Chip War: The Fight for the World’s Most Critical Technology

The Chip War by Chris Miller came out in October 2022, around the time COVID-19 was going down. I remember reading about the book release and thinking this book will be a must read for me. As a researcher working in the semiconductor chip industry, there was no excuse on skipping it. However, it was not until recently that I was able to start reading it.

From within the world of science and technology, today, the field of semiconductors is arguably one of the most politically active and discussed topic. In this book, Miller is attempting to not only illustrate the significance and our unimagined dependence on semiconductor chips or simply ‘chips’, but also underscoring the staggering vulnerability of the chip manufacturing technology (politically, technologically and logistically) and therefore its impact on our everyday life.

Chips are ubiquitous; but it was not until the disruptions in the chip manufacturing that were felt through the automobile industry [1,2,3,4], the “chip shortage” period, that we realized how much we depended on this technology. Interestingly, the chip shortage topic was covered so heavily by the media and affected so many people who wanted to buy a car, that since then, whenever I meet someone new and tell them that I work in the semiconductor industry, their response isn’t a blank face anymore, but one that is ready with many questions. With this blog article, I would like to share the conclusions made by Miller and my thoughts on them (and on the book). I also encourage all potential readers to openly participate in the discussion if interested.

If you would like to directly jump to my thoughts about the book, please click here to go to the end.

Introduction

Miller does not hesitate, and dives straight into the international politics of chip technology, and why not, after all the title of the book is “Chip War”. Throughout the introduction, he often compares microelectronics or the chip technology to other, more familiar, critical technologies, such as atomic, steel and oil, to emphasize that chips are not merely to enable a better life with smartphones, video games, TV, but are vital technology critical for national defense, healthcare, finance, communications etc. However, unlike, oil and steel, which can be imported through many channels, sources and countries, chip manufacturing depends on an extremely complex interwoven network of companies, countries and channels, and any glitch in the perfectly synchronized flow can disrupt the whole process, which was what happened during the chip shortage period [1,2,3,4]. At the center of the politics of chip technology lies the US – China relationship and the struggle to dominate the chip technology.

“World War II was decided by steel and aluminum, and followed shortly thereafter by the Cold War, which was defined by atomic weapons. The rivalry between the United States and China may well be determined by computing power.” – Chris Miller

Keeping aside the politics and the logistics that enable chips, the technology in itself is one of the greatest engineering feats. Central to computing are the ones (1) and the zeros (0). Everything in the digital universe is just a series of 1s and 0s, stored into the chips as electric charge present (on or 1), and as electric charge absent (off or 0). Each individual 1 and 0 is stored in a tiny unit called a transistor, which can be smaller than the size of a coronavirus. The Financial Times has done a great job illustrating this with an intuitive infographic [5]. Consider, iPhone 12’s chip A14 processor, it is composed of 11.8 billion of such tiny transistors [6]. To emphasize on this impressive engineering achievement, Miller mentions that just 60 years ago, the state-of-the-art chip had not 11.8 billion, but only 4 transistors [7]. Since then, the number of transistors on a chip, which can roughly represent the computing power has almost doubled every year, better known as “Moore’s Law”, predicted by Gordon Moore in 1965 [8], which still pushing the chip industry to its limits.

Miller concludes the introduction by reminding the readers about how the lifeline technology of today is openly vulnerable to international politics, natural disasters and technological challenges. Never before has ever existed such a technology that depends on extremely complex international interconnections, and “… has been shaped not only by corporations and consumers but also by ambitious governments and the imperatives of war” ends Miller.

Part 1: Cold War Chips

The story of chips start in the backdrop of the World War 2, when the key players of the semiconductor industry were wrapped up in the ugliness of this major conflict. Akio Morita (co-founder of Sony) barely avoided the frontline by working as an engineer in the Japanese Navy, Morris Chang (the founder of TSMC) fled from place to place evading the conflict brutalities, Andy Grove (the third CEO of Intel and a key player in establishing the chip manufacturing process) suffered first at the hands of Hungary’s far-right government and later by the Red Army troops.

As described in the introduction, the outcome of the World War 2 was decided by the industrial capacity of the countries involved, however, it was clear that with advancement in technology, new systems will be replacing the military might in the future. The existing systems for calculations during the time heavily depended on humans as computers, hence prone to delays and errors. It was clear that the need for calculations was not going to decline, but only increase. Early calculators were based either on mechanical gears and levers, or on vacuum tubes, for example, the Electronic Numerical Integrator and Computer or simply ENIAC. These systems were useful, to a limit, and the need for better and more reliable systems than vacuum tubes was growing – something small, fast and reliable.

The vacuum tubes were eventually replaced by the ‘transistors‘. The story of the invention and the perfection of this device is interesting but a long one. It was through the joint effort of Bell lab scientists Shockley, Brattain and Bardeen, that resulted in a working device that could amplify electrical signals and work as an electrical valve – current on or ‘1’ and current off or ‘0’. The device was called ‘transistor’ and could work as a switch. The trio won the 1956 Nobel Prize in Physics for this invention. If transistor had to replace the vacuum tubes, it needed simplification and the possibility to be manufactured at scale. Jack Kilby, an engineer from Texas Instruments, started working on how to put multiple transistors together and connect them with minimal wiring. In 1958, he solved this problem by realizing multiple transistors on a single slab of semiconductor (monolithic), calling it ‘integrated circuit (IC)’, colloquially also known as a chip. However, the first true monolithic IC was designed half a year later by Robert Noyce, since Kilby’s invention still had gold wires connecting the transistors, while Noyce’s invention had wires integrated in the semiconductor slab itself [9]. This resulted in a smaller device with the possibility to be produced at scale.

There were still 2 big issues to solve, one financial – finding a market for transistors, and one technological – further simplification of the production process, which was needed to manufacture chips in bulk. The USSR was leading the space race by launching Sputnik and the first human in the space, leading the US president Kennedy to declare that the US would send a man to the moon. Suddenly the market for chips created itself – rockets. Both Fairchild semiconductors and Texas Instruments harnessed this market by manufacturing chips for the Apollo missions [10] and the Minuteman missiles [11]. The successful integration and operation of chips for both these complex tasks (Apollo and Minuteman) provided a seal of trust, resulting in further increase in the demand. However, without further simplifying the production process, meeting these demands was going to be cumbersome. The solution for this came from Jay Lathrop and James Nall of Texas Instruments, who figured out a way to ‘print’ the patterns for transistors and the wires on a silicon slab, resulting in a significant simplification of the process. Lathrop called it ‘photolithography‘ literally meaning ‘writing with light on stone’. Further efforts by Morris Chang at Texas Instruments and Andy Grove at Fairchild Semiconductors to improve the manufacturing yield helped to push for successful high volume production of chips.

Noyce of Fairchild knew that NASA and the military were crucial for the initial success to establish a market, but he always envisioned a much larger civilian market for the chips. To push for this, Fairchild started selling chips at a significantly lower price, sometimes even below the manufacturing cost, hoping for an increased adoption by the civilian market. The strategy paid off, and by 1968, Fairchild was selling as many chips to the computer market as to military. The result was a significant increase in the venture capital funding to manufacture chips for computers.

Part 2: The Circuitry of the American World

When Jack Kilby was awarded the Nobel prize in Physics for inventing the integrated circuits in the year 2000, he shared the prize with Zhores Alferov, a Soviet and Russian scientist who in 1960s studied on the ways semiconductor devices could produce light. The Sputnik was launched in 1957, Yuri Gagarin became the first human in space in 1961 and in a secret research facility, Yuri Osokin fabricated an integrated circuit (about 2 years after Noyce’s invention). These evidence point to the fact that the USSR was becoming a scientific superpower. However, the next decades did little push this momentum of innovation in microelectronics in catching up with the US.

Miller elaborates, even though the Soviet scientists’ understanding was at par with American scientists, the orders from the top were to simply to copy the design. However, the semiconductor world was changing at an exponential speed (remember Moore’s Law), hence, copying a process, finding the right materials, equipment took time, and by then the technology was too old. It is intuitively difficult for human brains to process exponential growths.

It was a difficult time for the US too. The direct military involvement of the US in Vietnam was dragging on, and then Washington pulling out resources and investment from many Asian countries was like someone just shot the albatross. But here was an opportunity too. Much of the final stages of chip assembly in the 1960s – attaching the silicon chip to the plastic casing – was done manually by hands. With chip production increasing, the manufacturers needed cheap labor for the final step. The American worker’s hourly wage in 1960s was about $2.5, compared to 23 cents in Hong Kong, 19 cents in Taiwan, 15 cents in Malaysia, 11 cents in Singapore and 10 cents in South Korea [12]. Globalization was never in the plans of the people who laid the foundation of the semiconductor industry but there was no choice. First it was Fairchild Semiconductors, then Texas Instruments, Motorola and other, to start offshore assembly lines in Asia. The plan benefitted Asian countries too, “laying the grounds for the Asia-centric supply chains as we know today.“.

“From South Korea to Taiwan, Malaysia to Singapore, anti-Communist governments were seeking assurance assurance that America’s retreat from Vietnam wouldn’t leave them standing alone. They were also seeking jobs and investment that could address the economic dissatisfaction that drove some of their populations toward Communism.” – Chris Miller

The integration of Asia in the semiconductor industry did not remain restricted to the cheap labor for chip assembly. In Japan, Akio Morita, who laid the foundation of Sony with his colleague Masaru Ibuka, took advantage of this situation. However, instead of focusing on transistor or chip manufacturing, Sony focused on using these chips to innovate and invent new consumer electronics. Such strategies by many Japanese companies resulted in Japanese export of electronics to increase from $0.6 billion in 1965 to $60 billion by 1985 [13,14], a 100-fold increase. Similarly, the integration of other Asian countries into the semiconductor industry provided them with various benefits, Taiwan’s integration helped it economically and strategically, 7% of Singapore’s GNP by early 1980s came from electronic industry, employment boomed in Hong Kong (second only to textiles) and in Malaysia (15% of population that moved from rural to urban areas) [14-17].

The chips were essential not only for consumer electronics but also for military and for the weapons of the future. During the cold war, this made the both the US and the USSR significantly dependent on chips, resulting in escalated development of ‘guided’ or ‘smart’ weapons using semiconductors. All such applications of chips typically required custom chip manufacturing, which was not a good path to enable mass manufacturing. The breakthrough came from Intel, when Ted Hoff realized that with enough (DRAM) memory, computers will be able to handle complex calculation and so a standard logic chip coupled with large DRAM memory and specialized software could replace the need for custom chips. This was a significant leap, as it enabled mass production of standardized logic chips, with the first one being 4004 – world’s first microprocessor.

Part 3: Leadership Lost?

In 1970, only two years into its foundation, Intel introduced DRAM, a revolutionary product into the market [19]. However, just 10 years later, when 64K DRAM was in the market, Intel’s contribution to the global DRAM market was about only 1.7%, while more than 60% of the market was owned by Japanese companies [20]. This reversal of the market ownership was restricted not just to DRAM chips, but to many more sectors, such as consumer electronics (Sony Walkman, one of the most popular consumer electronic device), chip fabrication tools (by 1980, Japan supplied 70% of all lithography equipment) and many more. The initial American decision to encourage Japan to build a strong electronics industry, a strategy to prevent communist takeover during the Cold War [21, 22], was turning out to be costly.

This was not only restricted to the quantity of electronics and equipment, but also the quality. When Hewlett-Packard (HP) tested DRAMs chips, they found that the American made chips malfunctioned 4.5 times more than Japanese made chips [23]. Similarly, in the field of chip fabrication equipment, such as lithography, not only Japan dominated the market, but Nikon’s lithography equipment outperformed than GCA’s and had an average of 10 times longer continuous use age (versus GCA’s), before requiring maintenance [24]. Furthermore, as Miller elaborates, “… the mundane reality was that GCA didn’t listen to its customers, while Nikon did. Chip firms that interacted with GCA found it “arrogant” and “not responsive.” No one said that about its Japanese rivals”.

The reason for this phenomenal success of Japanese companies were multifold. First, the Japanese government provided incentives and subsidies to its chip and equipment making companies (although that was also true for the US government). Secondly, “Unlike in the U.S., where antitrust law discouraged chip firms from collaborating, the Japanese government pushed companies to work together”, writes Miller; given my zero understanding of anti-trust laws I found it easier to directly quote Miller here. Furthermore, to enable this collaboration, the Japanese government set up and funded a research consortium called the VLSI Program in 1976. Additionally, the cost of capital was low in Japan compared to Silicon valley. The Japanese banks, which had huge deposits – because of post-war economics, the Japanese society was structurally shaped to encourage massive savings – the Japanese companies had easier access to capital and for longer durations.

Worried, the chip makers in the US rushed to the American government requesting for support. After much back and forth, the US government took a few steps. It cut a deal with Japan, to limit the number of DRAM chips exported by Japan to the US, hoping it will encourage the domestic demand, which due to complex bunch of reasons did not work as expected. Additionally, the US government, just like Japan, set up a semiconductor consortium “Sematech”, funded half and half by industry and by the Pentagon. This intervention did help to some extent, but by this time the industry to so significantly reliant on many Japanese equipment, that replacing them again with American made systems was extremely difficult. So, by the end of 1990s, the Japanese firms were accounting for 50% of the whole world’s investment in chip manufacturing facilities and equipment

Part 4: America Resurgent

After the collapse of Intel’s market share in DRAM, they needed to find a new product, which came in the form of a standard-use chip called microprocessor as mentioned above. The market needed to commercialize this product came from IBM, who in 1981, introduced their personal computer with an Intel chip inside [25,26]. During the same time, a new company ‘Micron’, founded by the Parkinson brothers in 1978, entered the DRAM market. It was an unusual time to enter the memory business, when the market was dominated by Japanese companies and everyone else was closing down. Even Intel, who pioneered the DRAM chips, were getting out of DRAM manufacturing. However, one Jack Simplot, who became a billionaire through his innovations in the potato chip business and knew little about chip manufacturing, was insightful in observing how the Japanese competition had turned the DRAM chip business into a commodity market, and through his extensive experience knew that the best time to get into a commodity market was when the prices were low and everyone was closing their shops.

The entry of Intel into PC market and Micron into a depressed DRAM market were just the first steps. Both the companies were already diving deep into improving their manufacturing process yield, which was going to be the cornerstone in their eventual success. Andy Groove of Intel and Ward Parkinson of Micron, both were known to be obsessed with improving the manufacturing process efficiency. They were prepared to push hard to tweak the process knobs and squeeze out every bit of improvement possible, and every optimization was directed towards cost reduction. The location of Micron, in Boise, Idaho, helped too, as the land and electricity were cheaper than in California or Japan. The structural factors that initially sided with Japanese manufacturers now were moving. During mid 1980s, the Japanese Yen doubled in its value, making American exports cheaper. Additionally, the cost of capital also dropped in the US, helping both Intel and Micron. Compaq and other firms started launching PCs with Intel chips, expanding their microprocessor market. Although these additional factors were helpful, but the struggle was real. At one point, Micron’s cash balance was so low that it could only afford to pay for the next 2 weeks. Many of the employees recall working extremely hard with a “sweatshop mentality”.

A key part of American strategy was to find cheaper alternatives to its heavy reliance on Japanese firms for chips, which is where Samsung came into play. Lee Byung-chul founded Samsung (meaning “Three Stars”) in 1938, and initially traded in fish and vegetables, diversifying later into textiles, fertilizers, banking and insurance. By 1960, he became the richest person in South Korea. He wanted to get into semiconductor business but seeing the already intense competition in the chip industry made him reluctant to bet on it. After much thought, and the possibility of support from the South Korean government as well as banks, he took the bold step. To add to that, support also came from half a world away, from the silicon valley. The US collaborated with Samsung not only by providing a market for the product but also as technology transfer in the form of design licensing from Micron, and Intel selling Samsung manufactured chips under its own brand. Eventually, the South Korean firms overtook Japanese firms in DRAM manufacturing, resulting in a drop of Japanese market share from 90% in 1980s to 20% by 1998.

The success of Intel and Micron in combination with the Samsung collaboration helped revive the American chip industry. Additionally, the work of Lynn Conway [27] and Carver Mead also contributed the American resurgence by enabling digital methods for chip designing. Initially, the circuit designs were prepared by hands using a ruler, pencil, a red film and a knife. Lynn’s background in computer architecture along with Carver Mead, enabled them to draw up mathematical design rules that could be programmed and digitized. This in turn enabled circuit designers to focus only on circuit design optimization rather than also worrying about sketching it out themselves. Much later, only three American firms controlled about 75% of the circuit designing market – Cadence, Synopsis and Mentor. Additionally, the boundaries were pushed further by new startups like Qualcomm, which foresaw the emergence of powerful chips that could potentially support – then only theoretical – new communication methods (for example, Viterbi’s algorithms) for cramming more data into existing communication spectrum.

(Although this unit has 4 more chapters (20 pages) with interesting discussions, however, I think that main message of those chapters is in alignment with unit 1, so I will not repeat noting them here).

Unit 5: Integrated Circuits, Integrated World?

“At age 54, he was looking for a new challenge” writes Miller, referring to Morris Chang, who, in 1980s, was passed over the CEO position at Texas Instruments. His first trip to Taiwan in 1968 was as a Texas Instruments employee, but in 1985, the situation was different. He was hired by the Taiwanese government to head and set up a new semiconductor fabrication facility, Taiwan Semiconductor Manufacturing Company (TSMC). The Taiwanese government provided about half of the funding, while the remaining capital came from Philips and from wealthy Taiwanese businesses. Chang’s unique idea of contract-based manufacturing of chips (designed by customers) was initially pitched by him to his Texas Instruments’ colleagues (who subsequently rejected it), now found a new patch of land to grow at TSMC. Chang argued that with rise in the need for computing power, the requirement for chips will increase, and the companies that do not know how to manufacture can outsource that task to chip manufacturing experts. With increase in computational complexity, only those firms that can manufacture in bulk volume will be profitable. TSMC’s business exploded in the 1990s, and today, TSMC owns 68% of the advanced chip manufacturing capacity [28], thanks to Chang who foresaw the revolution.

Around the same time, in 1979, just across the Taiwan strait, things were very different; China had negligible semiconductor facilities and only 1500 computers in the entire country, according to Richard Baum [29] and National Research Council [30]. Miller writes, in 1960s, during the Cultural Revolution, many Chinese scientists were sent to the countryside farms and asked to study the proletarian politics than science; additionally ties with foreign technology were also severed. However, things started to change in the 1980s. With the ending of the Cultural Revolution, investment in semiconductor research increased, resulting in rise of business like Huawei. Initially, China progressed in electronics assembly, and it was not until the year 2000 that big investments to set up a chip manufacturing company like Semiconductor Manufacturing International Corporation (SMIC) happened. Interestingly, a significant capital for SMIC was provided by the US investors (up to 50%), according to estimates by Chase et. al. [31]. By 2005, the competition in the chip industry in East Asia was heating up, with SMIC in China, Chartered Semiconductor in Singapore, TSMC in Taiwan and Samsung in South Korea.

Going back in time to the 1990s, a few people were brainstorming about how the technology would progress in the next 2 to 3 decades. In order to make smaller and smaller features on chips, the selected wavelength of light to “print” patterns on silicon was critical. A smaller wavelength light can print smaller features. To further shrink the features or the number of transistors on a chip, newer technology like extreme ultraviolet (EUV) lithography was needed. Compared to the more commonly used 248 or 193 nm light, EUV was 13.5 nm in wavelength. Since the cost of developing EUV lithography was going to be enormous, most lithography firms did not dare to take a step forward. However, Intel took the initiative to push for EUV development, spending a significant capital to enable partnerships with National Labs and setting up EUV consortiums. The answer to the question of who would develop such a technology was also becoming clearer. The US had no lithography firms that were technologically advanced enough to work on EUV as the market was mainly captured by Nikon and Cannon. A partnership with them was difficult due to US-Japan politics. This left only one lithography firm, ASML in the Netherlands.

Fast forward to the mid-2000s, when the PC business was booming, Intel was shining due to its hold on the PC microprocessor market. It’s processors used x86-architecture which they still used till today, not because it is the best but because the first IBM PC happened to use it, and it was just carried forward. In the 1990s, Intel did consider to switch to RISC architecture which was simpler and energy efficient, but the cost of change to simply too high. In the meantime, new device markets were emerging such as smartphones, but expanding into them seemed like a gamble to Intel given that they were performing very well in the PC business. So, Intel declined the contract to make chips for the iPhone – big mistake. Eventually, Arm, a British semiconductor firm who used the energy efficient RISC architecture for their chips got the contract with Apple. Later, Intel CEO Otellini recounted this mistake in an interview as mentioned below.

“They (Apple) wanted to pay a certain price, and not a nickel more. . . . I couldn’t see it. It wasn’t one of these thing you can make up on volume. And in hindsight, the forecasted cost was wrong and the volume was 100X what anyone thought” Intel CEO Otellini to journalist Alexis Madrigal [32].

Unit 6: Offshoring Innovation?

Can a semiconductor company work without a owning a fab? AMD’s Jerry Sanders would have said a straight no. Early in the semiconductor world, owning a fab was not only advantageous but practically mandatory. However, as the technology advanced, owning a fab became more and more expensive and today, starting a new advanced fab could cost about $20 billion. As the competition became extremely cutthroat in the memory industry (DRAM and NAND) many companies sold off their fabs, with only a few companies surviving the ordeal like SK Hynix, Samsung, Western Digital and Micron. More recently, this has also been the case for fabs manufacturing logic chips like Intel. This is where TSMC business model stood out. Morris Chang’s bet on contract based bulk chip manufacturing helped reduce cost and outcompete many fabs.

In the 1990s, many companies started moving towards the fabless model, resulting in (1) fab owning companies selling off their fabs to focus on chip designing (e.g., AMD), and (2) opportunities for new startups not worry about chip manufacturing and only work on product innovation and software engineering (e.g., Apple, Google etc). The physical act of manufacturing the chips could be delegated to giants like Intel, Samsung and TSMC. The US, particularly California, the Silicon Valley, benefitted hugely from this fabless model, as the innovation could happen in tech without worrying about expensive in-house chip manufacturing.

Around the same time, a new player in chip manufacturing joined the chip manufacturing industry. When AMD sold off its fabs, they were purchased by an investment arm of the Abu Dhabi government, enter GlobalFoundries. The business model of GlobalFoundries was similar to that of TSMC, contract based chip manufacturing. As the world was passing through the 2008 financial crisis, the semiconductor market slumped as people stopped purchasing electronics. This threatened the fabs and they went into cost cutting mode. Morris Chang of TSMC had been through enough ups and downs of the industry to have the insight that investing in the downturn will later help them grab the market share when the crisis is over. This was also due to his vision on how the mobile devices will change the digital landscape. Chang fired Rick Tsai, his hand-picked CEO of TSMC and took the charge himself [33, 34]. Around 2009, he announced multi-billion dollar investments, increased the R&D spending and re-hired the fired workers.

While the industry was going through this financial and political turmoil, a new technological innovation was about to enter the landscape, EUV lithography. ASML, through Intel’s investment 1980s, and later also by many other chip manufacturers, had been working on making EUV lithography work, a task deemed impossible by many [35]. Producing EUV light is an extremely difficult task. The most efficient method currently used in ASML’s EUV machines involves shooting tiny tin balls (millionth of a meter in size) that are moving at the speed of 300 km/h with a laser twice. The first shot to warm it up and the second to pulverize it into plasma reaching temperatures of about 0.5 million ˚C (hotter than the surface of the Sun), and repeating this 50,000 times per second. This results in EUV light production, enough to fabricate chips. To enable each of these steps, ASML had to collaborate with many different companies who were highly specialized, for e.g., Cymer for light source, Trumpf for laser engineering, Zeiss for optics, and many more. There was no plan B, and ASML had to push hard it’s suppliers to deliver the innovations in a timely manner. The final EUV lithography machine by ASML was tool with thousands of components coming from multiple suppliers and funded by many chip manufacturers. Each EUV lithography tool costs over $100 million, a big investment as each fab employs multiple of such tools for high volume chip production.

“If you don’t behave, we’re going to buy you,” ASML’s CEO Peter Wennink told one supplier [36]. It wasn’t a joke: ASML ended up buying several suppliers, including Cymer, after concluding it could better manage them itself.

TSMC, Intel, Samsung and GlobalFoundries, all were certain that EUV will be the way forward, but each had different strategies on how to integrate EUV into their manufacturing process. In 2018, after purchasing multiple EUV machines, GlobalFoundries’ executives ordered to scrap the EUV program [37], as it would still cost about $1 billion more to get EUV going, so, instead they would stop the research for the production of the most advanced and the smallest transistors and stay a medium-sized fab. Intel, who had been funding EUV development since 1990s, and had spent over $10 billion a year throughout 2010s, delayed its adoption of the EUV machine [38]. Miller writes, “By 2020, half of all the EUV lithography tools, funded and nurtured by Intel, were installed at TSMC. By contrast, Intel had only barely begun to use EUV in its manufacturing process” [39].

Unit 7: China’s Challenge + Unit 8: The Chip Choke

The last 2 units explain a great deal about the contemporary politics of semiconductors, regarding the emergence of China in the semiconductor landscape, not in its traditional role as a hub for electronic assembly (which it held for previous decades) but as a competitor in chip manufacturing as well as a huge consumer market for existing players in the semiconductor world.

“One U.S. semiconductor executive wryly summed things up to a White House official: “Our fundamental problem is that our number one customer is our number one competitor. [40]”

Miller also talks about the scenarios where politics entangles with technology and discusses a few cases of technology transfer where such intermixing of politics and tech becomes obvious and how it impacts the field. He talks about Huawei, 5G, and technological advancement as a way of strengthening national defense. He also talks about how extremely vulnerable our current globalized semiconductor world is, whether it be risks due to politics or simply natural phenomenon like earthquakes impacting chip production and hence our everyday life.

I am choosing to not write about the last 2 units for a few reasons. Firstly, although I learned a lot by reading those chapters, I feel that the topic of the political and technological interaction is extremely complex and that by just reading about 100 pages on this I am only scratching the surface; so writing about it or even just summarizing would not be a good idea at the moment. I need to read more books. Secondly, Chip War did what a good book must, especially the last 2 units, it raised more questions for me than it answered. Given my educational background only in technical and scientific fields, I am currently not well equipped to quickly looks for those answers dealing with politics and economics of the semiconductor industry.

So, I would like to invite proposals from the readers of my blog (if any), for books that I must read to improve my understanding of the basics of politics and economics, in general or specific to semiconductor industry.

My few thoughts

As mentioned before, this book was a must read for me as I am working in the semiconductor industry and on one of the hottest topics (technically and politically), EUV lithography. I really enjoyed reading the book and the nerd in me loved the many many semiconductor industry anecdotes that Miller wrote. I would strongly recommend this book to anyone in the field.

I typically go over the Contents /index page, to get an idea what is in each chapter, so commenting on the style of the Contents/Index and the way of naming the units and sub-units/chapters, I am a bit torn. I can understand why Miller probably choose the way he created the Contents page with catchy names for units and chapters (to grab the attention of the audience), but if I had to guess the contents of the units and chapters just by reading the Contents page, it would be impossible. My personal preference would be more old school traditional way of writing a Contents page.

I liked that he included a couple of pages on who are the main characters in the field and also a glossary to get everyone up to speed on some technical words. I also liked that each chapter is a short quick read of not more than 10 pages, which helped me in retaining a lot of info chapter after chapter and it was also easy to take breaks. Since it is non-fiction, referencing was critical, and Miller does include many references to the mentioned facts and figures, however the choice of the unconventional style of referencing – no citation numbers but instead using page numbers and phrases – felt not very comfortable to follow. Additionally, references to many critical and sometimes alarming facts were missing that should have been cited in the book, as it helps in winning credibility to the arguments that the author is making.

On the content, I think it is rich with information and it never felt like a chore while reading. The language used is simple and easy to understand, with some technical jargon. Since I was already familiar with the technical words, it was a smooth read for me, but I am curious about how would it be received by someone who is not in semiconductors. The direction of thoughts and arguments throughout the book are primarily from an American point of view, which Miller makes very clear already in the introduction, when he talks about the the American destroyer USS Mustin passing through the Taiwan Strait. Although there are many discussions about the industry in the east and south-east Asia, those discussions are still about their impact on American semiconductor industry. Little was discussed about the semiconductor landscape in Europe, except for the Dutch company ASML. This choice to discuss such a vast industry from one point of view is understandable, as it could be quite difficult and perhaps also confusing for readers when multiple perspectives are included in discussing such complex interactions.

Overall, I thoroughly enjoyed reading Chip War, and would like to congratulate Chris Miller on this very nice and successful publication.

References (* marked ones are as mentioned in the book)

1. JP Morgan: Chip Shortage
2. Statista: Chip Shortage
3. Statista: Semiconductor supply chain
4. Forbes: Chip Shortage
5. Financial Times chip infographic
6. Apple A14
7. Invention of integrated circuits
8. Moore’s Law
9. Integrated Circuits
10. Chips in Apollo mission
11. Chips in Minuteman II
12. *William F. Finan, “The International Transfer of Semiconductor Technology Through U.S.-Based Firms,” NBER Working Paper no. 118, December 1975, pp. 61-62.
13. *Kenneth Flamm, “Internationalization in the Semiconductor Industry,” in Joseph Grunwald and Kenneth Flamm, eds., The Global Factory: Foreign Assembly in International Trade (Brookings Institution, 1985), p. 70
14. *Bundo Yamada, “Internationalization Strategies of Japanese Electronics Companies: Implications for Asian Newly Industrializing Economies (NIEs),” OECD Development Centre, October 1990, https:// web-archive. oecd.org/2012-06-15 /167646-33750058.pdf.
15. *Kenneth Flamm, “Internationalization in the Semiconductor Industry,” in Grunwald and Flamm, The Global Factory, p. 110
16. *Lim and Pang Eng Fong, Trade, Employment and Industrialisation in Singapore, p. 156
17. *Hong Kong Annual Digest of Statistics (Census and Statistics Department, 1984), table 3.12, https:// www. censtatd. gov.hk/en/data /stat_report/product/B1010003/att/ B10100031984AN84E0100.pdf
18. *G. T. Harris and Tai Shzee Yew, “Unemployment Trends in Peninsular Malaysia During the 1970s,” ASEAN Economic Bulletin 2, No. 2 (November 1985): 118-132
19. DRAM by Intel
20. DRAM (Wikipedia)
21. *Choi, Manufacturing Knowledge in Transit, pp. 191-192.
22. *”Marketing and Export: Status of Electronics Business,” Electronics, May 27, 1960, p. 95.
23. *Rosen Electronics Newsletter, March 31, 1980
24. *Henderson, “The Failure of Established Firms in the Face of Technical Change,” pp. 220-222, 227; interview with former AMD executive, 2021
25. *Gerry Parker, “Intel’s IBM PC Design Win,” Gerry Parker’s Word Press Blog, July 20, 2014, https://gerrythetravelhund wordpress.com/tag/ibm-pc/
26. *Jimmy Maher, “The Complete History of the IBM PC, Part One: The Deal of the Century,” ars TECHNICA, June 30, 2017, https:// arstechnica.com/gadgets/2017/06/ibm-pc-history-part-1/
27. *Miller’s interview with Lynn Conway, 2021, where she surprised me by wanting to discuss the nuances of John Gaddis, The Landscape of History (Oxford University Press, 2004).
28. TSMC’s 68% share in advanced chip tech manufacturing
29. *Baum, “DOS ex Machina,” pp. 347-348
30. *National Research Council, “Solid State Physics in the People’s Republic of China,” pp. 52-53
31. Chase et al., “Shanghaied,” p. 149
32. Alexis C. Madrigal, “Paul Otellini’s Intel”
33. *Lisa Wang, “TSMC Reshuffle Stuns Analysts,” Taipei Times, June 12, 2009
34. *Yin-chuen Wu and Jimmy Hsiung, “I’m Willing to Start from Scratch,” Common Wealth, June 18, 2009.
35. *Miller’s interview with John Taylor, 2021.
36. *Miller’s interview with executive at ASML supplier, 2021.
37. *Miller’s interviews with three former Global- Foundries executives, one of whom focused on EUV, 2021; on R&D spending, see GlobalFoundries’ IPO prospectus, Security and Exchange Commission, October 4, 2021, p. 81, https://www.sec. gov /Archives/edgar/data/ 0001709048/00011931 2521290644/ d192411df1.htm. See also Mark Gilbert, “Q4 Hiring Remains Strong Outlook for Q1 2019,” SemiWiki, November 4, 2018, https://semiwiki. com/semiconductor-manufacturers /globalfoundries/7749-globalfoundries- pivot-explained/q.
38. *Miller’s interview with Pat Gelsinger, Bloomberg, January 19, 2021, https://www.bloomberg. com/news/videos/ 2022-01-19/intel -ceo-gelsinger-on-year-ahead-for -global-business-video.
39. *Ian Cutress, “TSMC: We Have 50% of All EUV Installations, 60% Wafer Capacity,” AnandTech, August 27, 2020.
40. *Miller’s interview with a former senior administration official, 2021.