Bloomberg Audio Studios, podcasts, radio News.
Hey, I got to say that there are moments in time like today, with all the spend and build out and companies doing things related to AI and questions around the return on investment, the ROI and the impact of artificial intelligence, that we kind of wish we could pull out that crystal ball and see a little bit into the future.
I think it's fair to say Chris Miller saw something years ago. He's the author of the twenty twenty two bestseller Chip War, The Fight for the World's Most Critical Technology. He joins us from Pittsburgh this afternoon. Chris, it's good to have you back with us. Carol and I were just talking, and you know, you wrote this book close to four years ago. At this point, how different is the world of semiconductors and the world's reliance on semiconductors now than when you wrote the book.
Well, I think the key difference is just the scale of spending that we've got right now on building data centers and buying all of the chips that are required inside of them, and Nvidia first and foremost, but not only, we've seen companies transformed by the data center build out but in a lot of ways, not much just changed other than that we're still sourcing our key semi conductors from Asia and above all from Taiwan, and so the industry is supercharged, its size, growing faster than ever, but
dealing with some of the same supply chain choke points that it had a decade ago.
You know, we talk a lot with one of our brilliant voices on the chip world, and that is our own Ian King, who has seen a lot of cycles, the ups, the downs, and we talk a lot about that supply demand imbalance and what happens when there's a lot of demand, there's not enough supply. The buildout continues, the investment supply goes up, and then there's a glut. So I'm just curious what your view is when it
comes to the cyclical nature of the semiconductor world. Has AI changed it or we we'll see also and over build and oversupply and then a glut.
I think if you look historically, certainly you see cycles up, cycles down, but there are moments when you see step changes in terms of demand for certain types of chips.
We saw that with smartphones, for example, there was no demand for smartphone chips and then now everyone needs a new smartphone every couple of years, and what we're seeing in AI right now is that type of step change, a huge increase in just the baseline amount of chips that we're going to need for data centers driven by AI, and so I think we shouldn't expect cycles to be over Certainly there'll be ups and downs in the future,
but it's now very clear. I think that we're just going to need a lot more compute for AI purposes in the future, and as a result, will need a lot more of the AI chips that go inside of data centers.
In your view, is the promise that many people think AI is supposed to deliver, Will it actually be delivered? Like, what's the future that you envision after all this capex is spent?
Why? I think it's in some ways funny when we ask will AI deliver? Because if you look just three years ago, even after chat schipt was released, there are so many things that were not possible that are possible today, whether it's the number or the scale of hallucinations and answers that chatbots give you, or the scale of work that you can put together today that just wasn't possible
in the past. We've already had so much new capability generated just in the couple of years since Chat schipt that in some ways, I think it's an absurd question to ask will AI deliver it already has in a lot of ways. But I think I understand why there's plenty of questions about what about the investment is happening right now, will that investment pay off? And I think there are reasons to think carefully about how much each
company is putting in over what time horizon. But I guess when I zoom out, I ask myself, do I want a world in which there is more access to compute or last? And it seems to me that we should, rather than being concerned that there's too much compute being put in the ground, be excited that there's actually companies willing to invest in the infrastructure that's going to deliver all these capabilities.
Well, I think it's an important question to ask for a few reasons, and one of those is because if you look at just the capex that these companies are spending, you know, one company two hundred billion dollars in a single year. That's a big commitment, and that's something that they have to convince investors that is money that is going to the right place, that's money that has to
be earned back. Plus on companies coming to a hyper scaler and saying, yes, we think that this money is not only being well spent, but then we can use this compute to actually create a product that will provide a return on investments. So, yes, we've seen a lot of the hyperscalers benefit the you know, the anthropics and the open ais and the mistraws like that's amazing stuff.
But at the end of the day, there are a lot of companies that are not necessarily technology companies that maybe aren't yet necessarily seeing an increase in productivity as a result of these tools. So my question is do those companies start to see that do we live in a world where this is a layer, just like the Internet was a layer of technology.
I think there's no doubt that we're going to have AI as a layer that's embedded into all technology that we use. And when I look at the economic question, I say, first off, is it profitable to serve AI systems today, not train the next generation, but serve today.
And if you listen to what's publicly reported about OpenAI or anthropic, their margins on their inference business are not just positive but quite good, and so that I think is pretty strong evidence that the delivery of already existing AI services is a pretty profitable business. The next question is should we be investing in R and D in the next generation, which is of course very very expensive.
But I think it's hard to argue we should dramatically slow down R and D and AI just given all of the extraordinary improvements that we've seen over the past couple of years in terms of capability. And so when you start breaking it down and ask, well, which specific investment do we think is excessive? Which specific investment would rather not be doing? Would you really like to be the CEO of the only big tech company that's not investing in AI. That doesn't seem like a very comfortable
place to be. And I'd rather have a situation in which the big technology companies are investing more rather than putting money in the bank or buying back their stock because they didn't have any comfitable by investment opportunity.
Well, you know, I would say Apple's doing it differently, and I guess time will tell whether or not their approach works out. Chris, I am curious to you know, who do you talk to, what research do you follow, what are the leading voices, what are the leading companies that you watch to figure out kind of how the semiconductor space is evolving. I mean, we still know TSMC is still the big manufacturer of all chips in the world,
and there's geopolitical attentions to that. But give us an idea of where do you think kind of investors and just the world at large need to be focusing their attention on. When it comes to the semiconductor world. Is it Nvidia, is it China? Is it something else?
I think the hard problem is that it's all of the above, and we've seen this play out in the GPU supply chain over the past couple of years, where we've had shortages and different types of components, different materials, and the memory chips that go next to GPUs and AI servers. Each part of the supply chain has had to dramatically ramp up its production capacity to respond to the surge and demand. And so if you only look at one part of the supply chain, you miss the
challenges that other parts are often facing. And so it's the chip designers. It's the chip makers, but it's also the materials suppliers, which are often not even thought of as being semiconductor firms, but produce many of the capabilities that are critical to actually manufacture the AI chips and servers that we need.
Do you think the world do you think the US specifically does need to be restricting sales of its most sophisticated chips to China or on others? I mean, we did see Nvidia. They still face the uncertainty in China, that's their largest market, or which it is the largest market for chips. The government granting some licenses to ship a small amount of some of its processors to customers there, but and Vidia is not sure if the Chinese government
will give its approval. So there's still some back and forth here. But is it still an arms race of source.
Arms race of sorts?
Excuse me?
Yeah, I think armed races is not a bad analogy. When you listen to tech CEOs, they'll speak in the same language. They're struggling to get access to all the computing power that they need, that they envision needing more tomorrow than they've got today. And if you listen to Chinese technology leaders. What you hear from them is challenges
and getting access to computing power. And the primary reason they've struggled to deploy AI products at scale is because they've struggled to get access to all of the computing capabilities. And it's on a regular basis we see new Chinese models launched that can actually be deployed at scale because they don't have access to the advanced chips that they need.
So this does, I think, seem to me that this is still a very powerful card that the US has to play, and so I think we should be very careful or on any decisions to give China more access or at least make sure that we're getting something in exchange for that.
On that, Chris, the US support of TSMC and the US support of Intel different types of support, And I guess you could call the TSMC one encouragement to build here in the United States. What is the right industrial policy to reduce reliance on companies outside of the US.
It's a really hard problem because TSMC is such a capable and efficient manufacturer in their home base in Taiwan. But I think the US government is right to say that we need a more diversified manufacturing base for the world's most important semiconductors. I think we've learned over the last couple of years there's no silver bullet. President Trump's tried tariffs that comes with obvious downside. President Biden tried subsidies that worked to a degree, but only to a degree.
Here's the reality. The chip industry has involved hundreds of billions of dollars of cappax over the last several decades, and so it's just not going to move fast. And so if you want to slowly change the structure of where chips are produced, you've got a plan for years of years of implementation of measures designed at the shift economics of the GIP industry.
All right, Kenly with there, Chris Miller, thank you so much. Good to check in with you. Professor of International History at the Fletcher School at Tufts University, author of Chipward joining us on this Thursday.