i mean my line is always "when." people have been saying ai is a bubble and will crash since the beginning. the question has always been "when."
i think the best framing of the ai crash is not in terms of software, but in terms of real estate and infrastructure. the real cost is hidden in construction contracts and people buying up land (and chips, transformers, generators, etc), not in technology. this will be the 2008 real estate crash, not the 2000 dot com bubble. (altho this point is not my most well researched point, i tbh need to look more into hard numbers of where the most risk is distributed, ai software startups or data center contracts. but i also am in grad school i have papers to publish and no time.)
im also gonna push back on the "diseconomy of scale" point. while its true that the best models rn use the most resources, we are building better smaller models as well. i recently set one up on my 12 gb vram and while its not nearly as good as claude, it works. i make it write really simple code for me. and we are still improving small models. i think theres some level of hope to economy of scale. obviously there are going to be physical limits, but im sure there will be a way to get economies of scale to work.
When it comes to the AI crash honestly my biggest concern is what is going to happen to the job market during and the years following the crash. Not sure how things are in the rest of the world, but as someone in their early 30s working in the IT industry in Sweden I’ve never seen the market this competitive before, even for mid level and senior roles, and it worries me what the future of employment is going to look like.
Maybe those older than me have been through this kind of thing before in 2008-2009 and in the early 2000s but the state of the IT job market in the last year or two has been really concerning to me. Anecdotally I’ve also heard it’s very tough for new graduates these days.
AI investment will crash but AI itself (the technology) will continue thriving, learning, improving and there is absolutely no way to stop it. The only reading on the crystal ball is if US companies fail, China will take the lead by leaps and bounds, so the only solution is to keep pushing the cart until the wheels come off or we all cross the finish line, together.
It can't 'learn' on its own. Models only get better with mountains of RnD for data, training, and lots of fine tuning.
So the moment investment dries up, models stop improving.
However, it's likely we'll get good 80/20 solutions where you get most of the performance of the then-unsustainable high end models for significantly less compute.
Those analyses tend to not consider the government interest in having AI as a weapon: now AI labs are fully part of the military sector, making them much more supported by the system
> Most new technologies have been welcomed by the public with open arms.
I'm not going to predict how this is going to turn out in either direction but this statement gives me pause. I don't think that's ever been true. Yes, the siren song is strong but initially most new tech is met with skepticism. Are we so quick to forget "the internet/computers are just a fad"-type thinking?
To parallel to The Big Short this is the point in the movie where folks realize it’s mathematically impossible for things to not implode and so players are quietly positioning themselves for that eventuality before things are allowed to blow.
It’s been a dramatic shift these last six months but everywhere I look now folks are quietly preparing their battle armor to survive what’s about to unfold.
The tech will stay, but the AI business landscape will have a market-cleansing forest fire.
There’s a whole generation in tech now that’s never seen what happens when a bubble like this unravels. I fully expect we’ll see the likes of offices just abandoned overnight with food still in the fridge as AI company after AI company just vaporizes.
this is a free fiat printining era. you fiction eating people will not realize that they can and will print T$ for years to save all and they are already printing too.
seriously, how hard is to get this? finance changed in 2020/21 forever. fiat printing level size several times bigger than all money total before
There's a very nonzero chance that the AI expenses are not a race to recoup investments as they are a bet (with other people's money) that AI will lead to longevity or immortality, a bet made by aging tech oligarchs. Notice that Ellison is betting his entire company on it, which due to his secured debt is the same as betting his personal fortune in the most extreme case.
I think after the AI crash the biggest winner will be Apple.
They will release MacBook Pro M10 or whatever that can comfortably run Opus5 levels of performance local model for software development/general ai that is baked in the MacOS for 7k USD.
I generally like this but I think the diseconomies of scale bit is narrow. Open models are showing us that it’s just the frontier companies in the west which are bloating and ignoring efficiency and that they’re leaving a lot of efficiency on the table.
When I joined my first startup, they said if you didn't have a good plan for going public or getting acquired by series C, then the D in series D stood for death. Uber et al raised like a J round. None of the rules around markets or investments are real, it's all vibes-based, and the decision-makers love talking to their ChatGPT mistresses too much to let this all come crashing down.
What implications might an AI crash have on the software developer job market? It’s likely that many AI software companies would fold and thus have to layoff their employees, but what would be the implications for the rest of the tech job market?
I often see developers giddy about the idea of AI being a bubble and waiting for the crash, but I suspect this event could actually be particularly terrible for us.
> Circular Revenues: A small handful of tech firms, chip manufacturers, and AI companies are propping each other up by investing and buying from each other.
> Increasing Corporate Skepticism: The news is full of stories of corporations that are throttling the employee use of AI since the costs to use the software are a lot higher than expected.
Real AI spend is out of control, with the news is full of stories about corporations trying to keep a lid on it, but also real AI spending is low and concentrated to a few firms.
I don't know. This doesn't feel very coherent to me, but rather like a collection of assertions that are adopted because they individually say something bearish about the industry.
Certainly some investments will have been overreaches, but I find it pretty unlikely that any of the compute build-out to date is going to be left sitting idle one or two or five years from now.
Maybe it would be good if more electricity generation became available for other purposes like heat pumps and electric cars? Although, perhaps it doesn’t help as much if it’s in the wrong place.
These days, I find it hard to predict the future. Things feel too complex, and the times seem different.
People say the AI bubble will burst, and I also think it will. But I want to think about how it might differ from other bubbles.
One positive factor is that during the internet startup era, there was almost no revenue. But now, big tech companies are generating profits and can absorb AI-scale losses.
The dot-com bubble burst, but the internet ended up being far more valuable than the bubble itself. I think the value was priced in early, and AI will be similar.
The core issue is always the same: there's a bubble and there are warnings, but you never know when it will collapse. And the people who act first on the warnings have to give up on the upside. The only winners are those who sell everything just before the crash. The problem is knowing when that moment is.
One difference from the dot-com bubble is that after it burst, a second wave of entrepreneurs leveraged the cheap telecom infrastructure that was left behind. So after the AI bubble bursts, will OpenAI and Anthropic fall and new companies emerge? I'm skeptical about that. Something feels different this time.
My biggest concern, though, is that if I suddenly couldn't use AI anymore, I'm worried about how long it would take to recover the coding skills I've lost after nearly a year of barely hand-coding. AI has become too deeply embedded in my life
Previous bubbles were characterised by "mass participation in the relevant markets"
This hasn't happened yet for AI, and there's a good chance it won't happen. Ie neither OpenAI nor Anthropic go public, or if they go public, the reception is "meh"
Nevertheless, independent of wider market sentiment, imho there is still a bubble in "closed models". One could, eg, pay attention to OpenAI/Anthropic/Copilot/Google's monthly new consumer subscriptions (Oversubscribed but in a hyperhyperreal sense lol. It seems that there are too many people queuing for this new restaurant, and those that have tasted the food all think it's ok but not healthy enough ("productivity mirage"). but people on the street can't see it because the bookings are all online. It may be that Gemini pro's subscriptions may be the last one to fall off the cliff. When that happens even the "open cloud models" bubble may pop (after everyone sees Google sunsetting Gemini, or making it exclusive to Apple lol)
Most of the post makes sense to me, even if I don't agree. This part however:
> The news is full of stories of corporations that are throttling the employee use of AI since the costs to use the software are a lot higher than expected
is not something I've seen. News is not "full of stories" of this behaviour. There are a few anecdotal stories here and there.
I've been reading a lot of stories like this lately. I'm no business genius, but you'd assume that investors are. Are they just blind or are they burning cash on purpose. What's the steelman argument here?
> I'm no business genius, but you'd assume that investors are.
Absolutely not a reasonable assumption at all. Most people are not very good at their jobs.
> What's the steelman argument here?
If you're right about your prediction, but wrong about the timing, then you're wrong. The markets can stay irrational longer than you can stay solvent. Meaning, even if you know it's going to crash eventually, you don't know if it's tomorrow or in three years. Betting on stocks going down is extremely risky -- you have to get the timing pretty much exactly right, or you lose everything (potentially much more than you put down, depending on how you do it). Meanwhile, bull positions are basically free money until this point. Getting completely out of the market means you lose to inflation, although many investors have taken this strategy anyway.
Looking at these investments through the lens of traditional businesses (which is the perspective taken by articles like this) won't make sense. It's not until you appreciate the expectation of how disruptive this technology will actually be does any of this make sense.
These people think they're on the verge of creating a technology which, at a minimum, would constitute an unprecedented superweapon (and, at the extreme, would usher in a new era of civilization). Even if you don't buy-in to the take that one of these companies will reach a singularity and create a superintelligence, the cybersecurity implications alone is enough to put these products into a category outside the confines of profitability. We're already starting to see these implications become reality.
If the US NEEDS an advanced AI on an existential level, then it doesn't really matter how much it costs to make or whether or not it can produce a profit. It'll be valuable one a scale where financials like that just don't apply.
I don't have much background in the area, but I am surprised to see that everyone here basically agrees a crash is imminent. There are disanalogies to past crashes that don't convince me that a big crash is definitively coming in the near term.
For example:
- Anthropic makes a profit right now and is seemingly on an exponential upward trajectory, so the debt being too much for it doesn't seem compelling to me.
- AI technology continues to get better exponentially and doesn't have any clear sign this trend is flattening. If anything, it's accelerating. So it's plausible the investor value is legitimate for these companies given the massive potential for continued profitability.
- I would say markets are typically very good predictors of the future. Many sophisticated investors know about the case for the future crash and are still buying at these valuations.
I am open to being wrong, but the assessment in the blog seems one-sided to me.
Polymarket currently puts the chance of such a downturn at 20% by December 2026. Seems like most people would put higher chances here, but I'm not convinced by the arguments. (https://polymarket.com/event/ai-bubble-burst-by)
> analysts have estimated that it will take $2 trillion a year in revenue to pay for the infrastructure that has already been built
I doubt any credible analyst has claimed that. What's the total AI capex that's already been spent? About $1T? It's a pretty absurd idea that those DCs need to make $2T/year for 5-7 years -> $10T-14T over their lifetime to break even.
(Yes, this is nitpicking in the sense that there are probably analysts talking about how projected and sustained capex at >1T/year will require 2T/year in revenue, so patching the article won't be a biggie. But this article is cosplaying as financial analysis and leads off with such an obviously incorrect argument. What does that say about the credibility of the rest of the article?)
> Diseconomies of Scale.
Another very basic mistake here. The author starts talking about efficiency in the context of past technologies. That's lower unit costs as scale increases.
But for AI, they seem to switch from talking about unit costs to total costs. Or at least I can't explain what they say about models getting more expensive over time in any other way, because that is not true about unit costs.
We've never seen economies of scale as large as for AI. For a given quality level, the cost has been dropping at >10x per year, not increasing.
The post is just full of half-baked conjectures masquerading as facts... combined with "things they read" by unspecified authors and sources... the author seems to prefer engaging in AI doomerism as opposing to actually understanding.
> Diseconomies of Scale. Every new technology I can think of thrived, in part, due to economies of scale, where the larger the industry grew, the more efficient it got. AI is going in the opposite direction, where every new AI model consumes more resources than its predecessors. This may turn out to be the fatal flaw – the bigger the industry gets, the more its operating costs increase.
[Agreed] Newer gens of models are more power-efficient per task, not less.
It's possible that the demands on the model are growing faster than it gets efficient though. A lot of the improvement in output quality/metrics come from more churning and passes.
You could argue that a RTX5090 needs less power than a 1996 3Dfx Voodoo 1 to render GLQuake at 320x200/60fps, but it's rarely being used for that.
I agree a lot of the dooming is from people who don't really do the maths. Total capex so far is probably about $1T. Recent annual earnings are $132bn at Google, $101bn at msft, 60bn at Meta, $120bn nvda adding up to $413bn after tax. They could pretty much write down the whole capex to zero against two years earnings from those. It's not that out of whack.
39 comments
[ 0.23 ms ] story [ 6.6 ms ] threadMaybe those older than me have been through this kind of thing before in 2008-2009 and in the early 2000s but the state of the IT job market in the last year or two has been really concerning to me. Anecdotally I’ve also heard it’s very tough for new graduates these days.
So the moment investment dries up, models stop improving.
However, it's likely we'll get good 80/20 solutions where you get most of the performance of the then-unsustainable high end models for significantly less compute.
I'm not going to predict how this is going to turn out in either direction but this statement gives me pause. I don't think that's ever been true. Yes, the siren song is strong but initially most new tech is met with skepticism. Are we so quick to forget "the internet/computers are just a fad"-type thinking?
To parallel to The Big Short this is the point in the movie where folks realize it’s mathematically impossible for things to not implode and so players are quietly positioning themselves for that eventuality before things are allowed to blow.
It’s been a dramatic shift these last six months but everywhere I look now folks are quietly preparing their battle armor to survive what’s about to unfold.
The tech will stay, but the AI business landscape will have a market-cleansing forest fire.
There’s a whole generation in tech now that’s never seen what happens when a bubble like this unravels. I fully expect we’ll see the likes of offices just abandoned overnight with food still in the fridge as AI company after AI company just vaporizes.
seriously, how hard is to get this? finance changed in 2020/21 forever. fiat printing level size several times bigger than all money total before
They will release MacBook Pro M10 or whatever that can comfortably run Opus5 levels of performance local model for software development/general ai that is baked in the MacOS for 7k USD.
Let’s argue about SQL vs NoSQL again. I miss that.
- if every autonomous self-driving car is going to need some sort of edge AI data center nearby (for instantaneous exceptional incident handling)
- if every household robot is going to need some sort of edge AI data center nearby (for instantaneous exceptional incident handling)
- if there are going to be millions of self-driving cars and autonomous robots joining us in the next decade
...then we will need lots more semi-conductor chips, and data centers in lots more places, in the next decade.
I often see developers giddy about the idea of AI being a bubble and waiting for the crash, but I suspect this event could actually be particularly terrible for us.
> Increasing Corporate Skepticism: The news is full of stories of corporations that are throttling the employee use of AI since the costs to use the software are a lot higher than expected.
Real AI spend is out of control, with the news is full of stories about corporations trying to keep a lid on it, but also real AI spending is low and concentrated to a few firms.
I don't know. This doesn't feel very coherent to me, but rather like a collection of assertions that are adopted because they individually say something bearish about the industry.
Certainly some investments will have been overreaches, but I find it pretty unlikely that any of the compute build-out to date is going to be left sitting idle one or two or five years from now.
Also, cheaper RAM would be nice.
People say the AI bubble will burst, and I also think it will. But I want to think about how it might differ from other bubbles.
One positive factor is that during the internet startup era, there was almost no revenue. But now, big tech companies are generating profits and can absorb AI-scale losses.
The dot-com bubble burst, but the internet ended up being far more valuable than the bubble itself. I think the value was priced in early, and AI will be similar.
The core issue is always the same: there's a bubble and there are warnings, but you never know when it will collapse. And the people who act first on the warnings have to give up on the upside. The only winners are those who sell everything just before the crash. The problem is knowing when that moment is.
One difference from the dot-com bubble is that after it burst, a second wave of entrepreneurs leveraged the cheap telecom infrastructure that was left behind. So after the AI bubble bursts, will OpenAI and Anthropic fall and new companies emerge? I'm skeptical about that. Something feels different this time.
My biggest concern, though, is that if I suddenly couldn't use AI anymore, I'm worried about how long it would take to recover the coding skills I've lost after nearly a year of barely hand-coding. AI has become too deeply embedded in my life
This hasn't happened yet for AI, and there's a good chance it won't happen. Ie neither OpenAI nor Anthropic go public, or if they go public, the reception is "meh"
Nevertheless, independent of wider market sentiment, imho there is still a bubble in "closed models". One could, eg, pay attention to OpenAI/Anthropic/Copilot/Google's monthly new consumer subscriptions (Oversubscribed but in a hyperhyperreal sense lol. It seems that there are too many people queuing for this new restaurant, and those that have tasted the food all think it's ok but not healthy enough ("productivity mirage"). but people on the street can't see it because the bookings are all online. It may be that Gemini pro's subscriptions may be the last one to fall off the cliff. When that happens even the "open cloud models" bubble may pop (after everyone sees Google sunsetting Gemini, or making it exclusive to Apple lol)
US, not Korea https://archive.ph/WBSCj
> The news is full of stories of corporations that are throttling the employee use of AI since the costs to use the software are a lot higher than expected
is not something I've seen. News is not "full of stories" of this behaviour. There are a few anecdotal stories here and there.
https://www.advisorpedia.com/media/media/2026/02/23/trailing...
Somehow we've managed to turn software from a machine for printing money to a machine for setting it on fire.
Absolutely not a reasonable assumption at all. Most people are not very good at their jobs.
> What's the steelman argument here?
If you're right about your prediction, but wrong about the timing, then you're wrong. The markets can stay irrational longer than you can stay solvent. Meaning, even if you know it's going to crash eventually, you don't know if it's tomorrow or in three years. Betting on stocks going down is extremely risky -- you have to get the timing pretty much exactly right, or you lose everything (potentially much more than you put down, depending on how you do it). Meanwhile, bull positions are basically free money until this point. Getting completely out of the market means you lose to inflation, although many investors have taken this strategy anyway.
These people think they're on the verge of creating a technology which, at a minimum, would constitute an unprecedented superweapon (and, at the extreme, would usher in a new era of civilization). Even if you don't buy-in to the take that one of these companies will reach a singularity and create a superintelligence, the cybersecurity implications alone is enough to put these products into a category outside the confines of profitability. We're already starting to see these implications become reality.
If the US NEEDS an advanced AI on an existential level, then it doesn't really matter how much it costs to make or whether or not it can produce a profit. It'll be valuable one a scale where financials like that just don't apply.
For example:
- Anthropic makes a profit right now and is seemingly on an exponential upward trajectory, so the debt being too much for it doesn't seem compelling to me.
- AI technology continues to get better exponentially and doesn't have any clear sign this trend is flattening. If anything, it's accelerating. So it's plausible the investor value is legitimate for these companies given the massive potential for continued profitability.
- I would say markets are typically very good predictors of the future. Many sophisticated investors know about the case for the future crash and are still buying at these valuations.
I am open to being wrong, but the assessment in the blog seems one-sided to me.
Polymarket currently puts the chance of such a downturn at 20% by December 2026. Seems like most people would put higher chances here, but I'm not convinced by the arguments. (https://polymarket.com/event/ai-bubble-burst-by)
I doubt any credible analyst has claimed that. What's the total AI capex that's already been spent? About $1T? It's a pretty absurd idea that those DCs need to make $2T/year for 5-7 years -> $10T-14T over their lifetime to break even.
(Yes, this is nitpicking in the sense that there are probably analysts talking about how projected and sustained capex at >1T/year will require 2T/year in revenue, so patching the article won't be a biggie. But this article is cosplaying as financial analysis and leads off with such an obviously incorrect argument. What does that say about the credibility of the rest of the article?)
> Diseconomies of Scale.
Another very basic mistake here. The author starts talking about efficiency in the context of past technologies. That's lower unit costs as scale increases.
But for AI, they seem to switch from talking about unit costs to total costs. Or at least I can't explain what they say about models getting more expensive over time in any other way, because that is not true about unit costs.
We've never seen economies of scale as large as for AI. For a given quality level, the cost has been dropping at >10x per year, not increasing.
That figure is $2T in the next 4 years though. $2T/year would be quite silly.
> Diseconomies of Scale. Every new technology I can think of thrived, in part, due to economies of scale, where the larger the industry grew, the more efficient it got. AI is going in the opposite direction, where every new AI model consumes more resources than its predecessors. This may turn out to be the fatal flaw – the bigger the industry gets, the more its operating costs increase.
[Agreed] Newer gens of models are more power-efficient per task, not less.
This is a low quality post full of basic errors.
You could argue that a RTX5090 needs less power than a 1996 3Dfx Voodoo 1 to render GLQuake at 320x200/60fps, but it's rarely being used for that.
Not mentioning that we are kind at the point in the Wait but Why cartoon between that's cute and wft https://waitbutwhy.com/wp-content/uploads/2015/01/Intelligen...
From the article from ten years ago https://waitbutwhy.com/2015/01/artificial-intelligence-revol...
It's interesting that it shows human level at 2025 and we've just seen a wtf story on HN https://news.ycombinator.com/item?id=49048681