 91 octane anyway.
You are almost certainly spending more money on gas even if you eke out a percentage point or two extra mpg on higher octane fuels, as they are priced at higher margins and have lower scale.
It's a sad trend for "the rest of us" and history in general. The economic boom of the 80's thru the 2010s has been a vast democratization of computation - hardware became more powerful and affordable, and algorithms (at least broadly if not individually) became more efficient. We all had supercomputers in our pockets. This AI movement seems to move things in the opposite direction, in that us plebeians have less and less access to RAM, computing power and food and...uh...GPUs to play Cyberpunk; and are dependent on Altermanic aristocracy to dribble compute onto us at their leisure and for a hefty tithe.
I am hoping some of that Clayton Christensen disruption the tech theocracy keep preaching about comes along with some O(N) decrease in transformer/cDNN complexity that disrupts the massive server farms required for this AI boom/bubble thing.
> This AI movement seems to move things in the opposite direction, in that us plebeians have less and less access to RAM, computing power and food and...uh...GPUs to play Cyberpunk; and are dependent on Altermanic aristocracy to dribble compute onto us at their leisure and for a hefty tithe.
Compute is cheaper than ever. The ceiling is just higher for what you can buy.
Yes, we have $2000 GPUs now. You don't have to buy it. You probably shouldn't buy it. Most people would be more than fine with the $200-400 models, honestly. Yet the fact that you could buy a $2000 GPU makes some people irrationally angry.
This is like the guy I know who complains that pickup trucks are unfairly priced because a Ford F-150 has an MSRP of $80,000. It doesn't matter how many times you point out that the $80K price tag only applies to the luxury flagship model, he anchors his idea of how much a pickup truck costs to the highest number he can see.
Computing is cheaper than ever. The power level is increasing rapidly, too. The massive AI investments and datacenter advancements are pulling hardware development forward at an incredible rate and we're winning across the board as consumers. You don't have to buy that top of the line GPU nor do you have to max out the RAM on your computer.
Some times I think people with this mentality would be happier if the top of the line GPU models were never released. If nVidia stopped at their mid-range cards and didn't offer anything more, the complaints would go away even though we're not actually better off with fewer options.
> Yes, we have $2000 GPUs now. You don't have to buy it. You probably shouldn't buy it. Most people would be more than fine with the $200-400 models, honestly. Yet the fact that you could buy a $2000 GPU makes some people irrationally angry.
This is missing the forest for the trees quite badly. The 2000 price GPUs are what would've been previously 600-700, and the 200-400 dollar GPUs are now 600-700. Consumers got a shit end of the deal when crypto caused GPUs to spike and now consumers are getting another shitty deal with RAM prices. And even if you want mid range stuff it's harder and harder to buy because of how fucked the market is.
It would be like if in your example companies literally only sold F-150s and stopped selling budget models at all. There isn't even budget stock to buy.
> Some times I think people with this mentality would be happier if the top of the line GPU models were never released. If nVidia stopped at their mid-range cards and didn't offer anything more, the complaints would go away even though we're not actually better off with fewer options.
If the result was that games were made and optimised for mid-range cards, maybe regular folks actually would be better off.
Well...Cyberpunk for example. It will run on a 3070 but look like shit...on my 4k monitor, if I want the gorgeous photorealism of the game, the $2000 (formerly $700 pre-crypto/AI) GPU still gets about ~40 FPS with some stuttering if you don't turn on the fake AI generated frames...
> xx60 is mid range at best. Maybe even xx50 if those still exist
"It's mid range if it exists" doesn't make sense.
Also you're missing that they're talking about 3070, a card from 2020 (5 years ago), 2 generations behind this year's 50xx series. The 30xx matters more than the xx70 here. It was an upper midrange card when it came out, and it's solidly midrange for Nvidia's product lineup today. You can have cheaper and decent just fine (integrated Ryzens like you mentioned are fine for 1080p gaming on most titles).
A GTX 1080 came out in the first half of 2016. It had 8 GB of VRAM and cost $599 with a TDP of 180W.
A GTX 1080 Ti came out in 2017 and had 11 GB of VRAM at $799.
In 2025 you can get the RTX 5070 with 12 GB of VRAM. They say the price is $549, but good luck finding them at that price.
And the thing with VRAM is that if you run out of it then performance drops off a cliff. Nothing can make up for it without getting a higher VRAM model.
> "They say the price is $549, but good luck finding them at that price."
I did one Google search for "rtx 5070 newegg usa" and they have MSI Ventus GeForce RTX 5070 12G down from $559 to $499 for Black Friday, and ASUS Prime RTX 5070 12GB for $543.
-the whole reason why the GPU is $2000 is because of said AI bubble sucking up wafers at TSMC or elsewhere, with a soupçon of Jensen's perceived monopoly status...
-for a good part of the year, you could not actually buy said $2000 GPU (I assume you are referring to the 5090) also because of said AI bubble
(granted, while Jensen does not want to sell me his GPU, I would like to point out that Tim Cook has no problem taking my money).
on that point, I can go and buy a Ford F150 tomorrow. Apparently, per the article, I would have problems buying bog standard DDR5 DIMMS to build my computer.
One can see it that way, granted. When I zoom all the way out, all of consumer computation has existed as sort of an addendum or ancillary organ to the big customers: government, large corporations, etc. All our beloved consumer tech started out as absurdly high priced niche stuff for them. We've been sold the overflow capacity and binned parts. And that seems to be a more-or-less natural consequence of large purchasers signing large checks and entering predictable contracts. Individual consumers are very price sensitive and fickle by comparison. From that perspective, anything that increases overall capacity should also increase the supply of binned parts and overflow. Which will eventually benefit consumers. Though the intervening market adjustment period may be painful (as we are seeing). Consumers have also benefited greatly from the shrinking of component sizes, as this has had the effect of increasing production capacity with fixed wafer volume.
SGI and 3Dfx made high-end simulators for aerospace in the beginning. Gaming grew out of that. Even Intel's first GPU (the i740) came from GE Aerospace.
Flight simulators just had more cash for more advanced chips, but arcade games like the Sega Model 1 (Virtua Racing) was via Virtua Fighter an inspiration for the Playstation, and before that there was crude games on both PC and Amiga.
Games were always going to go 3d sooner or later, the real pressure of the high volume competitive market got us more and more capable chips until they were capable enough for the kind of computation needed for neural networks faster than a slow moving specialty market could have.
> Flight simulators just had more cash for more advanced chips
Yes. That is my point. The customers willing to pay the high initial R+D costs opened up the potential for wider adoption. This is always the case.
Even the gaming GPUs which have grown in popularity with consumers are derivatives of larger designs intended for research clusters, datacenters, aerospace, and military applications.
No question that chip companies are happy to take consumers money. But I struggle to think of an example of a new technology which was invented and marketed to consumers first.
Computers themselves were non-consumer to begin with, but the Personal Computer broke the technology moat to consumers before anything else and once that had passed it was mostly a matter of time imho.
Many 3d games like doom, quake1, flight unlimited,etc ran purely on software rendering since CPU's were already providing enough oomph to render fairly useful 3d graphics in the mid 90s. CPU power was enough but consoles/arcades showed that there was more to be gotten (but nothing hindered games at that point).
And already there, the capital investment for game consoles (Atari,NES,SNES,PS1,PS2, etc) and arcade games(like the above mentioned 3d games) were big enough to use custom chipsets not used or purposed for anything else (I think also that in the 80s/90s the barrier of entry to making competitive custom chips was a tad lower, just consider the cambrian explosions of firms during the 90s making x86 and later ARM chips).
Yes, there was vendors that focused on the high end commercial customers, and yes many alumnis of those firms did contribute a ton of expertise towards what we have today.
But if you look at what companies survived and pushed the envelope in the longer run it was almost always companies that competed in the consumer market, and it was only when those consumer chips needed even more advanced processing that we breached the point where the chips became capable of NN's.
In fact I'd say that had the likes of SGI prevailed we would've had to wait longer for our GPU revolution. Flight simulators,etc were often focused on "larger/detailed" worlds, PS2-era chips with higher polycounts and more memory would have been fine for simulator developers for a long time (since more details in a military scenario would have been fine).
Leisure games has always craved fidelity on a more "human" level, to implement "hacks" for stuff with custom dynamic lighting models, then global illumination, subsurface scattering,etc we've needed the arbitrary programmability since the raw power wasn't there (the most modern raytracing chips are _starting_ to approach that levels without too ugly hacks).
Wolfenstein 3d was released before 3DFx existed, was purely CPU rendered, and generally considered the father of modern 3d shooters. Even without the scientific computing angle, GPUs would have been developed for gaming simply because it was a good idea that clearly had a big market.
> When I zoom all the way out, all of consumer computation has existed as sort of an addendum or ancillary organ to the big customers: government, large corporations, etc.
Perfectly stated. I think comments like the one above come from a mentality that the individual consumer should be the center of the computing universe and big purchasers should be forced to live with the leftovers.
What's really happening is the big companies are doing R&D at incredible rates and we're getting huge benefits by drafting along as consumers. We wouldn't have incredible GPUs in our gaming systems and even cell phones if the primary market for these things was retail entertainment purchases that people make every 5 years.
The iPhone wasn't designed or marketed to large corporations. 3dfx didn't invent the voodoo for B2B sales. IBM didn't branch out from international business machines to the personal computer for business sales. The compact disc wasn't invented for corporate storage.
Computing didn't take off until it shrank from the giant, unreliable beasts of machines owned by a small number of big corporations to the home computers of the 70s.
There's a lot more of us than them.
There's a gold rush market for GPUs and DRAM. It won't last forever, but while it does high volume sales at high margins will dominate supply. GPUs are still inflated from the crypto rush, too.
3Dfx was not the inventor of the GPU. There’s a long history of GPU development for corporate applications.
The iPhone wasn’t the first mobile phone. Early mobile phones were very expensive and targeted as businesses who wanted their executives in touch
You’re still thinking from a consumer-centric view. Zoom out and those consumer companies were not the first to develop the products. You didn't even think about the actual originators of those types of products because you don’t see them as a consumer.
The consumer centric view is powerful - as was stated above, the originators often were niche/expensive items. The power of the consumer market was what really drove the engineering/science forward and made those companies into market powerhouses. It also arguably killed intel and build TSMC into the most technically advanced company in the world.
I'm sorry, but I'm not sure if you're implying you dislike Apple's approach to what the user is allowed to do, or suggesting we should only talk about general purpose computing devices. If it's the latter, sure, the iPhone's not an innovation in that space, discard it from my list of examples. If it's the former, I'll give you that too, but it was still the first of its kind, by a large margin.
(I remember the huge window in which phone companies desperately put out feature phones with sub-par touch screens, completely missing the value to consumers. The iPod Touch should've been warning enough... and should've been (one of) my signal(s) to buy Apple stock, I guess :-)
Advances in video cards and graphics tech were overwhelmingly driven by video games. John Carmack, for instance, was directly involved in these processes and 'back in the day' it wasn't uncommon for games, particularly from him, to be developed to run on tech that did not yet exist, in collaboration with the hardware guys. Your desktop was outdated after a year and obsolete after 2, so it was a very different time than modern times where you example is not only completely accurate, but really understating it - a good computer from 10 years ago can still do 99.9% of what people need, even things like high end gaming are perfectly viable with well dated cards.
> a good computer from 10 years ago can still do 99.9% of what people need, even things like high end gaming are perfectly viable with well dated cards.
HN is strange. I have an old gaming build from 7-8 years ago and while it can do high end games on low settings and resolution, it doesn’t hold a candle to even a mid-range modern build.
“viable” is doing a lot of work in that claim. You can tolerate it at low res and settings and if you’re okay with a lot of frame rate dips, but nobody is going to mistake it for a modern build.
You’re also exaggerating how fast video cards became obsolete in the past. Many of us gamed just fine on systems that weren’t upgraded for 5-6 years at a time.
I'll take the absurd extreme end of my claim. Here [1] is a video of somebody running modern games on a GeForce GTX 1080 Ti, a card that was high end... 8 years ago. And he's doing it on high-ultra settings in 4k, and it's still doing fine. Spend a couple of hundred on a "new" video card and he'd be rocking a stable 60+FPS on everything, with some games he's still hitting that even with his card!
And back in the early 2000s, even bleeding edge current-year rigs would struggle with new games like Doom 3, Far Cry, Crysis, and so on. Hardware was advancing so rapidly that games were being built in anticipation of upcoming hardware, so you had this scenario where high end systems bought in one year would struggle with games released that year, let alone systems from 5-6 years prior.
Obviously if you're referencing CRPGs and the like, then yeah - absolutely anything could run them. The same remains even more true today. Baldur's Gate 3's minimum requirement is a GTX 970, a card more than 11 years old. Imagine a 1989 computer trying to run Baldur's Gate 2!
I'm still on PCIe 3.0 on my main machine and the RX580 works fine for my needs. Outside of the scope of OP, I recently picked up a (new) 5060 not due to the impending memory production apocalypse but because I wanted to extend my current setup with something I recently read about on LSFG, previously posted here but garnered no interest/comments.
I wonder about this...I had thought I would be on PCIe 5.0 by now but I'm still on my AM4 PCIe 4.0 board since AM5 and PCIe 5.0 seem...glitchy and heat prone. And apparently I'm still not saturating PCIe 4.0...
> We wouldn't have incredible GPUs in our gaming systems and even cell phones if the primary market for these things was retail entertainment purchases that people make every 5 years.
Arguably we don't. Most of the improvements these days seem to be on the GPGPU side with very little gains in raster performance this decade.
> with very little gains in raster performance this decade.
I have a flagship 7-8 year old GPU in one machine and a mid-level modern GPU in another.
It’s flat out wrong to claim “very little gains” during this time. The difference between those two GPUs is huge in games. The modern GPU also does it with far less power and noise.
I can’t understand this HN mentality that modern hardware isn’t fast or that we’re not seeing gains.
100%. We’ve seen crazy swings in RAM prices before.
A colleague who worked with me about 10 years ago on a VDI project ran some numbers and showed that if a Time Machine were available, we could have brought like 4 loaded MacBook Pros back and replaced a $1M HP 3PAR ssd array :)
You still do. There is no "AI movement" you need to participate in. You can grab a copy of SICP and a banged up ten year old thinkpad and compute away, your brain will thank you. It's like when people complain that culture is unaffordable because the newest Marvel movie tickets cost 50 bucks, go to the library or standardebooks.org, the entire Western canon is free
Well put. Since the 1980's consumer has been driving the segment. Even supercomputers were built out of higher end consumer hardware (or playstations in one example).
The move to cloud computing and now AI mean that we're back in the mainframe days.
True, it is reminiscent of a time before me when people were lucky to have mainframe access through university. To be fair this was a long time in the making with the also quite aggressive move to cloud computing. While I don't mind having access to free AI tools, they seem to start taking possession of the content as well
It's not like you need 64GB to have "democratized computation". We used to have 64MB and that was plenty. Unfortunately, software got slower more quickly than hardware got quicker.
> Unfortunately, software got slower more quickly than hardware got quicker.
Hard disagree. A $600 Mac Mini with 16GB of RAM runs everything insanely faster than even my $5000 company-purchased developer laptops from 10 years ago. And yes, even when I run Slack, Visual Studio Code, Spotify, and a gazillion Chrome tabs.
The HN rhetoric about modern computing being slow is getting strangely disconnected from the real world. Cheap computers are super fast like they've never been before, even with modern software.
You brought up a light computing load that a laptop from like 2005 wouldn't struggle with?
People ran multiple browser windows, a 3D video game, irc (chat application), teamspeak/ventrilo (voice chat) and winamp (music) all at once back in the early 2000s. This is something an 8 year old phone can do these days.
I’m responding to the comment above claiming that modern software is slow on modern hardware. It’s an HN meme to claim that Electron apps are unusable.
They are slow though. There are noticeable delays on any machine I've used with them compared to native desktop applications. It's absolutely not just a meme.
It is pretty important if you are doing things like 3d animation, video editing, or advanced CAD software. Plus software in general has ballooned its memory requirements and expectations. Even my 11 year old PC had to have a RAM upgrade a few years ago just because software updates suck up so much extra memory, and there is almost nothing consumers can do about it.
Bullshit. It was cramped and I wasn't able to do half of what I was wanting to actually do. Maybe it was plenty for your usecases, but such a small amount of memory was weak for my needs in the late 90s and 2000s. 64MB desktops struggled to handle the photo manipulations I wanted to do with scanned images. Trying to do something like edit video on a home PC was near impossible with that limited amount of memory. I was so happy when we managed to get a 512MB machine a few years later, it made a lot of my home multimedia work a lot better.
There are some use cases that simply require a lot of memory because they do, but I'm talking in general. Software that doesn't have a good excuse, used to run in 64MB how it now runs in 64GB.
Besides, you just said you only needed 512MB, which is still nothing these days.
> Besides, you just said you only needed 512MB, which is still nothing these days.
I didn't say I "only needed 512MB", only that things were a lot better once we got a 512MB machine. Things continued to get massively better as I upgraded to a 1GB machine, an 8GB machine, etc.
> I'm talking in general
Isn't doing some light picture editing/organizing, playing back multimedia, etc. pretty dang general computer use these days? Or what, is "general" computer usage entirely limited to 80 column text manipulation? You'd have a hard time even just keeping my displays drawn with 64MB of memory at the resolutions and bit depths and multiple desktops that are common.
I play around with retro computers (especially the early/mid 90s for that nostalgia) and I'm constantly reminded of how little memory we really had to play with back then, and these are on pretty much fully loaded home desktop machines. Have a Word document open and you're trying to play back an MP3 and have a couple browser windows open? Oof, good luck! You want to stream a video? I hope its about 10FPS at 320x240! Opening one photo from my camera today and you'll have used half the memory before its even hit the framebuffer.
I don't disagree per-se, but this is the sort of thing which happens when only a few businesses exist in a commodity market with high entry costs. IOW, it's not great, but it is predictable. See: Oil.
There is a zero lower bound on the interest rate. Excess capital means negative returns on capital. The money system can't express the state of the real world so either companies close down until the yield is positive, or the companies pass on the artificial minimum price onto the consumer. In both cases, the real world is forced to match the state of the money system.
Being shocked that companies try their best to deal with the bad cards they have been dealt with should be expected. The money system simply cannot express the concept of surplus capital or abundance. Positive interest means capital is scarce, so capital must be made scarce even if there is abundance.
Before you come up with the argument that the interest rate is supposed to reflect a market property and therefore does not force itself upon the market, remember that I said that there is an artificial restriction in the money system that prevents the state of the real market to be expressed. The non-profit economy has never had a chance to exist, because our tools are too crude.
The non-profit economy includes resilient production with slight/minor overproduction.
Think about how stupid the idea of a guaranteed 0% yield bond is (aka cash). The government obligates itself to accept an infinite amount of debt if the real return on capital would ever fall negative. No wonder it has an incentive to inflate the value of the bond away.
I wonder how long it will take for China to flood the market with state-of-the-art modules. It's a pretty decent opportunity for them. They probably can hasten the build of new fabs more than many other nations.
But my guess is that this shortage is short-lived (mostly because of the threat above). There's no OPEC for tech.
Letting things go this unmanaged with a 3 year run way for AI demand seems a little hard to understand. In this case, not anticipating demand seems to creates more profit.
If LLMs' utility continues to scale with size (which seems likely as we begin training embodied AI on a massive influx of robotic sensor data) then it will continue to gobble up memory for the near future. We may need both increased production capacity _and_ a period of more efficient software development techniques as was the case when a new 512kb upgrade cost $1,000.