Yo, technology.

What is it all about?

What we've really solved is human language.

That's huge.

For 50 years or 60 years of computer science, we couldn't even get close.

And now we can make things that are actually fluent in human language.

It's amazing.

And so what we can do with that is all kinds of new business cases where we can make things more efficient.

We can we can do more with fewer people with less resource.

And I think that's why the economics start to matter.

Hello and welcome to Danny in the Valley, your weekly dispatch from behind the scenes and inside the minds of the top people in tech.

This week, you're in for a treat.

On the program, we have Navine Rao. He's a serial entrepreneur who started working on AI way before it was cool. And you may have heard of him recently because in June, he sold his

two-year-old startup, Mosaic ML, to Databricks for $1.3 billion that is with a B, which is not bad going. Mosaic, which was developing

low-cost ways to build and train AI models, only had 55 employees. So quite the result. In any event,

I wanted to have Navine on because one of the really interesting aspects of the market right now

is that just how much it

costs to train these AI models. It means that in most cases only the very biggest or very

best funded companies can build these really big super capable models. And this, of course, is a big

obstacle for smaller developers and kind of the development of AI more broadly.

But what Mosaic showed is that you don't need tens or hundreds of millions of dollars.

You can really build powerful AIs for just a few hundred thousand.

...