The chip and memory stock frenzy

Katie Martin (0:06)

Pushkin.

A new divide is opening up in stock markets between tech stocks and, well, other tech stocks. Investors still just cannot get enough of the whole AI thing. Stocks in the ecosystem of semiconductors or chips to their friends are still ripping higher, absolutely killing it. But software stocks have been getting clobbered. What does that tell you? It's that AI will kill at least a large part of the industry that forged it in the first place. There's gratitude for you. Today on the show, Chips and Dips, we unpack these divergent fortunes and ask, are chips in a bubble or a super cycle?

This is Unhedged, your friendly markets and finance podcast from the Financial Times and Pushkin. I'm Katie Martin, a markets columnist at FT Towers in London, where it is aggressively hot. Joining me all the way from over there in New York City is Rob Armstrong, my partner in crime on the Unhedged newsletter with his fancy new glasses. Rob, how's it going?

Rob Armstrong (1:15)

You know where else it's hot, Katie, is in the FT audio studio in New York. I want all our listeners to imagine a large middle-aged man slowly turning into a puddle in here.

Katie Martin (1:29)

A large middle-aged man in quite a small box, actually. You could probably reach each wall, can you?

Rob Armstrong (1:34)

Yeah, just about touch the walls. Not quite, but just about. Today, I'm a little sweaty.

Katie Martin (1:39)

Okay. That's a pleasant thought for all of us. So what's going on here?

Ships are absolutely off to the races, right?

Rob Armstrong (1:48)

Yeah. So we all know that there is a mad rush in the United States to build as many data centers as possible, as quickly as possible. So that we can run AI models on them and do all sorts of wonderful, efficient things with those AI models.

The bottleneck or one of several bottlenecks to building all those data centers is having enough chips to support them. And it's like every kind of chip you need. It's not just NVIDIA's fancy GPUs, which are the kind of main central brain for training and using an AI. It's also old-fashioned CPUs. It is very much memory chips. It's networking chips.

It's the whole range. And so everything-

Katie Martin (2:42)

All my favorite types of chips, yeah.

Rob Armstrong (2:44)

Yeah, exactly. Tortilla chips, corn chips, potato chips, through the roof.

And so there's been wild stuff. Like Micron, which makes memory chips, just to pick one example out of a hat, is up almost 1000 percent this year.

Katie Martin (3:03)

Shut up. It's absolute.

Rob Armstrong (3:05)

Because they just can't. They cannot make-

it's not almost 1000, but it's in the 800s, and they can't make this stuff fast enough.

Katie Martin (3:14)

So this is like the one that's become a trillion-dollar company kind of from nowhere.

Rob Armstrong (3:18)

It has become a trillion-dollar company. And this raises a lot of interesting questions for the following reason, which is that in one very important respect, computer chips are like cows.

So follow me here for a second.

The hard thing about the beef industry, cattle industry, is that production follows prices, and production leads inevitably to gluts and then to lower prices. So the price of beef goes up. Everybody's like, holy crow, let's make more cows. And so you make loads of cows. And that process of making more cows, which involves getting cows together and all this kind of, takes several years.

Katie Martin (4:12)

Mommy cow and daddy cow love each other very much.

Rob Armstrong (4:16)

By the time you have more cows ready to be turned into hamburger, the price has changed.

And similarly, in the chip industry, it is cyclical in exactly the same way. Prices rise, people are like, oh, let's add capacity. Let's make as many chips as we can quickly, quickly. This is actually a process that takes a long time. And then by the time all that, all those chips are made, the prices have fallen, you have a huge gut. You don't have a huge gut. That would be the...

Katie Martin (4:47)

Well, you might do if you eat all the hamburgers.

Rob Armstrong (4:49)

It's true. Good point. You have a huge glut. So, you know, it's like the classic cyclical industry where if you look at price charts or revenue charts or whatever, profitability, it's like this sine wave that rolls through. And so the question confronting us is, are we merely at the top of a chip cycle, an especially big chip cycle, and we're headed for a glut, or has the world economy changed in a permanent way, such that demand for chips is permanently higher and the whole industry is going to rerate for good and things are different now?