Item — HN Reader

I mean, sure we're in a bubble, but the trick is to call it, with that old Keynes quote about the market staying irrational longer than you can stay liquid.

But also: > "So, you're saying a third of the stock market is tied up in seven AI companies that have no way to become profitable and that this is a bubble that's going to burst and take the whole economy with it?"

> I said, "Yes, that's right."

that is something different in this case isn't it? those seven companies making up a third of the market do not need to become profitable, they are insanely profitable. Mostly they invest a lot in AI but if that doesn't pay out, all but NVidia have their day job to go back to.

3 comments

> I mean, sure we're in a bubble, but the trick is to call it, with that old Keynes quote about the market staying irrational longer than you can stay liquid.

It might be worth it just to call it now. All you really have to do is get out of the S&P, you don't have to get out of everything.

It can run for another year or two. Long time to sit on the sidelines with your cash inflating away.

This is the crux of bubbles - timing and where do you move assets so they have protection.

The things I'm trying to keep in mind, because hard to suppress the instinct to be reactive:

1. "Time in market beats timing the market."

2. When diversifying, your profession is is already part of your portfolio.

There's also the political mismanagement of the United States, but that's a whole 'nother can of worms.

Is it that impractical that the GPUs would find another use case? Of NVidia cut enterprise gpu prices by 10x (which they absolutely could!) the gpu+memory combo would be cheaper than all other compute for database-like operations, simulations, and whatever ai is left.

Is gpu good for that? It seems like it is a very niche capability that can't easily be re-purposed the way regular cpu compute could be.

I did a lot of projects with Kafka (big data) in cloud environments and companies had big, pie in the sky dreams that came crashing back to reality when they got the bill for compute services. It's happened several times on projects I was on.

Older GPUs had nowhere near enough memory to be really interesting for big data crunching.

Now however, the biggest limit for AI workloads is GPU memory capacity (and bandwidth). The billions invested are going into improving this aspect faster than any other. Expect GPUs with a terabyte of ultra-fast memory by the end of the decade. There are lots… and lots… of applications for something like that, other than just LLMs!

What’s OpenAI’s and anthropic’s day job?

Which of the seven OP referred to are OpenAI or Anthropic?