Hi, David Brancaccio here. This week on Marketplace Morning Report, I was asked to share a personal

story. My wife and I worked on a plan to move back to California with two of our offspring living in the

Golden State. In November, we made that dream a reality when we purchased a cozy bungalow in the city of Altadena.

Just a few weeks later, our home, along with so many thousands of others, is in ruins, total loss.

This week, we return to survey what little the wildfire left and to start planning what might be next for us.

I'm learning a lot, and some of it may be useful to others rebuilding after disaster,

from cleaning up the mess to navigating insurance, mortgages, taxes, to securing contractors,

and more. Listen to Marketplace Morning Report wherever you get your podcasts.

And then there was DeepSeek.

From American Public Media, this is Marketplace Tech.

I'm Megan McCarty Carrino. Everyone was obsessed with the new white whale of the AI world this week.

We'll get into it on today's Marketplace Techbytes week in review.

Plus, Trump floats tariffs on semiconductors from overseas.

And a bipartisan Senate bill to ban kids from social media

is getting another look. But first, back to that DeepSeek drama. The Chinese AI company

took the world and the markets by storm with claims that its class-leading large language model

was built at a fraction of the cost of Silicon Valley rivals.

DeepSeek claims it spent only $6 million on compute power, at least 16 times less than leading U.S. companies.

Here to discuss all that is our regular contributor, Peresh Daveh, senior writer at Wired.

Part of it was that $6 million number. It seemed like maybe these AI developers, whether it be a meta or an open AI or a Google, could get away with spending a lot less money, buying a lot less chips, and still coming up with these powerful models.

But I think as people have digested it,

there's a lot of caveats in the number that $6 million doesn't really cover everything.

...