Can we just observe what a bizarre fucking moment we are living in?
Everyone knows the collapse of the AI bubble is imminent and most of those people know it’s going to be the proximal trigger for a global financial collapse and at the very least a long recession.
Everyone knows except the people who are investing billions of dollars into AI and their press.
I wonder if some people don’t understand that saying anymore. “Get that bag” means getting a full money bag. “Left with the bag” or “left holding the bag” means you have an empty money bag that has been emptied by others.
Your interpretation, to be left with an empty bag that previously had something valuable, is one of several interpretations, but not the earliest.
The original seems to be, the member of the group holding a bag of contraband when the authorities intervene.
I think this second meaning is the more compelling: the group sought a benefit but only one bears the cost. Or, each member of the group took the risk but that risk only materialised for one member.
I mean, after you share out a bag of gold, including giving yourself a share, the bag will be empty.
No, I know what it means. I don’t think it applies here. It would apply to something like a pump and dump scheme, where the people know the bag is already empty, but they fluff it up and try to pass it off on someone else before the whole thing collapses.
That’s not what these companies are doing. They legitimately believe that AI will eventually become profitable and are trying to corner the market and be the dominant player. They WANT the crash. They just want to be the ones that survive it. That’s not the same as trying to leave someone else with the bag at the end of it.
I don’t think that’s really the same. Not being left with the bag means you know already know the bag is empty and pass it on to someone else before they realize it’s empty. Like a pump and dump scheme.
That’s not what these companies are doing. They’re not trying to leave investors holding the empty bag. They are legitimately trying to turn profitable and corner the market.
The fibre in the ground from the dotcom bubble was still used… eventually.
Most of the 100’s of billions of dollars of GPUs that have just been sitting on shelves in warehouses and will continue to sit there for years even if there isn’t a crash will be nearly worthless (relative to the prices paid) whether they try to mass liquidate at a small fraction of the purchase price to pay back some of their 10%+ interest loans they used to buy them or they sit around for a decade until there’s a place to put them and at that point they could be so inefficient, they’re not worth the money to install them and power to run them.
That’s if they don’t decide the best course of action is to launch them into LEO lol. If they do (they will–they’re that stupid), the infrastructure will burn up in less than five years.
The LOE stuff is the dumbest shit possible I’ve ever heard for data centers I swear to God. Who the hell comes up with these ideas 😭 space is such a difficult environment to work in, absolutely not suited for data centers. You barely get any benefit out of it at all, just somewhat more efficient solar panels that could have been worked around by having more down on earth
There’s a pretty big benefit to the solar panels: more efficient and also can be placed in a solar-synchronous orbit that is always illuminated. I can’t imagine that power advantage outweighs the disadvantages of cooling power-hungry hardware with no way to reject that heat to air or water. I’m with you: WTF are they thinking?
I heard Google just launched a test bed. Basically one mini server the size of a refrigerator that can only operate for 15 minutes before it needs to shut off to prevent heat damage. I’m guessing that thing will prove the case against celestial data centers, but we’ll see. The only good thing to come of this is that it might help advance cooling tech for satellites and suchlike.
Yeah, chips are more of a perishable good than any kind of durable infrastructure, and at the end of the day, the models themselves have no durable value because information is never truly scarce, regardless of how they try to make us pretend it is or poison the well with lies. There’s no magic solution to making LLMs actual AGI or even effective tools outside of their narrow scope. Not only that, the way they’re using it is extremely reckless, deeply inefficient, and supremely cost intensive, on a level that makes the Manhattan Project look wholesome. There isn’t going to be a “last man standing” to win, it’s just going to go up like flash powder and take everyone in the dogpile with it.
Just so we are clear the dot com crash was a result of over commercialization and venture capitalization of the World Wide Web (hence “dot com” crash and not Newsgroup crash). The Internet was nearly 30 years old at that point. What became AOL went online in 1985, 8 years before commercial interests were allowed on the WWW, just as one for instance.
I didn’t say AI is going to go away. I said there are many unscrupulous investors that are trying to game the system to take money from suckers knowing full well there is a crash coming and planning to pull out at the top.
Oh they definitely know, it’s a deliberate and methodical way to pump the money around between corporations, move it from consumers to corporations and from corporations to the C-suite. They know the bubble will burst. They only care about how much money they have extracted from it to still come out on top after it all comes crashing down.
It’s a pyramid scheme, we’re the ones at the bottom and we can’t get out.
Everyone knows the collapse of the AI bubble is imminent
Paul Krugman was predicting the Housing Crash as early as 2004 in his NYT column.
Everyone knows except the people who are investing billions of dollars into AI and their press.
The theory of the modern tech industry is that you can simply outrun your problems. Meta/Facebook has done this repeatedly, leaping over multi-billion-dollar failure after failure based entirely on the inertia of their core product growth combined with low interest rates and a bullish long term economic outlook.
Even in the worst-case scenario, we have such an enormous array of tools to manipulate markets - as demonstrated in 2008 and 2014 and 2020 - that investors are reluctant to sell with a massive federal bailout always on the horizon. And we’re far from the worst-case scenario. '26 Q2 was a historic earnings season. We are in the middle of the biggest bull market in history.
Here’s the thing about bubbles - when people believe that collapse is imminent, they hold back and don’t dive in. It is only when the overwhelming majority believe that this new thing can’t lose, is going to change the world, and that we are in a “new paradigm” where the old rules don’t matter, that the bubble has peaked and collapse is imminent. We’ve probably got another year or so before collapse. Too many people are still skeptical.
I think one thing that’s helping people hang on to delusion is the scarcity of real information. For instance, it’s prefectly possible that 2/3rds of NVIDIA GPUS purchased for data centers are just sitting in a warehouse somewhere, uninstalled. But that’s just a guess based on available data, but we can’t know for sure.
I think with Anthropic and OpenAI’s upcoming IPOs, we’re going to get a lot of information that will eventually filter through the slavish business press and into public consciousness as facts instead of opinions.
I think this bubble is a bit different and not really reliant on the general public’s view or support. They will attempt to ram this down our throats, with violence where necessary, because the more buy-in the better, but the clock is already ticking because they made promises of profit they cannot keep and when the debt comes due they will not be able to pay
Bond market already knows.
But this time it’s an ugly double whammy as it’s both the AI companies and the US government that won’t be able to pay their debt, and they are interlinked.
If it comes to it, the US is constitutionally barred from just defaulting on its debts. And the US government can just print more money. Which will massively devalue the USD, tanking the economy and killing their ability to borrow more in the future, but it’ll still technically allow them to pay off the current debt.
They know their promises cannot be kept. They are building a surveillance state for the government. When the bubble bursts, the government will swoop in and bail them out because they are too big to fail and hand over most of these AI resources to the US government.
I feel like we just left the era of delusions of “new paradigm” and entering the skepticism part. Of course plenty of people have always been skeptical, but not the people with enough money to cause this.
OTOH, timing can still be difficult to predict. Surely people knew tulips weren’t worth more than homes, yet it took years after tulips reached that price for the peak to be reached iirc. Even if people know the crazy is irrational and cannot go on forever, there is still FOMO. The peak is likely to happen after a lot of people know the thing will have to crash eventually, but enough continue to try to eek would a little more profit while not being left holding the bag.
The peak only happens after a good majority think the crash is imminent and are ready to sell off to avoid being left with nothing. Hence why many of these companies are trying to be added to indicies quickly so that pension/retirement funds can be the one stuck with much of the losses while the billionaires just to onto their next grift.
I wondering if there is any benefit for me in having all the RAM my motherboard allows for when this bubble burst, and Ram prices fall like a rock to rock bottom.
I think we’ve reached the top of the S-Curve for personal computing hardware. People are staying with their old hardware longer and software is no longer targeting high end and future devices.
So unless there’s a reason you think you might need that RAM, I’d put the money somewhere else.
Absolutely 100% my experience too, my computer is a mix of 2-5 year old hardware, and I’m still on the AM4 socket which is 9 years old now! It is a newer motherboard but I started on that socket way back in 2017! The upgrade path for AM4 has been insane. 😋
In the 90’s I wouldn’t have a single component that was 2 years old.
IDK if that’s because I’m older, but definitely it seems the advantage of upgrading every year is diminishing, and the price is increasing. Sometimes I would even buy a new GPU 2 times in the same year. 😜
My desktop is 10-year-old gaming PC. If you look at the Steam Survey, it’s about middle-of-the-road today.
It’s not just PC. It’s an open question whether anyone is going to be the next gen of game console, and that’s partially because gaming itselfhas been enshittified, but also for the same reason people aren’t running out to buy 8K TV’s–the minimal improvement in subjective experience is not worth the price.
People are more likely to replace phones because they break than because they are upgrading–partially why Apple and Samsung are doing their DARNEDEST to lock down repairs, so they can drive up the cost of repairs to force people to buy a new phone.
We’re approaching a world where we think of what was formerly “high technology” the same way we think of refrigerators–as basic devices that can generically and permanently do a necessary set of tasks.
I always hated consoles, as a waste of a good computer you couldn’t do anything with but game.
But I think consoles will be fine, consoles are mostly for children, and they will get exclusives to drive up sales.
But this is a bad time to make a new console, still AFAIK both Sony and Microsoft are planning new ones next year?
It used to be you’d get highish-end hardware at cheaper price because of mass production and manufacturer incentives (made up for on games).
Now with people carrying massive game libraries, they’re less likely to buy new games, at the same time as console gaming becomes enshittified cash grabs, at the same time as tech has hit diminishing returns in an industry where generational hardware sales drive software purchases.
Playstation has a few exclusives, but XBox has literally none–it’s all cross compatible with Windows, and almost every single one of those titles is playable via Steam on Linux. You can build an equivalent PC for the same price or less than the equivalent console, and have improved upgrade and repair paths, plus mods, plus Steam sales… consoles make less and less financial sense every day.
The only console that is doing well is Nintendo, and that’s because they built around content not hardware.
And that said, most kids are growing up gaming on tablets and phones, and they haven’t gone through the evolutionary learning process that a 35-year-old life-long gamer. Chances are you hand an 8-year-old a complex two-stick game and they’ll do about as well as my father would.
I think likely we’ll have a profusion of quirky handheld consoles, but the big box consoles will be replaced with computers.
the big box consoles will be replaced with computers.
I think you are wrong, the PC is dying, and the consoles will have even bigger advantage from mass production in the future.
Most people don’t care to maintain a PC just to play games, and most people don’t really have any other use for a PC anymore.
Even without subsidies from Microsoft and Sony, the current gen consoles were cheaper than an equivalent performing PC when they arrived. But in the future that may not be true, it’s all down to profit maximization, because there is no doubt a console can be sold cheaper than an equivalent performing PC.
Depends on what you wanna use it for. If you wanna host a bunch of services, do scientific computing, or play modded video games with really bad memory leaks and memory optimization, perhaps?
The problem is memory is expensive because fabs have switched to making HBM3 and EEC LPDDRX5 (soldered memory, like what’s in a phone and some laptops). So even when the crash happens, we won’t get a mass sell-off of useful consumer hardware. Likewise, the “GPUs” don’t even have video outputs.
I’m just a casual user today, I used to stress my computers to the limit, but not anymore.
So I most definitely would make very poor use of maxing my RAM.
Back in the day I loaded DVD 's into virtual drives in RAM, to make games load faster.
For some reason the Linux kernel works best with a swap disk, so I had that in RAM too, and of course temporary files.
These things are of course banal although they were extremely efficient back in the day, and this was a thing I learned already back in the 80’s where the Amiga made similar things possible.
Today with fast SSD drives I don’t feel the need, and to be honest I already have more RAM than I need. And since I don’t program anymore either, which I always found interesting to explore the capabilities of new hardware, I have become an “average” user. 😳
Can we just observe what a bizarre fucking moment we are living in?
Everyone knows the collapse of the AI bubble is imminent and most of those people know it’s going to be the proximal trigger for a global financial collapse and at the very least a long recession.
Everyone knows except the people who are investing billions of dollars into AI and their press.
Pretty sure they know too. They’ve simply arranged things so that they will still come out well ahead even so.
A lot of them do know and are planning to be the ones not left with the bag
Eh kind of. AI is not going to go away. It’s going to be just like the internet in the early days. With a similar crash to the dot-com crash.
They’re just all trying to be the one that doesn’t go bankrupt during the crash.
trying to be the one that doesn’t go bankrupt during the crash == planning to be the ones not left with the bag
I wonder if some people don’t understand that saying anymore. “Get that bag” means getting a full money bag. “Left with the bag” or “left holding the bag” means you have an empty money bag that has been emptied by others.
There’s a full discussion of the phrase here
Your interpretation, to be left with an empty bag that previously had something valuable, is one of several interpretations, but not the earliest.
The original seems to be, the member of the group holding a bag of contraband when the authorities intervene.
I think this second meaning is the more compelling: the group sought a benefit but only one bears the cost. Or, each member of the group took the risk but that risk only materialised for one member.
I mean, after you share out a bag of gold, including giving yourself a share, the bag will be empty.
No, I know what it means. I don’t think it applies here. It would apply to something like a pump and dump scheme, where the people know the bag is already empty, but they fluff it up and try to pass it off on someone else before the whole thing collapses.
That’s not what these companies are doing. They legitimately believe that AI will eventually become profitable and are trying to corner the market and be the dominant player. They WANT the crash. They just want to be the ones that survive it. That’s not the same as trying to leave someone else with the bag at the end of it.
I thought “get that bag” was specifically to encourage women chasing money to buy synecdochic luxury purses.
I don’t think that’s really the same. Not being left with the bag means you know already know the bag is empty and pass it on to someone else before they realize it’s empty. Like a pump and dump scheme.
That’s not what these companies are doing. They’re not trying to leave investors holding the empty bag. They are legitimately trying to turn profitable and corner the market.
The fibre in the ground from the dotcom bubble was still used… eventually.
Most of the 100’s of billions of dollars of GPUs that have just been sitting on shelves in warehouses and will continue to sit there for years even if there isn’t a crash will be nearly worthless (relative to the prices paid) whether they try to mass liquidate at a small fraction of the purchase price to pay back some of their 10%+ interest loans they used to buy them or they sit around for a decade until there’s a place to put them and at that point they could be so inefficient, they’re not worth the money to install them and power to run them.
That’s if they don’t decide the best course of action is to launch them into LEO lol. If they do (they will–they’re that stupid), the infrastructure will burn up in less than five years.
The LOE stuff is the dumbest shit possible I’ve ever heard for data centers I swear to God. Who the hell comes up with these ideas 😭 space is such a difficult environment to work in, absolutely not suited for data centers. You barely get any benefit out of it at all, just somewhat more efficient solar panels that could have been worked around by having more down on earth
There’s a pretty big benefit to the solar panels: more efficient and also can be placed in a solar-synchronous orbit that is always illuminated. I can’t imagine that power advantage outweighs the disadvantages of cooling power-hungry hardware with no way to reject that heat to air or water. I’m with you: WTF are they thinking?
I heard Google just launched a test bed. Basically one mini server the size of a refrigerator that can only operate for 15 minutes before it needs to shut off to prevent heat damage. I’m guessing that thing will prove the case against celestial data centers, but we’ll see. The only good thing to come of this is that it might help advance cooling tech for satellites and suchlike.
Yeah, chips are more of a perishable good than any kind of durable infrastructure, and at the end of the day, the models themselves have no durable value because information is never truly scarce, regardless of how they try to make us pretend it is or poison the well with lies. There’s no magic solution to making LLMs actual AGI or even effective tools outside of their narrow scope. Not only that, the way they’re using it is extremely reckless, deeply inefficient, and supremely cost intensive, on a level that makes the Manhattan Project look wholesome. There isn’t going to be a “last man standing” to win, it’s just going to go up like flash powder and take everyone in the dogpile with it.
Just so we are clear the dot com crash was a result of over commercialization and venture capitalization of the World Wide Web (hence “dot com” crash and not Newsgroup crash). The Internet was nearly 30 years old at that point. What became AOL went online in 1985, 8 years before commercial interests were allowed on the WWW, just as one for instance.
I think a more concise explanation is that everyone knows that shares are over-valued, but no one really knows by how much.
I didn’t say AI is going to go away. I said there are many unscrupulous investors that are trying to game the system to take money from suckers knowing full well there is a crash coming and planning to pull out at the top.
Thats why they are pumping these IPO valuations
We all know that an industry that spends more than it earns should collapse but I’m not so certain it is imminent.
“Markets can remain irrational longer than you can remain solvent.” feels like a quote that holds very true in today’s market.
Oh they know
There is going to be a LOT of used hardware on the market if it does.
Oh they definitely know, it’s a deliberate and methodical way to pump the money around between corporations, move it from consumers to corporations and from corporations to the C-suite. They know the bubble will burst. They only care about how much money they have extracted from it to still come out on top after it all comes crashing down.
It’s a pyramid scheme, we’re the ones at the bottom and we can’t get out.
The investors know too. They just need to hype it up so the next is bigger than them and bail them out.
It’s a pyramid scheme for billionaires. They just hope they’re not at the bottom of the pyramid.
Same as NFTs and every other bubble where it was clear that it was a bubble
Paul Krugman was predicting the Housing Crash as early as 2004 in his NYT column.
The theory of the modern tech industry is that you can simply outrun your problems. Meta/Facebook has done this repeatedly, leaping over multi-billion-dollar failure after failure based entirely on the inertia of their core product growth combined with low interest rates and a bullish long term economic outlook.
Even in the worst-case scenario, we have such an enormous array of tools to manipulate markets - as demonstrated in 2008 and 2014 and 2020 - that investors are reluctant to sell with a massive federal bailout always on the horizon. And we’re far from the worst-case scenario. '26 Q2 was a historic earnings season. We are in the middle of the biggest bull market in history.
Well, sure, as far as government policy is concerned, the stock market is the economy, no wonder it’s going to be goosed until it collapses.
Put it together aligning with the housing collapse and we gonna be living great.
That is why you sell shovels rather than mine the gold yourself. These memory companies know what they are doing…
Here’s the thing about bubbles - when people believe that collapse is imminent, they hold back and don’t dive in. It is only when the overwhelming majority believe that this new thing can’t lose, is going to change the world, and that we are in a “new paradigm” where the old rules don’t matter, that the bubble has peaked and collapse is imminent. We’ve probably got another year or so before collapse. Too many people are still skeptical.
Timing a bubble pop is notoriously difficult.
I think one thing that’s helping people hang on to delusion is the scarcity of real information. For instance, it’s prefectly possible that 2/3rds of NVIDIA GPUS purchased for data centers are just sitting in a warehouse somewhere, uninstalled. But that’s just a guess based on available data, but we can’t know for sure.
I think with Anthropic and OpenAI’s upcoming IPOs, we’re going to get a lot of information that will eventually filter through the slavish business press and into public consciousness as facts instead of opinions.
I think this bubble is a bit different and not really reliant on the general public’s view or support. They will attempt to ram this down our throats, with violence where necessary, because the more buy-in the better, but the clock is already ticking because they made promises of profit they cannot keep and when the debt comes due they will not be able to pay
Bond market already knows. But this time it’s an ugly double whammy as it’s both the AI companies and the US government that won’t be able to pay their debt, and they are interlinked.
If it comes to it, the US is constitutionally barred from just defaulting on its debts. And the US government can just print more money. Which will massively devalue the USD, tanking the economy and killing their ability to borrow more in the future, but it’ll still technically allow them to pay off the current debt.
Technically, yes - but not in a way that foreign investors would want to lend any more money in the future.
They know their promises cannot be kept. They are building a surveillance state for the government. When the bubble bursts, the government will swoop in and bail them out because they are too big to fail and hand over most of these AI resources to the US government.
I feel like we just left the era of delusions of “new paradigm” and entering the skepticism part. Of course plenty of people have always been skeptical, but not the people with enough money to cause this.
OTOH, timing can still be difficult to predict. Surely people knew tulips weren’t worth more than homes, yet it took years after tulips reached that price for the peak to be reached iirc. Even if people know the crazy is irrational and cannot go on forever, there is still FOMO. The peak is likely to happen after a lot of people know the thing will have to crash eventually, but enough continue to try to eek would a little more profit while not being left holding the bag.
The peak only happens after a good majority think the crash is imminent and are ready to sell off to avoid being left with nothing. Hence why many of these companies are trying to be added to indicies quickly so that pension/retirement funds can be the one stuck with much of the losses while the billionaires just to onto their next grift.
I wondering if there is any benefit for me in having all the RAM my motherboard allows for when this bubble burst, and Ram prices fall like a rock to rock bottom.
Honestly, probably not.
I think we’ve reached the top of the S-Curve for personal computing hardware. People are staying with their old hardware longer and software is no longer targeting high end and future devices.
So unless there’s a reason you think you might need that RAM, I’d put the money somewhere else.
Absolutely 100% my experience too, my computer is a mix of 2-5 year old hardware, and I’m still on the AM4 socket which is 9 years old now! It is a newer motherboard but I started on that socket way back in 2017! The upgrade path for AM4 has been insane. 😋
In the 90’s I wouldn’t have a single component that was 2 years old.
IDK if that’s because I’m older, but definitely it seems the advantage of upgrading every year is diminishing, and the price is increasing. Sometimes I would even buy a new GPU 2 times in the same year. 😜
My desktop is 10-year-old gaming PC. If you look at the Steam Survey, it’s about middle-of-the-road today.
It’s not just PC. It’s an open question whether anyone is going to be the next gen of game console, and that’s partially because gaming itselfhas been enshittified, but also for the same reason people aren’t running out to buy 8K TV’s–the minimal improvement in subjective experience is not worth the price.
People are more likely to replace phones because they break than because they are upgrading–partially why Apple and Samsung are doing their DARNEDEST to lock down repairs, so they can drive up the cost of repairs to force people to buy a new phone.
We’re approaching a world where we think of what was formerly “high technology” the same way we think of refrigerators–as basic devices that can generically and permanently do a necessary set of tasks.
I always hated consoles, as a waste of a good computer you couldn’t do anything with but game.
But I think consoles will be fine, consoles are mostly for children, and they will get exclusives to drive up sales.
But this is a bad time to make a new console, still AFAIK both Sony and Microsoft are planning new ones next year?
The economics of consoles are in shambles.
It used to be you’d get highish-end hardware at cheaper price because of mass production and manufacturer incentives (made up for on games).
Now with people carrying massive game libraries, they’re less likely to buy new games, at the same time as console gaming becomes enshittified cash grabs, at the same time as tech has hit diminishing returns in an industry where generational hardware sales drive software purchases.
Playstation has a few exclusives, but XBox has literally none–it’s all cross compatible with Windows, and almost every single one of those titles is playable via Steam on Linux. You can build an equivalent PC for the same price or less than the equivalent console, and have improved upgrade and repair paths, plus mods, plus Steam sales… consoles make less and less financial sense every day.
The only console that is doing well is Nintendo, and that’s because they built around content not hardware.
And that said, most kids are growing up gaming on tablets and phones, and they haven’t gone through the evolutionary learning process that a 35-year-old life-long gamer. Chances are you hand an 8-year-old a complex two-stick game and they’ll do about as well as my father would.
I think likely we’ll have a profusion of quirky handheld consoles, but the big box consoles will be replaced with computers.
I think you are wrong, the PC is dying, and the consoles will have even bigger advantage from mass production in the future.
Most people don’t care to maintain a PC just to play games, and most people don’t really have any other use for a PC anymore.
Even without subsidies from Microsoft and Sony, the current gen consoles were cheaper than an equivalent performing PC when they arrived. But in the future that may not be true, it’s all down to profit maximization, because there is no doubt a console can be sold cheaper than an equivalent performing PC.
Depends on what you wanna use it for. If you wanna host a bunch of services, do scientific computing, or play modded video games with really bad memory leaks and memory optimization, perhaps?
The problem is memory is expensive because fabs have switched to making HBM3 and EEC LPDDRX5 (soldered memory, like what’s in a phone and some laptops). So even when the crash happens, we won’t get a mass sell-off of useful consumer hardware. Likewise, the “GPUs” don’t even have video outputs.
I’m just a casual user today, I used to stress my computers to the limit, but not anymore.
So I most definitely would make very poor use of maxing my RAM.
Back in the day I loaded DVD 's into virtual drives in RAM, to make games load faster.
For some reason the Linux kernel works best with a swap disk, so I had that in RAM too, and of course temporary files.
These things are of course banal although they were extremely efficient back in the day, and this was a thing I learned already back in the 80’s where the Amiga made similar things possible.
Today with fast SSD drives I don’t feel the need, and to be honest I already have more RAM than I need. And since I don’t program anymore either, which I always found interesting to explore the capabilities of new hardware, I have become an “average” user. 😳