Howard Marks

But then there are

bubbles which are based on technological progress, starting with the steam engine, the railroad, the radio, the automobile, computers, internet, et cetera.

And these actually push society ahead and change it irreversibly.

But in the process, there is a bubble surrounding their implementation which is overly accelerated and overly financed and goes to excess.

and end up destroying a lot of capital, but leave society greatly changed.

And I'm sure that AI is in the latter category in terms of effect on society.

And the question is, will the implementation prove to have been excessive in scope and in the way it's financed?

No, but it...

You know, when investors hate everything and won't touch it with the 10-foot pole, chances are it's going to be on sale because nobody has pushed up the price.

In fact, their disinterest has pushed down the price.

But when everybody likes something and is excited about something, chances are it may be overhyped and overpriced.

So you just have to be careful.

What does that look like?

What I say in the memo is that it's okay to lend for activities, even if they're uncertain, but not if the outcomes are purely conjectural.

I mean, in order to be a smart lender, you have to have good visibility on the extent to which the thing is likely to repay interest in principle.

If it's just purely conjectural, you shouldn't be a lender.

And in fact, I think the memo says that...

where that's the case, you should actually, if you wanna participate, you should be in the equity so at least you get the upside.

The lender has no upside.

You make a 9% loan, all you're gonna get is 9% no matter how well the thing does.