Tim Beyers

In the third quarter of fiscal 2026, that's the last quarter they reported, $51.2 billion in data center revenue, which was up 66% year-over-year.

They also have

control of, they might be the only one that does full-stack system design for AI because they have the CUDA software for programming GPUs and CPUs for these AI systems, and they have visibility.

I'm not sure if I should believe this or not, but Jason, you tell me, they have visibility into what they say is

$500 billion, so half a trillion dollars of Blackwell and Vera Rubin revenue.

These are the two most recent chipsets coming out of NVIDIA right now.

Then longer term, $3 trillion to $4 trillion in annual AI infrastructure build by the end of the decade.

Again,

Jason, these are just outrageous numbers.

You tell me, which one is the more important participant in the AI value chain?

Is it Taiwan Semi or is it NVIDIA?

I'll tell you one that's already fighting NVIDIA to get some TSMC time and attention.

That would be Apple.

Apple is getting crowded out by NVIDIA right now inside TSMC factories.

So that's interesting.

Okay.

So you both agree that TSMC is at the head of the AI value chain here.

Last question on this, and then we're going to move to our second segment.

Because, Jason, you mentioned something interesting, and I had forgotten about this, even though it wasn't that long ago, the way DeepSeat got around CUDA.

So, again, CUDA is the programming language.