What would happen if tomorrow the AI bubble would burst?

🗣️ Off-Topic Lounge
1

Lets do some brainstorming about what could happen if the AI bubble actually bursts in the future.

Not that AI disappears or becomes useless, but more like the money side of it gets completely out of hand before the actual revenue catches up.

So the premise is that billions are going into GPUs, data centers and AI infrastructure. A lot of it is financed with debt. If one important company gets into trouble, GPU values drop, lenders get nervous, then things can start hitting Nvidia, cloud providers, AI startups, jobs, the stock market, etc.

What I found interesting is that even in this scenario, AI itself still wins long term.

Kind of like the dot-com bubble. The crash didn’t kill the internet. It killed a lot of bad companies, crazy valuations and unrealistic expectations.

So I was thinking… what would an AI crash actually mean for people like us?

Would AI APIs suddenly become much cheaper because there is too much compute? Or much more expensive?

Would half of the “AI startups” disappear?

Would smaller developers actually benefit because the insane funding advantage goes away?

Or do you think this whole AI bubble idea is exaggerated and the revenue will eventually catch up with all the spending?

Check this site for details: https://isaiprofitable.com/

1 Like
2

Interesting question…

If we look at companies that survived the dot-com bubble/crash, e.g. Amazon or eBay I don’t think the price of their products really changed, like I don’t think Amazon shipping costs suddenly went 5X up in price?

So maybe the bubble and crash is in the valuations, not the price of the product itself.

I’d expect AI prices to go up, since they’re currently subsidized, but how much can that be? What’s the “real” cost?

There are a lot of big players so I think it should stay quite competitive, i.e. prices should trend towards Cost of Compute + some small margin?

3

There is no bubble. Let’s be serious: developers have adopted AI like it’s been here forever, and everybody wants it now not just devs. Considering how fast AI is advancing, it will only generate more and more income from all directions.

My biggest fear, seeing how it has evolved, is that it will wreak havoc across all industries. It will create the biggest crisis humanity has ever had to solve, and that’s a fact. Just look at what it’s already done to IT, it will come for everything, literally any type of job is in danger, even handy laber is a matter of time until robots will take over at this point.

If there will be no regulations on this it will cause a disaster that will be really hard to fix, not to mention the terminatior and matrix scenario that do not look so SF anymore?

4

I agree that AI is in demand and more people start to use it, however, I don’t point out the bubble in the demand part, but in the infrastructure part.

The entire AI industry is based on the premise that new datacenters will continue to open up, chips will continue to be added over time and processing power continues to grow.

Also, the weakness here is that some companies purchase the chips with borrowed money, and they guarantee their loans with the same chips which they purchased. If the chips loose value (for any reason), these companies will be in trouble, as the lending companies will come after them. I know, nowadays, chips to loose value is a bit unlikely, but it is still on the table (life can be surprising in many ways).

The other scenario in which there can be problems for companies is if the chip supply would be artificially limited. This scenario can become reality if China decides that they want Taiwan for good and chip production is cut off from Taiwan for the rest of the world. This is another extreme scenario, but things are brewing in this direction for a while now.

In this case, chips will become scarce and expensive, no new datacenters can be opened up. In this case, AI demand could continue increasing while the infrastructure behind it runs into big problems at the same time.

That wouldn’t necessarily kill AI. In this case, AI would become more expensive. Smaller AI infrastructure companies could disappear or get acquired. Compute prices could increase a lot.

So, what I thought of is not the end of AI, but I meant that the industry might be overbuilding and overfinancing the infrastructure around AI, based on assumptions that everything will continue moving in the same direction: cheaper compute, more chips, more datacenters and unlimited access to capital. Which is not how usually things work, on the long run.

5

Yes, these are issues, but the latest advances are impressive as sh*t, and not just in coding. You should see what it can do with 3D in Blender. I was blown away! It recreated my logo from an SVG with incredible precision in just a few minutes. That would easily have cost me $1,000 and taken a few days to get that level of quality and boy that wouled have been such a mess since is really hard to find proffesionals this days, everybody wants easy money.

Yes, the chips are an issue. They consume too much power and only last 4–6 years at most. Unless they invent something else, transistor-based chips will always have this problem. But more and more people are adopting AI and paying for Pro plans. For example, my sister is an accountant and pays for OpenAI’s $200 Pro plan ( and this is monthly plan which is a lot). It does most of her work, and she’s never had so much free time. She’s in heaven, and that’s just one example.

Personally, I don’t see this bubble popping. I used to think it would, but seeing how good AI is now, I’m no longer convinced. We’ll see. :slightly_smiling_face:

What worries me is the possibility of a Matrix or Terminator scenario, or something even worse, given the lack of regulation around this, thos at the top who are sick in the brain and all they see are trillions of $$$ they shuld know that the terminatior will get them as well LOL

1 Like