Chip War

The story in brief

Chris Miller is an economic historian at Tufts University’s Fletcher School. Chip War, published in October 2022, tells the history of the semiconductor from the invention of the transistor at Bell Labs in 1947 to the escalating contest between the United States and China in the early 2020s. It won the Financial Times Business Book of the Year award for 2022.

The story moves through several eras. In the 1950s and 1960s, American companies such as Texas Instruments and Fairchild Semiconductor invented the integrated circuit, and the US military and space programme were their first major customers. In the 1970s and 1980s, chips moved into consumer electronics, and Japanese manufacturers, especially in memory chips, came to rival and then overtake American firms. From the late 1980s, production began to concentrate in East Asia, culminating in the dominance of the Taiwan Semiconductor Manufacturing Company. And in the 2010s and 2020s, as China invested heavily in its own industry, chips became a central issue in the rivalry between the United States and China.

The decisions that made it

American chip firms decided early to move assembly, the labour-intensive stage of production, to Asia. Fairchild opened an assembly plant in Hong Kong in the early 1960s, and others followed to Taiwan, South Korea, Singapore and Malaysia. The decision lowered costs and, Miller argues, also served US foreign policy by tying Asian allies to American industry. It also began the shift of production capacity across the Pacific.

In the 1980s Japanese firms captured much of the memory chip market, through high quality, heavy investment and cheap capital. Intel, under Andy Grove, decided to leave memory chips and concentrate on microprocessors, a decision that preserved the company and made the personal computer era an Intel era.

In 1985 Taiwan’s government recruited Morris Chang, a senior executive who had been passed over for the top job at Texas Instruments, to lead its industrial research. In 1987 he founded TSMC on an untested model: a foundry that would manufacture chips designed by other companies, and never design its own. The model removed the need for chip designers to own factories, enabling a new generation of “fabless” firms such as Qualcomm, Nvidia and later Apple’s chip business. Over three decades TSMC became the leading manufacturer of the most advanced chips in the world.

Lithography, the process of printing circuits onto silicon, became the narrowest point in the supply chain. After decades of research, much of it funded by American companies and government, only one company, the Dutch firm ASML, succeeded in making extreme ultraviolet lithography machines, which are required for the most advanced chips. Each machine contains components from suppliers across several countries.

Intel, which had led manufacturing for decades, fell behind in the 2010s, as its transition to new manufacturing processes was delayed and it declined an opportunity to make chips for Apple’s first iPhone. TSMC and Samsung moved ahead.

Finally, from 2019 the US government used its control of chip technology as a weapon, cutting the Chinese telecoms company Huawei off from the advanced chips TSMC had been making for it. Miller presents the move as the first large use of the supply chain’s narrowest points as a source of power.

What the book sees in them

Miller’s central argument is that chips have become what oil was in the twentieth century: a resource on which economic and military power depends, whose supply is concentrated in a few places. But chips are more concentrated than oil ever was. A single company in Taiwan makes most of the world’s most advanced processors, and a single company in the Netherlands makes the machines they require.

He also shows how much of the industry’s history was shaped by government, through defence contracts, research funding, industrial policy in Japan, Korea and Taiwan, and export controls. The image of chips as a pure triumph of private enterprise is incomplete.

And he argues that the concentration was a consequence of economic logic, not a plan. Each decision to specialise, outsource or concentrate made sense for the companies involved. Together they produced a system of remarkable efficiency and remarkable fragility. The site’s idea page Path Dependence and Lock-In explores how small early choices can harden into structures that are hard to change.

Since it was published

The book’s subject has moved quickly since 2022. Days after it appeared, on 7 October 2022, the US government imposed broad export controls on the sale of advanced chips and chip-making equipment to China, and pressed allies to follow. The CHIPS and Science Act, signed that August, began funding new factories in the United States, including TSMC’s plants in Arizona. The boom in artificial intelligence after the release of ChatGPT in late 2022 made Nvidia’s chips, manufactured by TSMC, among the most sought-after products in the world, and turned access to them into a question of national strategy. The United States tightened its export controls further in 2023 and 2024, and debates continue about their effectiveness, including after Chinese developers released capable AI models trained with restricted hardware. The site’s essay The Universal Machine and the End of the Free Lunch follows the computing side of this story.

What it asks of someone deciding

The book asks where your critical dependencies lie, and how many of them pass through a single supplier or place. Efficiency and resilience pull against each other, and most organisations do not notice how much resilience they have given up until something breaks.

It asks what a specialisation decision commits you to. TSMC’s choice to manufacture only, and Intel’s choice to stay vertically integrated, shaped both companies for decades. Choices about what not to do are as consequential as choices about what to do.

And it asks what happens when an industry becomes strategically important. Companies that once sold to everyone find their markets shaped by governments. Leaders in technology now need to understand geopolitics as part of strategy.