Frank Lee

It's not as clean-cut as it was before.

Yeah, so again, I mean, I think the spending for hyperscalers have continued to increase more and more than anybody's expected.

So we saw in January, you know, another big revision for most of the hyperscalers.

But you're getting to a point now where hyperscalers are having to fund some of this externally.

So you see some of these hyperscalers are coming to the market to fund.

And this is the first time we're really seeing it.

Previously, the first couple of years of AI, it was self-funded, right?

Right.

So I think there's more scrutiny now in terms of what people are paying for, what hyperscalers are investing in and perhaps they were earlier.

I'm still an AI bull.

I think the difference is that because of some of these uncertainties persisting right now, companies need to deliver earnings, right?

In the first two years of this AI trade, it didn't really matter if the earnings were coming through or they were generating.

NVIDIA is not in that camp.

They have delivered on earnings.

But I think the question is the multiple of how much people are willing to pay for AI names individually.

is definitely under more scrutiny than they were previously.

So I think that's one of the challenges.

In general, what you're seeing is this diffusion of AI moving into hardware names.

Clearly, memory has been one of them.

Networking, which NVIDIA also mentioned on this call, is another area where we see very clear content growth and perhaps higher growth than the traditional compute, which is where we saw their initial beneficiary of AI.