this post was submitted on 29 Sep 2025
96 points (99.0% liked)
Fuck AI
4175 readers
1491 users here now
"We did it, Patrick! We made a technological breakthrough!"
A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
... says every bubble, ever.
Having lived through the dot-com bubble in real time as a professional, allow me to draw the following comparisons:
This couldn't be farther from the truth. Here's a great read, but the long and short of it is that this is entirely based on debt and smoke-and-mirrors.
On the premise that they will be needed, and that the companies ordering them will be solvent when they're completed. These big AI companies have limited runway, and the data centers they need won't be available until after the runway is gone.
Power generation, maybe. That's like all the fiber that MCI laid before they went tits-up in an accounting scandal of their own making, due entirely to creating a thing that nobody needed yet. Yeah, it's useful now (several decades later) but it was speculatively created without a way to pay for it. These ai-specific data centers cannot be repurposed for regular servers, short of gutting the whole thing. Add to this that the chips used for LLMs cannot be repurposed for anything else, and are practically obsolete by the time they're installed.
There are dozens (probably way more) of new "AI" companies being created every day in the hopes of being scooped up by the big players. This is exactly the same as the dot-com boom. Create a cool-sounding company with an idea and a shell of a copycat idea, sell to rube investors, rinse and repeat. This is just as bad, if not worse than it was before.
Which 2 of them are solid? Even if that were the case, one leg will come out from the stool (or 2 or 3) and the whole thing will topple. They're all dependent on each other with fake money and promises, and it's a matter of time. I'm saying sooner rather than later, but who knows.
I'm on mobile, so I'm not able to respond as fully as I would like.
Similarities or not to previous bubbles - it's not ALL built on debt, building on spec, and a hope and a prayer. The MCI fiber bust was very similar to the railroad build out bust, similar to the dot com URL buyup bust. Bubbles have a similar basic structure, that's why we categorize them similarly and they work similarly. Don't get hung up on the minute details.
Not the same - Are you honestly telling me that is it wasn't for LLMs, humanity would have met all it's data storage needs until the end of time? We were always going to need data centers and power generation. The bubble on those fronts is as the runway runs out and prices for both hit a premium because of existing non-AI demand well remain anyway. Making data centers still needed but just not fulfilling AI needs, slowing the industry.
Every idiots' AI "company" and "game studio" is the first bubble to go, but what are these "companies" really funded by? It's all a 1-3 person show paying for Claude $200 a month in real money out credit card debt until it hits marginal success. So there's not that much debt and operating on spec other than peoples' time. When that demand falls off, then daily user stats for AI companies won't waiver enough to see a dent because of all the fools asking GPT how to tie their own shoes. There's no single metric that any of this will impact enough to cause an immediate market sell off freak out.
Data centers and power gen are solid. Enterprise agentic workflow hype garbage is the second bubble. Added productivity and replacing ancient code with modern code is solid for 5-10 years. Data analysis and processing is solid and good for huge contracts. Cottage industry is weak, thats the first bubble.