How the great chip war came to be
Today, more than a third of all logic chips—the processors in computers—are produced in Taiwan.
When it comes to the most advanced of them, this number goes up to 90 percent.
In his new book “Chip War”, American economic historian Chris Miller from Tufts University in Massachusetts, USA, highlights that a disaster in Taiwan that would bring this production to a standstill would have thousands of billions of dollars’ worth of follow-up costs.
The economic fallout would be far worse than that of the COVID-19 pandemic.
Replacing the huge, advanced chip factories in Taiwan is not exactly simple.
It would take at least five years to build a similar capacity elsewhere in the world, Chris Miller notes.
“These days, when we look five years out we hope to be building 5G networks and metaverses, but if Taiwan were taken offline we might find ourselves struggling to acquire dishwashers.”
How did we end up in a situation where advanced weapons systems, car manufacturing, mobile phones, and much more has become dependent on Taiwanese chip factories with machines that can even only be supplied by a single company based in the Netherlands?
This is the story Chris Miller unfolds in his well-researched book.
The transistor and the chip
The first transistor was manufactured in December 1947 by Bell Laboratories, but the semiconductor industry was truly born about ten years later, when Jack Kilby at Texas Instruments in 1958 and Bob Noyce at Fairchild Semiconductor the following year independently manufactured the first integrated circuits, i.e. chips.
It happened shortly after the Soviet Union had shocked the whole world by launching the world’s first satellite, Sputnik, in 1957.
Semiconductor technology became the American path to winning the technological race and the Cold War.
Transistor technology was refined and improved by highly skilled physicists and engineers, not least in the area south of San Francisco, which we today call Silicon Valley.
The focus quickly shifted to mass markets and consumer goods, although transistors were also absolutely crucial to the development of new precision weapons.
Chris Miller describes that at the beginning of the Vietnam War, American bombs fell an average of 125 meters away from their targets. For example, over the years, Americans had sent 638 bombs targeting a bridge near Thanh Hoa but failed to destroy it. When American planes dropped new bombs equipped with laser sensors and transistor technology on 13 May 1972, the bridge was destroyed in the first wave of the attack.
The technological advances in the semiconductor industry helped change the world order, creating winners and losers.
Access to and control over the most advanced chips has become so important that the USA seeks to, among other things, prevent China’s access to them. At the same time, China’s ambition to annex Taiwan is growing—perhaps by force if necessary—and we in Europe are trying to build a local chip production, so that, for example, the German automotive industry does not have to depend on what happens on the other side of the globe.
The decisive element of this development was that some companies very early on, and before anyone else, discussed globalization and understood how to take advantage of the cheap production conditions in Asia. Chips were designed in one country, manufactured in another one, and inserted into products in a third one.
This led countries in Asia with cheap labour, first Japan and then Korea and Taiwan, to capture large shares of the chip market, while neither the Soviet Union nor Europe managed to position themselves at the forefront.
Chris Miller explains that the Soviet Union failed because it pursued a copycat strategy that had proven useful in other industries, but could not be used at all in semiconductor technology.
Although the Soviets acquired chips through both legal and illegal channels as well as information about production methods through espionage, the country constantly lagged behind because mass production of chips involves know-how that cannot be copied.
De Gaulle and the transistor salesman
We do not hear much about developments in Europe, but Chris Miller illustrates with a small anecdote that European political leaders, in contrast to the leaders in Asia, did not understand one bit of the developments that were taking place in the rest of the world.
When Japanese Prime Minister Hayato Ikeda visited French President Charles de Gaulle in the pompous Élysée Palace in 1962, he brought a Sony transistor radio as a gift.
Sony was founded in 1946 by Akio Morita. The company’s first products were an electric rice cooker and a tape recorder, but Morita immediately realized the future importance of transistors, so when he first came to the USA in 1953, he managed to obtain a license to manufacture transistors. The Japanese licensing strategy worked much better than the Soviet copying strategy, Chris Miller notes.
Morita also had a sense of which new products consumers wanted before they were even aware of it—quite comparable to what Steve Jobs later managed at Apple.
Texas Instruments had attempted selling transistor radios before Sony but failed because they were priced and marketed incorrectly. Morita and Sony understood better how to capture a consumer market. For example, this success was later repeated with the Walkman.
However, de Gaulle was not impressed by his gift. He was a military man with a sense of culture, pomp, and pageantry—not cheap electronics. In his eyes and in his words, Ikeda was nothing more than a transistor salesman.
A silicon shield
When it comes to Taiwan later succeeding in gaining a foothold in the global chip production, a large part of the credit goes to Morris Chang.
He had fled China when the Communists assumed power after World War II. He was educated in the USA, and in 1958 he was employed by Texas Instruments to manage their production. He was quite successful and quickly rose through the ranks.
In 1985, Chang received a decisive call from Taiwanese economist and minister Li Kwoh-ting. Li and Chang had both been instrumental in Texas Instruments’ decision to build a factory in Taiwan in 1968.
Li’s ulterior motive at the time was that it would be helpful for Taiwan to have close economic ties with the USA. Perhaps the Americans, many of whom were already tired of waging war in Vietnam by then, would be more inclined to defend Texas Instruments than a poor island nation, Li thought.
But there was more to Li’s plans. He wanted to elevate Taiwan to a whole new level by establishing an independent chip production—a company that could produce the chips designed by other companies that did not have their own production facilities. He hired Chang to build a semiconductor manufacturing facility, or fab as they are called today. Chang was only asked one question: “How much money do you need?”
However, money alone was not enough. Although Chang knew the semiconductor industry inside out, he needed a technology partner. Neither Texas Instruments nor Intel thought the fab idea was a good idea and rejected a collaboration. Instead, Chang made an agreement with the Dutch company Philips.
Taiwan Semiconductor Manufacturing Company, or TSMC, as the new company was dubbed, had close ties with the USA, where most customers were located, and Chang brought in employees from Motorola, Intel, and Texas Instruments.
As a result, TSMC could mass produce chips at a low price. The customers were also satisfied that TSMC was not a direct competitor that sold its own products, as would have been the case, for example, if Intel had been in charge of the production.
TSMC was not the only company that offered this service, but as the years went by, the production equipment became increasingly more advanced and expensive. Each time a new generation of chips was introduced, fewer and fewer manufacturers were capable of producing the most advanced chips.
Today, only TSMC and Samsung can produce the most advanced chips—even Intel dropped out at some point, but they have plans to re-establish their position among the leaders in the future.
No one has a rational interest in destroying chip production in Taiwan. Not even China, because the country is also highly dependent on it. When Apple manufactures mobile phones in China, they do it with chips produced at TSMC. Without chips from Taiwan, large parts of Chinese electronics production would disappear.
This has led Taiwan’s President Tsai Ing-wen to describe the island’s chip industry as a defence—a silicon shield. The question is whether the shield will last. In June, Bloomberg News quoted a leading Chinese economist saying that China would simply have to capture TSMC if the USA imposed sanctions on China similar to those that are currently in force against Russia.
Chris Miller notes that with the Chinese military build-up in recent years, it is possible that Beijing would think that it is a realistic possibility.
The world’s most expensive machine
“Chip War” is first and foremost a business book, but its technical descriptions of why chip production is so complex are at a surprisingly high level.
The continuous improvement of the computing power of a small chip, as predicted by Gordon Moore in 1965, is possible because it is possible to manufacture chips with smaller and smaller details through lithography, where the semiconductor material is illuminated during the manufacturing process. By using shorter and shorter wavelengths, it has been possible to produce smaller and smaller transistors and thus more advanced chips.
The first generations used visible light. Today’s chips require extreme ultraviolet lithography (EUV)—radiation with a wavelength of 13.5 nm—quite close to the X-ray wavelength, which goes up to 10 nm.
Today, there is only one manufacturer of EUV lithography machines in the world. It is the Dutch company AMSL, originally a spinoff from Philips. Their equipment comes with a price tag of over USD 100 million. The next generation will cost around DKK 300 million.
They are the most expensive—and perhaps most complicated—mass-produced machines in history. Only the very largest chip manufacturers can afford to buy them.
The laser system alone consists of 457,329 parts that must be assembled in just the right way. This takes place in the Netherlands, but large parts of the components, including the light source itself, come from the USA. Therefore, the American authorities can control who is allowed to buy these machines and thus produce the world’s most advanced chips.
An excellent book
The world’s first microprocessor, Intel 4004, was launched in 1971—the same year I started studying electrical engineering at DTU (then DTH).
I have followed developments in chip technology more or less intensively since then and therefore know quite a bit about the subjects and the people Chris Miller mentions and highlights. But I have never before read such a concise and accurate description of how the semiconductor industry came into existence and became a key industry and technology.
I have also never before seen the geopolitical significance of this technology development presented in such a convincing way.
It is an exceptionally excellent book that Chris Miller has written. It has exciting elements that make the reader keep turning the page. It contains many notes, so one can check where Chris Miller got his information from. It is packed with a wealth of information and thought-provoking considerations.
The book is a must read for anyone with an interest in the history of the semiconductor industry and for anyone interested in the connection between technology and geopolitics.