Lisa Mateo

So I think that's what software providers are going to have to do.

They're going to have to service and show that they add value to the underlying business.

There is a link here between that and Nvidia's earnings.

Your main takeaway from the call last night is the same that we've been talking about all morning.

Compute equals inference, which equals revenues.

And the way that Jensen Wong explained that was profitable tokens.

In other words, customers of all kinds are willing to pay for the output of an AI model and that actually the economics have improved.

The thing is, I don't see any evidence of that in any of the software companies' earnings yet.

Do you?

Not yet.

I don't think so.

Part of that is capacity, though, Ed.

I mean, you heard Microsoft say they took some of their own capacity, but you heard them say, and Amazon too, we can fill whatever we get immediately in terms of demand, but they don't have enough capacity.

So I guess I would look at it two ways.

One is if you go back and look at CapEx in the 90s.

It was about a two standard deviation above the mean spend in tech CapEx.

We're not even one standard deviation away from the mean.

And one man's CapEx spend is another man's revenue.

So AMD, NVIDIA, Broadcom, names we all own, will continue to benefit.

And here's the last thing I'll say.