Tyler Crowe

And so...

Similar to that, the anthropic deal that they signed was something like $200 billion worth of compute power over the next five years.

So there is a return.

I almost feel like we might be stepping on some stuff we're going to talk about later on this, but it is kind of this idea of like what we're doing at the current pace doesn't necessarily like work.

There's some incongruencies here.

And to that point, it's like, yes, we could see compute power costs go way down for reasons X, Y, Z. But there's also like the possibility of algorithm efficiencies.

Right now, all of these companies are incented to put out the best product and kind of compute power be damned.

Eventually, it's going to come down to a cost per token efficiency from the actual computing that you're doing with these algorithms that will incent these companies to be within the range or try to bring down their costs, commensurate with what they're going to see with compute power.

Yeah, to me, this is a little like this isn't just the tide that raises all boats.

This is the tsunami that is bringing everyone 30 feet up higher than they thought they were going to go.

It's not a tsunami.

Yeah, just think about, again, we're talking about Alphabet spending $180 to $200 billion this year.

If we're going to take the over or under of how many revisions up there we're going to see on their CapEx, it's going up.

2027, it's going up.

They're raising more money to go higher.

All of them are going higher.

Meta's doing weird off-balance sheet stuff so they can go higher.

It almost is to the point where it's not like we need to get the best equipment to make this happen.

It's like, grab every spare part you can and slap it together and get some sort of compute power available.

I'm sure that a lot of these...