How Semiconductor Chips Became Too Big to Fail
Chris Miller on his new book, Chip War: The Fight for the World’s Most Critical Technology.
Gabfest Reads is a monthly series from the hosts of Slate’s Political Gabfest podcast. Recently, John Dickerson spoke with author Chris Miller about how Moore’s Law changed everything we know about the chip-making business and his new book, Chip War: The Fight for the World’s Most Critical Technology.
This partial transcript has been edited and condensed for clarity.
John Dickerson: Remind people what Moore’s Law is. And it feels like one short way to tell this story is essentially, everybody involved in the chip-making business, trying to keep up with or fulfill the promise of Moore’s Law, and that that’s led to the offshoring, the innovation, the fact that we are in this concentrated position we’re in today.
Chris Miller: Gordon Moore was an early engineer in the chip industry, he would later go on to co-found Intel. And he noticed in 1965 that the number of components on each chip was doubling every year or so. And that meant that chips were getting twice as powerful every year or two.
And that dynamic has been maintained up to the present. For the past several decades, every two years, chips have gotten twice as powerful. And what that means is that the chip industry has improved vastly more rapidly than any other part of the economy.
And I like to think about, what’s an analogy in other sectors of the economy? What if airplanes flew twice as fast every two years, and that rate of change was steady for the past century? It’s sort of impossible to imagine, but that’s why remembering data on some kind devices cost less than a millionth of what it did 50 years ago because of that exponential growth rate.
I always assumed that the idea was you always went to cutting-edge chips, and that there was no use for anything that wasn’t right at the cutting edge. And so, I was interested to learn about the different kinds of chips that are useful and various different kinds of things.
But essentially, keeping up with Moore’s Law means trying to fit more stuff in a smaller space, more powerful stuff in a smaller space. Which requires extraordinary feats of design. I mean, it was hard to conceive sometimes about how small the etchings you were talking about were.
If you go to an Apple Store today and buy a new iPhone, the primary chip on that iPhone will have 15 billion tiny transistors carved into it. And so, to fit 15 billion devices on a chip the size of your fingernail, each one of them is the size of a virus. And they’re produced with basically perfect accuracy.
I love, I think it was the ASML machine that could hit a golf ball and hit the moon.
That’s right. They could hit a golf ball sitting on the moon, firing that laser from Earth.
Oh my God. All right. So, this notion of concentration, why has the industry become so concentrated?
There are two main reasons. The first is that if you produce more chips, you can gain economies of scale, costs get lower because you’re able to buy machinery more efficiently, acquire materials more efficiently. So, the cost is part of it.
But the second is, the more chips you produce, the more you can hone your production processes and advance your technology. And so today, it’s not a coincidence that TSMC, the Taiwanese company we mentioned, is both the world’s largest chip-maker and the world’s most advanced chip-maker. Because they get data from every chip they produce, and then use that to produce the next chip more efficiently.