Since some times now the AI bubble is growing and its consequences with it, flash storage price increase, electricity went wild in some places, GPU… Don’t need to say anything sadly… NVIDIA became the most valuable company exceeding 4T
So when all of this will go crazy and grow to the burst? When does prices will go down and speculators rushing out of it?
Open question feel free to explain the wider you can, I’m not a financial so I’m really interested in some analysis of the situation :)
My prediction is: Not anytime soon. Seems we’re getting lots of coverage in the mainstream now, and people start to question if their financial investments are safe or if we’re gonna re-live the real estate bubble from a few years ago.
But there’s still a datacenter popping up every two weeks, an absurd amount of money being invested in these companies and everything around it up to small nuclear reactors. The people with lots of money are still firmly seated in the hype train. And if past investment bubbles taught us anything: It takes way longer than we’d expect until they finally burst.
So I’d say it could pop any day, but I think this is unlikely to happen within the next 2 years or so, maybe more. Though you might want to think about your retirement fund now and make sure there’s some money in other assets.
Would you think that an announcement about an impossibility of pushing transformers further would results in a burst of the whole bubble?
Hehe. We’ve already had several "AI winter"s. For me it’s pretty much settled by now that transformers and maybe even large language models itself won’t scale well past what we have today. And I don’t think they’ll lead to super-intelligence, which used to be the sales pitch. And I think several high profile scientists hold that opinion. I’m really split on this. These bubbles are volatile. All it takes is enough head fond managers to pull out and maybe a large company like Microsoft. It’s somewhat rare companies at that size just fail, but I guess it’s possible?! I mean I’ve also predicted Bitcoin to fail several times now because it’s not as usable as promised, yet that always climbs to a new record high. So I’m a bit more careful with my predictions these days. There’s still lots of things we can shoehorn AI in to. We’ll see it in robotics and that’s barely started. And ultimately this bubble isn’t fueled by realistic expectations in the first place, so I’m not sure if any factual statement itself can change this. It’ll grow as long as it grows, as long as Sam Altman is able to deliver promises and as long as there’s enough money pumped in. Which might change once there isn’t enough money available anymore or if it actually stagnates for long enough for investors to want out. I just don’t think they’ll listen to reason.
I remember flying into Las Vegas when the real estate bubble popped in 2009. You could see scraped land 30 minutes out, then pavement and services going in, then partial frames and then closed envelopes. All abandoned.
Stupid money flowed in until the very last minute. The stupidest money kept going in thinking they’d buy the dip and catch that falling knife.
People will add money even after it pops. I think it was Buffet who said that the only way to really make money is to sell too soon.
Speaking of those data centers, what do you suppose we’ll do with all that compute post-bubble?
Good question. But technology always finds ways to eat up resources. I mean we’ve come from computers with few megabytes of memory and slow internet connections to several Gigabytes of memory and Gigabit internet connections, smartphones with double the processing power every few years and they roughly do the similar things in a lot of aspects. Call, text, office work, surfing… We had those very same things in the early 2000s with a fraction of the resources… Oh well, and we’ve gained some awesome new things like smartphone cameras or video streaming. I guess we’ll come up with use-cases. Question is just who is gonna afford to pay for operation and electricity. Compute is going to become way cheaper once the taxpayer and people ate the investment losses and the hardware is paid for. But operation costs a pretty penny so I think I still don’t have any idea about profitable use-cases given the scale of it all…
I think it’ll happen within 2 years and I think the housing market will collapse around the same time.
Not sure if I WANT to be right but that’s my prediction. Two major bubbles popping back to back.
So you think a kind of snowball effect would affect the both market?