4 ms·

Is it really a bubble about to burst when literally everyone is talking about AI being in a bubble and maybe bursting soon? To me, we're clearly not peak AI ex

Is it really a bubble about to burst when literally everyone is talking about AI being in a bubble and maybe bursting soon? To me, we're clearly not peak AI exuberance. AI agents are just getting started and getting so freaking good. Just the other day, I used Vercel's v0 to build a small business website for a relative in 10 minutes. It looked fantastic and very mobile friendly. I fed the website to ChatGPT5.1 and asked it to improve the marketing text. I fed those improvements back to v0. Finished in 15 minutes. Would have taken me at least one week in the past to do a design, code it, test it for desktop/mobile, write the copy. The way AI has disrupted software building in 3 short years is astonishing. Yes, code is uniquely great for LLM training due to open source code and documentation but as other industries catch up on LLM training, they will change profoundly too.

zerosizedweasle 1y ago
Let him cook
watwut 1y ago
> Is it really a bubble about to burst when literally everyone is talking about AI being in a bubble and maybe bursting soon? Yes, it is even one of necessary components. Everybody is twitchy afraid of the pop, but immediate returns are too tempting so they keep money in. The bubble pops when something happens and they all start to panicking at the same time. They all need to be sufficiently stressed for that mass run to happen.
- aurareturn 1y ago
  So after the bubble pops, do you think the AI market will still be bigger in November 2025? In other words, do you think we're in 1995 of the dotcom or 2000?
camillomiller 1y ago
This is a very biased example. Also, it is possible only because right now the tools you've used are heavily subsidised by investors' money. A LOT of it. Nobody questions the utility of what you just mentioned, but nodoby stops to ask if this would be viable if you were to pay the actual cost of these models, nor what it means for 99.9% of all the other jobs that AI companies claim can be automated, but in reality are not even close to be displaced by their technology.
- aurareturn 1y ago
  Why is it biased? So what if it's subsidized and companies are in market share grab? Is it going to cost $40 instead of $20 that I paid? Big deal. It still beats the hell out of $2k - $3k that it would have taken before and weeks in waiting time. 100x cheaper, 1000x faster delivery. Further more, v0 and ChatGPT together for sure did much better than the average web designer and copy writer. Lastly, OpenAI has already stated a few times that they are "very profitable" in inference. There was an analysis posted on HN showing that inference for open source models like Deepseek are also profitable on a per token basis.
  - ido 1y ago
    LLMs are particularly good at web development (granted that's a big market), probably due to a lot of the training material being that.
  - qcnguy 1y ago
    We don't know what AI should cost but if you look at the numbers then 2x more expensive is much too low. Think about the pricing. OpenAI fixed everyone's prices to free and/or roughly the cost of a Netflix subscription, which in turn was pinned to the cost of a cable TV subscription (originally). These prices were made up to sound good to his friends, they weren't chosen based on sane business modelling. Then everyone had to follow. So Anthropic launched Claude Code at the same price point, before realizing that was deadly and overnight the price went up by an order of magnitude. From $20 to $200/month, and even that doesn't seem to be enough. If the numbers leaked to Ed Zitron are true then they aren't profitable on inference. But even if that were true, so what? It's a meaningless statement, just another way of saying they're still under-pricing their models. Inferencing and model licensing are their only revenue streams! That has to cover everything including training, staff costs, data licensing, lawsuits, support, office costs etc. Maybe OpenAI can launch an ad network soon. That's their only hope of salvation but it's risky because if they botch it users might just migrate to Grok or Gemini or Claude.
    - aurareturn 1y ago
StopDisinfo910 1y ago
I think you are missing the fundamental point here. The question is not really if AI has some value. That much is obvious and the exemple you give, increasing developer productivity, is a good one. The question is: is the value generated by AI aligned with the market projected value as currently priced in AI companies valuation? That's what's more difficult to assess. The gap between fundamental financial data and valuations is very large. The risk is a brutal reassessment of these prices. That's what people call a bubble bursting and it doesn't mean the underlying technology has no value. The internet bubble burst yet the internet is probably the most significant innovation of the past twenty years.
- Xelbair 1y ago
  Well it all started with usual SV style "growth hacking"(price dumping as a SaaS) of "gather users now, monetize later" by OpenAI - which works only if you attain virtual monopoly(basically dominance over segment of a market, with competition not really competing with you) over segment of the market. The problem is no one attained that position, price expectations are set and it turns out that wishful thinking of reducing costs of running the models by orders of magnitude wasn't fruitful. Is AI useful? of course. Are the real costs of it justified? in most cases no.
- aurareturn 1y ago
  The question is: is the value generated by AI aligned with the market projected value as currently priced in AI companies valuation? That's what's more difficult to assess. I agree it is difficult to assess. Right now, competitive pressure is causing big players to go all in or get left behind. That said, I don't think the bubble is done growing nor do I think it is about to burst. I personally think we are in 1995 of the dotcom bubble equivalent. When it bursts, it will still be much bigger than in November 2025.
- SchemaLoad 1y ago
  There's also a lot of debate over how long a GPU lasts. If a GPU loses most of it's value after 2 years because a newer model is much better/cheaper, that destroys the economics of the companies who have spent billions on now obsolete hardware.
TrackerFF 1y ago
It's not that the AI models or products don't work. It's how much money is being poured into it, how much of the money that is just changing hands between the big players, the revenue, and the valuations.
- aurareturn 1y ago
  Well, do you have a model for this? Or is it just regurgitating mass media that it's a bubble? If hyperscalers keep buying GPUs and Chinese companies keep saying they don't have enough GPUs, especially advanced ones, why should we believe someone that it's a bubble based on "feel"?
  - nixosbestos 1y ago
    > why should we believe someone that it's a bubble based on "feel"? because leaders in the space also keep saying it? And then making financial moves that us pleebs can't even dream of, which back that up? This whole "the media keeps reporting it" as a point against the credibility of something is utterly silly and illogical.
    - aurareturn 1y ago
      The same people who are saying it are also the same people increasing capex?
    - antonvs 1y ago
      > because leaders in the space also keep saying it? They have a lot of reasons for saying that, including to give themselves cover in the event of a crash. What’s happening now is a classic land grab. You’re always going to get inflated prices in that situation, and it’s always going to correct at some point. It’s difficult to predict when, though.
- jennyholzer 1y ago
  to be fair, they don't work
checker659 1y ago
The economics of it is what's the problem, not the power of LLMs.
- aurareturn 1y ago
  So tell us the economics of it? The vast majority of AI doomers in the mass media have never used tools like v0 or Cursor. How would they know that AI is overvalued?
  - checker659 1y ago
    Are the companies making a profit on every prompt I give it (factoring in the cost of upgrading GPUs every year or two)?
    - aurareturn 1y ago
      According to OpenAI, yes. They're "very profitable" on inference. It is the training and capex (driven by competition) that is driving loss.
      - PessimalDecimal 1y ago
        https://www.wheresyoured.at/oai_docs/ https://www.wheresyoured.at/oai_docs/ offers a different point of view on OAI's profitability on inference.
        aurareturn 1y ago
        https://news.ycombinator.com/item?id=45050415 https://news.ycombinator.com/item?id=45050415
      - checker659 1y ago
        But isn't that their "moat"? If the moat is the source of losses, then yes, they're operating at a huge loss.
    - 1y ago

zerosizedweasle 1y ago

Let him cook

watwut 1y ago

> Is it really a bubble about to burst when literally everyone is talking about AI being in a bubble and maybe bursting soon? Yes, it is even one of necessary components. Everybody is twitchy afraid of the pop, but immediate returns are too tempting so they keep money in. The bubble pops when something happens and they all start to panicking at the same time. They all need to be sufficiently stressed for that mass run to happen.

aurareturn 1y ago
So after the bubble pops, do you think the AI market will still be bigger in November 2025? In other words, do you think we're in 1995 of the dotcom or 2000?

camillomiller 1y ago

This is a very biased example. Also, it is possible only because right now the tools you've used are heavily subsidised by investors' money. A LOT of it. Nobody questions the utility of what you just mentioned, but nodoby stops to ask if this would be viable if you were to pay the actual cost of these models, nor what it means for 99.9% of all the other jobs that AI companies claim can be automated, but in reality are not even close to be displaced by their technology.

aurareturn 1y ago
Why is it biased? So what if it's subsidized and companies are in market share grab? Is it going to cost $40 instead of $20 that I paid? Big deal. It still beats the hell out of $2k - $3k that it would have taken before and weeks in waiting time. 100x cheaper, 1000x faster delivery. Further more, v0 and ChatGPT together for sure did much better than the average web designer and copy writer. Lastly, OpenAI has already stated a few times that they are "very profitable" in inference. There was an analysis posted on HN showing that inference for open source models like Deepseek are also profitable on a per token basis.
- ido 1y ago
  LLMs are particularly good at web development (granted that's a big market), probably due to a lot of the training material being that.
- qcnguy 1y ago
  We don't know what AI should cost but if you look at the numbers then 2x more expensive is much too low. Think about the pricing. OpenAI fixed everyone's prices to free and/or roughly the cost of a Netflix subscription, which in turn was pinned to the cost of a cable TV subscription (originally). These prices were made up to sound good to his friends, they weren't chosen based on sane business modelling. Then everyone had to follow. So Anthropic launched Claude Code at the same price point, before realizing that was deadly and overnight the price went up by an order of magnitude. From $20 to $200/month, and even that doesn't seem to be enough. If the numbers leaked to Ed Zitron are true then they aren't profitable on inference. But even if that were true, so what? It's a meaningless statement, just another way of saying they're still under-pricing their models. Inferencing and model licensing are their only revenue streams! That has to cover everything including training, staff costs, data licensing, lawsuits, support, office costs etc. Maybe OpenAI can launch an ad network soon. That's their only hope of salvation but it's risky because if they botch it users might just migrate to Grok or Gemini or Claude.
  - aurareturn 1y ago

StopDisinfo910 1y ago

I think you are missing the fundamental point here. The question is not really if AI has some value. That much is obvious and the exemple you give, increasing developer productivity, is a good one. The question is: is the value generated by AI aligned with the market projected value as currently priced in AI companies valuation? That's what's more difficult to assess. The gap between fundamental financial data and valuations is very large. The risk is a brutal reassessment of these prices. That's what people call a bubble bursting and it doesn't mean the underlying technology has no value. The internet bubble burst yet the internet is probably the most significant innovation of the past twenty years.

Xelbair 1y ago
Well it all started with usual SV style "growth hacking"(price dumping as a SaaS) of "gather users now, monetize later" by OpenAI - which works only if you attain virtual monopoly(basically dominance over segment of a market, with competition not really competing with you) over segment of the market. The problem is no one attained that position, price expectations are set and it turns out that wishful thinking of reducing costs of running the models by orders of magnitude wasn't fruitful. Is AI useful? of course. Are the real costs of it justified? in most cases no.
aurareturn 1y ago
The question is: is the value generated by AI aligned with the market projected value as currently priced in AI companies valuation? That's what's more difficult to assess. I agree it is difficult to assess. Right now, competitive pressure is causing big players to go all in or get left behind. That said, I don't think the bubble is done growing nor do I think it is about to burst. I personally think we are in 1995 of the dotcom bubble equivalent. When it bursts, it will still be much bigger than in November 2025.
SchemaLoad 1y ago
There's also a lot of debate over how long a GPU lasts. If a GPU loses most of it's value after 2 years because a newer model is much better/cheaper, that destroys the economics of the companies who have spent billions on now obsolete hardware.

TrackerFF 1y ago

It's not that the AI models or products don't work. It's how much money is being poured into it, how much of the money that is just changing hands between the big players, the revenue, and the valuations.

aurareturn 1y ago
Well, do you have a model for this? Or is it just regurgitating mass media that it's a bubble? If hyperscalers keep buying GPUs and Chinese companies keep saying they don't have enough GPUs, especially advanced ones, why should we believe someone that it's a bubble based on "feel"?
- nixosbestos 1y ago
  > why should we believe someone that it's a bubble based on "feel"? because leaders in the space also keep saying it? And then making financial moves that us pleebs can't even dream of, which back that up? This whole "the media keeps reporting it" as a point against the credibility of something is utterly silly and illogical.
  - aurareturn 1y ago
    The same people who are saying it are also the same people increasing capex?
  - antonvs 1y ago
    > because leaders in the space also keep saying it? They have a lot of reasons for saying that, including to give themselves cover in the event of a crash. What’s happening now is a classic land grab. You’re always going to get inflated prices in that situation, and it’s always going to correct at some point. It’s difficult to predict when, though.
jennyholzer 1y ago
to be fair, they don't work

checker659 1y ago

The economics of it is what's the problem, not the power of LLMs.

aurareturn 1y ago
So tell us the economics of it? The vast majority of AI doomers in the mass media have never used tools like v0 or Cursor. How would they know that AI is overvalued?
- checker659 1y ago
  Are the companies making a profit on every prompt I give it (factoring in the cost of upgrading GPUs every year or two)?
  - aurareturn 1y ago
    According to OpenAI, yes. They're "very profitable" on inference. It is the training and capex (driven by competition) that is driving loss.
    - PessimalDecimal 1y ago
      https://www.wheresyoured.at/oai_docs/ https://www.wheresyoured.at/oai_docs/ offers a different point of view on OAI's profitability on inference.
      - aurareturn 1y ago
        https://news.ycombinator.com/item?id=45050415 https://news.ycombinator.com/item?id=45050415
    - checker659 1y ago
      But isn't that their "moat"? If the moat is the source of losses, then yes, they're operating at a huge loss.
  - 1y ago