I won't disclose my source for privacy reasons. However, you're right. This information was available immediately after ChatGPT went public. It was known even before in initiates circles. At first we could build on the the promise of inference getting cheaper with scale; except the exact opposite happened. The US market still kept on going and is now screwed beyond historical limits, which is why they won't turn back. The burst is going to be ugly, with or without Greenland to cover for it.
What's interesting with this scam is how much it uses a similar dynamic as the blockchain "get rich quick" era. Now people make posts about "get a good paying job [...without effort]" or "build a business [...without idea]". Same difference. I'm concerned how many educated people fall for it though.
It's going to be a VERY ugly bubble burst. All the power companies that have built generation plants to run the machines are going to have problems when there's no one to buy the power. Power companies going broke isn't good for the economy.
It is going to be ugly. I'm not sure power investments will suffer the most though. At least, you can use the extra production for other parts of the economy. I'd think data centers will be the worse. The hardware is totally unfit for anything else than neural networks and still, it is used for raising debt. As soon the farce is over, the value of these investments will fall to zero. They're already way over-evaluated.
The other part of the economy that will use the power will have to be built. It doesn't exist now. One factory won't use as much power as a data center. They'll have to build 20.
Someone wanting to build a factory in the area where the data center is located has to find people to work there. The factories have to be built. Housing for the people needed to staff them does, too. Roads need to be built to move the goods. It's not as simple as you make it out to be.
I agree that the data center people are more screwed. They'll be worth nothing forever.
Someone wanting to build a factory in the area where the data center is located has to find people to work there. The factories have to be built. Housing for the people needed to staff them does, too. Roads need to be built to move the goods. It's not as simple as you make it out to be.
I agree that the data center people are more screwed. They'll be worth nothing forever.
Good points. I guess I'm oversimplifying here. Anyway. What a circus show.
The more you think about it, the worse it gets. What happens to the banks that have loaned them the money to build this stuff?
Global spending on artificial intelligence (AI), including infrastructure, software, chips, and services, is expected to total US$1.5 trillion in 2025. Most of that is in the US. That doesn't include power stations.
The chip making companies have built extra production factories. They cost a huge amount.
It just goes on.
Global spending on artificial intelligence (AI), including infrastructure, software, chips, and services, is expected to total US$1.5 trillion in 2025. Most of that is in the US. That doesn't include power stations.
The chip making companies have built extra production factories. They cost a huge amount.
It just goes on.
The implications are huge.
If I recall correctly, as it is right now, the bubble is mainly driven by a mix of venture capital and CAPEX issued by an handful of companies (Microsoft, Meta, Alphabet, Oracle, X). It's expected that capital expenditures will slow down at some point (as I mentioned above, Microsoft is already jumping ship; other, like Meta, could keep going a while using their war chest, as they did with the Metaverse). As for VC, it will hit a hard ceiling sometimes during 2026, 2027 at the later. There's simply no more money available from private investors around the world (and yes, I know I sound mad, but it's a fact).
The banks will surely absorb some of the loss through shady, subprime-type financial products but they won't be able to do it alone, or they'll sink. I can't wait for the reaction of the American people when they'll have to abandon their 401k to save "the economy" (aka: Sam Altman).
If I recall correctly, as it is right now, the bubble is mainly driven by a mix of venture capital and CAPEX issued by an handful of companies (Microsoft, Meta, Alphabet, Oracle, X). It's expected that capital expenditures will slow down at some point (as I mentioned above, Microsoft is already jumping ship; other, like Meta, could keep going a while using their war chest, as they did with the Metaverse). As for VC, it will hit a hard ceiling sometimes during 2026, 2027 at the later. There's simply no more money available from private investors around the world (and yes, I know I sound mad, but it's a fact).
The banks will surely absorb some of the loss through shady, subprime-type financial products but they won't be able to do it alone, or they'll sink. I can't wait for the reaction of the American people when they'll have to abandon their 401k to save "the economy" (aka: Sam Altman).
I believe it's actually a bubble, same as dotcom bubble. Is it profitable right now? Sure, you can make an AI service and make some money from it. Will it still be profitable in the future? Very unlikely.
I also have seen the results of the AI in some of the freelance projects, that were made by other people who relied heavily on AI, and on people who want to get into the field now. The projects look worse than what I've seen pre-AI from people who could barely code, because of its unnecessary complexity and inconsistency. New developers don't try to even think critically about the output of the LLM.
I also have seen the results of the AI in some of the freelance projects, that were made by other people who relied heavily on AI, and on people who want to get into the field now. The projects look worse than what I've seen pre-AI from people who could barely code, because of its unnecessary complexity and inconsistency. New developers don't try to even think critically about the output of the LLM.
As far as I know, none of the people building the data centers are making money out of it. The only really successful projects I've heard about have been call centers.
It's interesting that you mentioned data centers, since people who let you use the power and people who build it make quite a lot of money nowdays. It's also the reason why there's a raise of devops/sysadmin positions.
I was talking more about these saas projects that use some form of third party API to make their "future AI-powered app". Some of them actually make a good bank from it.
As for the existing projects, I guess that it's more about the investors who believe in AI, and that's why we see it everywhere. If you're a public company, you can't really ignore their opinion is, so you'll add some kind of AI to your product, even if most users will tell you it's stupid.
I was talking more about these saas projects that use some form of third party API to make their "future AI-powered app". Some of them actually make a good bank from it.
As for the existing projects, I guess that it's more about the investors who believe in AI, and that's why we see it everywhere. If you're a public company, you can't really ignore their opinion is, so you'll add some kind of AI to your product, even if most users will tell you it's stupid.
Members-only continuation
This discussion contains more posts.
Create an account or sign in to continue reading the full conversation. 11 additional posts await inside.