How battles for dominance in the chip market could determine the future of the economy and business
By Maria Carolina Abe, featuring Chris Miller, Associate Professor of International History at The Fletcher School
For Chris Miller, author of The Chip Wars, the future of the next world conflicts depends, in large part, on the negotiations that are being forged today about the production and sale of processors, which contain within them all the power of the new generative artificial intelligence.
All power in the modern world — military, economic, geopolitical — is built on a tiny silicon foundation. And the future of global conflicts depends largely on the negotiations that are being forged today over the production and sale of these processors, which contain within them all the power of the new generative artificial intelligence. This is the central thesis of the book The Chip Wars: The Battle for the Technology That Moves the World, a New York Times bestseller and the best business book of 2022 by the Financial Times. In the book, economic history expert Chris Miller explains in detail how this new world economic order was formed. Época NEGÓCIOS spoke with Miller, who works as an associate professor at Tufts University in Massachusetts, about the prospects for this trillion-dollar market, the future of the trade dispute between the United States and China, and what the coming years hold for Nvidia.
Your book was published in October 2022, before the launch of ChatGPT. How do you see the changes in the last two years?
Chris Miller: The biggest surprise for me was the extraordinary scale of investment in AI, with companies like Alphabet, Microsoft, Meta and Amazon spending much more than expected and building vast data centers focused on artificial intelligence. This made Nvidia not only the most important company in this new industry, but also the leader in a sector that is experiencing one of the largest investment booms in history.
In addition to investing in data centers, big tech companies are increasingly interested in producing their own chips. Do you think they will be able to steal market share from Nvidia?
Miller: I believe that big tech companies have two reasons for designing their own semiconductors. If they can design chips optimized for their specific workloads, they will be more efficient. And the second reason is the need to reduce their dependence on Nvidia, which currently sells 90% of the GPUs used in AI training. If companies can produce their own processors, it will create some competition and allow them to negotiate a bit with Nvidia on the price of those chips.
The US government has announced financial support for companies that want to produce chips on US soil. Do you think this initiative will yield results?
Miller: I think the main goal of the US is to reduce dependence on Taiwan and China, which is quite achievable. To that end, they have been financing new facilities, both from US companies like Intel and from foreign companies like TSMC in Taiwan and Samsung in South Korea. There has been a dramatic increase in investment in chip manufacturing in the US. We see this in the vast complexes that are being built in Arizona, Texas and elsewhere. I think the long-term issue is whether these facilities can become cost-competitive with Taiwan. Right now, no one can compete with their efficiency. That is because Taiwan has connections to all the suppliers in the chain: chemicals, materials, machinery, etc. As the US develops its own infrastructure, some of the pain points that have driven up costs will be eliminated, and the differences with Taiwan’s processes will become smaller. It will never be the same, but if the country comes to be seen as one of several viable locations for the production of advanced semiconductors, alongside Korea, Singapore and Japan, it will already be a step forward.
How are the US government’s measures to restrict chip sales to China affecting the industry?
Miller: They certainly hurt China’s ambitions, since it is unable to access the most advanced chips from Nvidia, for example, and cannot manufacture such sophisticated products. China is seeking to become self-sufficient in chips – unlike the US, Japan and Europe, which only want a slice of the market. And it is spending more money than any other government in the world to subsidize its own industry. But it faces a major challenge: it has very little production of some key components in the supply chain – such as certain chemicals that are used in manufacturing. China imports virtually everything it needs from the US, Japan and the Netherlands. They will spend a lot of money, time and energy to make substantial progress in this quest and reduce their spending on imported products.
What are your concerns about the situation in Taiwan? What might happen if China decides to invade the country, for example?
Miller: It’s a very scary situation. First, because the Chinese government says it reserves the right to use military force against Taiwan. Second, because the Chinese military’s moves around Taiwan are substantial. Every year, China narrows the gap between its military power and the United States, so that it’s no longer clear who would win today if there were a world war. Those two factors combined are enough to make anyone very nervous about the situation. People say that Xi Jinping [Chinese president] won’t invade Taiwan because he’s focused on growing the economy. Maybe, but I don’t think we should put our trust in the whims of one person who happens to be a dictator who has repeatedly arrested and executed his political rivals, including his own ministers. I don’t know what Xi Jinping will do tomorrow or in five years. But the reality is that for most of the last 50 years, if China tried to take Taiwan, it wouldn’t have a chance against an ally like the United States. And now that’s no longer a sure thing. So that changes the political calculus in Beijing and increases the likelihood that something will go wrong, even if everyone in the Chinese government is really rational and focused on the economy. Every year, the chances of something going wrong are getting higher.https://3ac960dff633e2dd722175397f9a2fce.safeframe.googlesyndication.com/safeframe/1-0-41/html/container.html
What other countries are emerging as major players in chip production?
Miller: I don’t think other countries have much of a chance, actually. For a long time, it will be a split industry between Taiwan, Korea, Japan, the US and the Netherlands – which will play a major but niche role. That will be the supply chain for Western countries.
How do you see Nvidia competing with companies like Intel or AMD? Do you think Jensen Huang’s company should continue to maintain its position as the leader?
Miller: What strikes me about this story is that Nvidia’s leadership position comes not just from the fact that its chips are very well designed, but also from its effectiveness in creating a software ecosystem around those chips. Other companies, like AMD, are trying to build their own ecosystem, but I think it’s widely recognized that they’re way behind.
How long will Nvidia be able to remain one of the most valuable companies in the world?
Miller: The key question for me is: Three, four, five years from now, how much money will big tech companies be spending on GPUs? If they’re spending as much as they are now, and if that continues for a long time, then Nvidia will be a very, very valuable company. But if we’re already at the peak of a cycle and in a few years the big tech companies pull back on their spending, then it’s going to be a very tough time for anyone selling GPUs. In fact, I think the biggest risk for Nvidia is not what Nvidia does. It’s what its core customers do.
You said in a recent interview that companies are overinvesting in AI and investors are going to demand returns soon. Do you think we’ve reached the peak of this cycle?
Miller: I think every cycle, when it starts to slow down, is because there’s a realization that too much has been invested and not getting the desired returns. Investors are starting to ask: Where’s the profit? We’ve seen tens of billions of dollars invested. What are the results? For now, I think they’re being told by tech company leaders that they need to wait a little longer for the technology to mature. But I think they’re starting to get impatient. Unlike a year ago, when most investors were just excited about AI, there are also a lot of serious questions being asked today. If there are good answers, the cycle could continue for a few more quarters. If not, they could start demanding to keep putting money into GPUs. And eventually, there could be a pullback in investment.
In your view, is there any chance that semiconductors will be replaced by some other material, ending the chip wars?
Miller: It will take a while for that to happen. The truth is that all computing is written on a silicon foundation. So as long as chips continue to get more sophisticated as they are today, it will be very difficult to replace them.
(This post is republished from Época NEGÓCIOS.)