Bell Lin (2:56)

And how have some consumers adjusted to living with both gas powered cars and EVs?

Patrick George (3:01)

We had an interesting interview with Jose Munoz, who is the CEO of Hyundai Motor. And he was saying out in California especially, you know, a lot of folks are looking at making an EV the second part of their two car household. And the EV is the commuter car, just driving to the office, sitting an hour in traffic or what have you. And then on the weekends, if you need to go to the mountains or go on a longer road trip, go see family. That's where the gas car or the hybrid comes in. So a lot of folks in America are kind of seeing the value of using both.

Bell Lin (3:27)

How are some carmakers responding to consumers changing preferences?

Patrick George (3:31)

The fair answer is to say not at all. There's a few automakers that have kept up their EV investments like Hyundai, and they're reaping the benefits of this. Now, Toyota has been really late to the game in electric vehicles, but it's brought several more to market just this year, and they're seeing some strong sales around those. But largely the industry dialed back its EV investments in America to focus on more profitable gas cars. And that's a trend that we're not gonna see reversed really anytime soon. It would take months and maybe a year plus of prolonged gas spikes before we're going to see decisions made on products here.

Bell Lin (4:06)

So it's safe to say that some carmakers are not ready for an EV boom.

Patrick George (4:10)

Most of the automakers we've interviewed and talked to say they're riding this out and seeing where it goes. It's a very costly endeavor to shift production and to shift your product plans in the future. And it's tough to do that when you have these potentially flash in the pan moments like the conflict in Iran,

Bell Lin (4:26)

hopefully is that was WSJ Detroit Bureau Chief Patrick George. Do gas prices have you rethinking EVs for yourself? If you're a listener on Spotify, be sure to leave us a comment with your thoughts. Coming up, demand for agentic AI is exploding and companies are scrambling to secure the computing capacity needed to serve their rapidly expanding customer base. We discuss how that's impacting the AI boom. After the break.

Alex Osola (4:59)

Hi, this is Alex Osola, host of the WSJ's what's News podcast. We bring you the biggest news of the day, from business and finance to global and political developments that move markets. If you're looking for more insights and tools to understand the latest headlines, consider becoming a subscriber to the Wall street journal. Visit subscribe.WSJ.com whatsnews to subscribe now.

Bell Lin (5:27)

If you've ever used an AI chatbot like ChatGPT or an AI agent system like Claude Code, you may be part of a population of AI users who are contributing to a shortfall in computing firepower. And that shortfall is happening fast, with consequences for AI companies and people who use a lot of AI. Companies like OpenAI and Anthropic have been racing to get the computing power needed to serve a base of customers who are significantly increasing their AI usage. And at the same time, prices for compute are going up. WSJ reporter Robbie Whelan is with us now to talk about this computing crunch and how it's affecting both AI companies and power users. Robbie, why are prices for compute going up?

Robbie Whelan (6:13)

The prices for the chips that produce this computing power have really risen dramatically in the last six months or so. We have reports from what are called the Neo clouds and the cloud service providers. These are the companies that basically run the data centers that are filled with these powerful AI computer chips. Those companies, those middlemen, are raising prices really sharply on customers. So the data center companies are having to pay more for the chips and then having to charge more by the hour to rent the power that those chips produce out to end users, to people who are designing AI tools and using them and querying them and using them for business as well. There's two reasons why this is happening. One is just a simple supply and demand equation, right? When demand goes through the roof, you have to raise prices. And the other one is just that some of these companies, for example Anthropic and OpenAI, just simply don't have enough compute to serve everyone who wants to use their products. And so they've got to sort of limit it, dole it out in a way that might feel unfair to some users, but basically they have to prioritize certain users ahead of others. And what that generally means in practice is asking them to pay more.

Bell Lin (7:21)

And what are some of the consequences for real world people for everyday developers of this compute shortage?

Robbie Whelan (7:29)

We've talked to people who have had to shift away from their favorite AI tools. For example, we spoke to a Gentleman who is a founder of a company that uses coding all the time. He loves using Anthropic's Opus 4.6 tool to create agents that he needs for his business. But because there have been so many stoppages, outages on Anthropic because of the lack of computing power that we're seeing in the market right now, he's had to shift to other models. He's using OpenAI a lot more than he used to because his preferred tool from Anthropic is really suffering from this supply crunch that we're seeing. So that's one type of consequence. Another consequence is just that maybe the revolution that AI is supposedly going to cause in the business world especially is going to proceed a little bit more slowly than we initially thought. Because obviously if you can't find the resources and build out the infrastructure needed to totally transform so many different parts of industry around the world, then you're going to have to maybe just take it a little more slowly and we might see things slow down a little bit until we can allow the infrastructure piece of it to catch up to the demand piece.

Bell Lin (8:38)

Speaking of taking things more slowly and OpenAI and Anthropic, have they changed maybe the cadence at which they release products or held back on certain products because of this shortage?

Robbie Whelan (8:50)

The most notable example of that is what OpenAI did with its Sora video generation platform. We saw about a month ago that OpenAI had basically ended this product, Sora, which people were using to make really fun videos and you could insert yourself into an action movie and make a little one minute clip. And OpenAI basically decided that that tool was not their top priority. Their top priority is competing in the agentic coding space. So they have a tool called Codex that they're concentrating on a lot more. And they basically just ended this project of Sora, which was a really, a lot of consumers thought was a really cool product. A lot of businesses and especially in the entertainment industry were excited about it and it only lasted six months. So we've got dramatic decisions being taken by companies like OpenAI to prioritize the type of computing tasks that they think are going to be most profitable in the long run that are going to be most widely used by customers. They're really having to make some hard decisions and trade offs about what to focus on.

Bell Lin (9:52)

So how do we solve for this computing shortfall? What are some possible solutions further down the line?

Robbie Whelan (9:57)

There's going to be some developments in the semiconductor industry. There's a lot of startups right now that are working on trying to develop more efficient chips. So in other words, the chips that you use to train a big AI model are not going to be the same ones you use to run it down the line. That's sort of a long term solution for the short term. A lot of consumers and a lot of businesses in this space are just going to have to live with the reality that you can't put up a data center immediately. The reality they're going to be facing these harsh trade offs for the next year or two, probably.

Bell Lin (10:27)

That was WSJ reporter Robbie Whelan. And that's it for Tech NEWS briefing. If you're a listener on Spotify, be sure to leave us a comment. Today's show was produced by Julie Chang with supervising producer Katie Ferguson logging off. I'm Bell Lynn for the Wall Street Journal. We'll be back later this morning with TNB Tech Minute. Thanks for listening.

Robbie Whelan (10:48)

Foreign.

Miriam Gottfried (10:53)

Hey, this is Telus Demos and I'm Miriam Gottfried.

Bell Lin (10:56)

We're reporters at the Wall Street Journal and The hosts of WSJ's take on the Week. It's a weekly show that gives listeners a leg up in the world of markets and investing.

Miriam Gottfried (11:04)

From the Fed's moves to market bubbles, we dive into the biggest deals, key players and business news ahead. If you're looking for more news and tools that you can use to help navigate the markets, consider becoming a subscriber to the Wall street journal.

Bell Lin (11:16)

Visit subscribe.WSJ.com take on click to subscribe now.