David George

Inference cost is going to go down over time.

I mean, there's so many existential questions about market structure that will predict inference cost, but the history of technology would suggest that it's going to go down over time.

The cost of inference has gone down at the same time that reasoning happened.

And so token token usage has gone way up.

So you haven't yet seen any improvement in gross margins.

But I think over time, that's likely to happen.

There's a big difference between having 30% gross margins and 70% gross margins.

So we do care.

Our expectation is if you're producing a lot of customer value, and if the models get a lot better over time, you're going to increasingly produce customer value, that the cost is going to go down.

There's not going to be so much market power, the model providers, that it's going to settle out where these businesses are probably higher margin businesses.

I think there'll be lower margin businesses than SaaS businesses.

Maybe they end up as 50% margin companies as opposed to 80%.

But the size of the impact and the usage and the amount that they'll be able to capture to our point on business model earlier is probably so high that it's fine.

There's so many ingredients to the best of.

There's sort of a foundational point, which is every great company either has unique product or unique distribution.

The best companies in the world have both.

The best companies in the world have such unique product that it leads to unique distribution.

But if you don't have either of those.

What's your favorite example of that?

I'll use a recent one that the product is so good that people just naturally gravitated to it as cursor.