Jeff Bezos says AI is in a bubble but society will get 'gigantic' benefits

Post by wmf » Fri, Oct 03, 2025, 9:55 PM UTC

ten_hands wrote: post #24

lumost wrote: post #9

Earlier quoted context omitted.

OpenAI has a reasonable path to reduce their costs by 10-100x over the next 5 years if they stop improving the models. That would make them an extremely profitable company with their only real risk being “local ai”. However customers have wanted their data in the cloud for years, local inference would likely just mean 0 cost tokens for OpenAI. The challenge is the rest of the industry funding dead companies with bill…

I don't see how this works though. OpenAI doesn't exist in a vacuum, it has competitors, and the first company to stop improving their model will get obliterated by the others. It seems like they are doomed to keep retraining right up until the VC funding runs out, at which point they go bankrupt. Some other company, that doesn't have a giant pile of debt will then pick up the pieces and make some money though. Once…

Post by bilsbie » Fri, Oct 03, 2025, 10:00 PM UTC

zerosizedweasle wrote: post #36

blast wrote: post #29

He's (unsurprisingly) making an analogy to the dotcom bubble, which seems to me correct. There was a bubble, many non-viable companies got funded and died, and nevertheless the internet did eventually change everything.

The biggest problem is the infrastructure left behind from the Dotcom boom that laid the path for the current world (the high speed fiber) doesn't translate to computer chips. Are you still using intel chips from 1998? And the chips are such a huge cost, and being backed by debt but they depreciate in value exponentially. It's not the same because so much of the current debt fueled spending is on an asset that has ve…

Post by munk-a » Fri, Oct 03, 2025, 10:01 PM UTC

rchaud wrote: post #39

blast wrote: post #29

He's (unsurprisingly) making an analogy to the dotcom bubble, which seems to me correct. There was a bubble, many non-viable companies got funded and died, and nevertheless the internet did eventually change everything.

Dotcom mania companies were not Internet providers. They tried making money on the internet, something people already saw as worth paying for.

Post by wmf » Fri, Oct 03, 2025, 10:03 PM UTC

somewhereoutth wrote: post #43

My understanding is that the cost of training each next model is very very large, and a half trained model is worthless. Thus when it is realised that this investment cannot produce the necessary returns, there will simply be no next model. People will continue using the old models, but they will become more and more out of date, and less and less useful, until they are not much more than historical artifacts. My poi…

Post by jorts » Fri, Oct 03, 2025, 10:05 PM UTC

yeasku wrote: post #62

hx8 wrote: post #48

Earlier quoted context omitted.

A lot of the infrastructure made during the Dotcom boom was shortly discarded. How many dial-up modems were sold in the 90s? The current AI bubble is leading to trained models that won't be feasible to retrain for a decade or longer after the bubble bursts.

Honestly, not that many people had modems.