Jim Chanos

We don't need all this fiber optic gear.

And order books just collapsed.

And S&P earnings dropped 40% from peak to trough.

The S&P dropped about 40%.

There was a mild recession.

GDP dropped about 1% for two quarters.

It really was not like the global financial crisis or even 89-90 when we had a banking crisis.

But it was terrible for corporate profits because there was so much basically steroidal profit participation, profit margins from all this.

So that's what we're kind of keeping an eye on is order books and looking for any signs that suddenly maybe we don't need

all of these router GPUs or all of these data centers or what have you.

Well, again, we want to see the underlying profitability of the engine, right?

Will AI come up with a profitable business model?

Because right now, if you think about it, open AI is, I think, roughly two-thirds of all queries in AI are open AI.

And they're going to do $13 billion of revenues this year, $30 billion next year.

with capital spending needs in the hundreds of billions of dollars.

So at some point, someone's got to come up to say, we actually have a model that will directly lead to cash flow and profits so we can service the debt that we're taking on to build the data centers and build the infrastructure.

And we'll have to see.

I mean, what I've said is I'd rather be long-term

the companies that are producing the magic from the chips than the landlords of where the chips reside, right?

So if this is going to truly be revolutionary, and it probably will be, I think you're going to have to see stuff coming out of the hyperscalers and open AI and really interesting applications that can be monetized.