Chip War: The Fight for the World’s Most Critical Technology – Chris Miller (2022)
It’s always refreshing to read something that opens up your eyes and teaches you about something you hadn’t stopped to consider for a long time. Chip War, the new book by Chris Miller (a professor of economic history at Tufts University) is absolutely one of those books. Miller’s subject is the microchip industry and how it forms the core of the growing strategic rivalry (emnity?) between the US and China.
The book begins by outlining the history of the semiconductor industry, taking up the tale during the early WW2 days of unreliable, bulky and expensive vacuum-tube based computing. The Bell Labs physicist William Shockley provided the breakthrough by co-inventing the transistor (with Bardeen and Brattain) and leaving Bell to start Shockley Semiconductor with a few of his protegés.
As it turns out, Shockley was an arrogant and tyrannical boss and was unable to transmute his intellectual genius into a successful business. His abrasive leadership style led to the ‘Traitorous 8’ (including future industry giants such as Gordon Moore, Robert Noyce and Eugene Kleiner) to split off and start Fairchild Semiconductor which pioneered the integrated circuit (IC), along with Texas Instruments (TI).
Fairchild and TI were the leaders in the early semiconductor industry and sold mainly to the US military, which was then in the grips of the Cold War arms race and was keen to use microelectronics and computing power to give it the technological edge over the Soviet Union which dominated in terms of men and materiel.
The book traces the early days of Silicon Valley and the development of the stock options based, mercenary culture that still dominates. It follows the transition from the early military-dominated tech industry to the consumer focused industry we are familiar with today. Miller explains how this was driven by the rise of Intel and personal computing, as well as Japan’s Sony which brought electronics to the masses.
For me, the most remarkable revelation in this book is the rise of the giant foundry of TSMC (Taiwan Semiconductor Manufacturing Corporation) and its utterly central role in the modern digital economy. In the early days of semiconductors, Taiwan was unable to make a large dent in the market due to the dominance of Japan, and later South Korea, in the DRAM market. Nonetheless, expanding the semiconductor remained a key priority of the Taiwanese government and when, in the 1980s, Morris Chang, the microchip pioneer and leader of TI’s chip business, was passed over for promotion to CEO, the Taiwanese government gave him a blank cheque to establish a semi conductor foundry on the island.
The approach taken by Chang at his new Taiwan Semiconductor Manufacturing Company (I guess they didn’t waste too much energy thinking of a name), was radical at the time. TSMC would dispense with designing any of its own chips and would focus exclusively on being a foundry (or fab) which would manufacture chips on behalf of others. Focusing solely on chip manufacturing was considered a radical move and was doubted by many at the time, but Chang stuck to his guns and was proven right a million-fold since then.
TSMC’s exclusive focus on manufacturing the latest chips has allowed it to benefit from economies of scale in manufacturing which generate the revenues needed to power the ongoing gigantic CAPEX required to stay at the forefront of chip manufacturing and further drive down costs. TSMC’s promise to stay out of chip design also means that it is a trusted partner of firms who have faith that TSMC will not steal their designs.
TSMC’s business model allowed all manner of new innovative firms to develop new categories of chip, without needing to invest heavily in manufacturing capacity. This allowed businesses such as Qualcomm (focused on advanced mobile spectrum chips) and NVIDIA (chips for advanced graphics and AI) to develop new categories and drive innovation with just their ideas (and a few million dollars) by outsourcing production to TSMC and other foundries instead of tens or hundreds of millions of dollars in developing their own capacity.
The most famous customer of TSMC’s foundry is of course Apple, which chose to outsource production of the chips used in iPhones and Macs to TSMC and Samsung. (Miller tells an ironic anecdote about how Steve Jobs first asked Intel’s CEO if Intel would be interested in making the chips for the iPhone – Intel rejected the opportunity because they didnt think the iPhone would take off…). Intel, dominant in the PC market, did make the chips for the MacBook until 2020 but this has been replaced with the TSMC-manufactured M series of Apple chips.
The ongoing investment cost required to keep up to date with the latest manufacturing processes and EUV (extreme ultraviolet) technology is now out of reach of all companies and governments except for TSMC, Samsung and, to a lesser degree, Intel. The latest generation of advanced lithography machines (from the Dutch company ASML) cost over $300 million each and upgrading a single fab to the latest 3nm process costs over $20 billion. Even the deepest pockets in the world (the US Pentagon or Chinese government) cannot afford such spending, particularly given the fact that the military now uses less than 3% of all chips produced today. As a result, the US government cannot simply dictate that the chip production is onshored again as they lack the heft in the world market, as well as the financial and technical wherewithal to make that happen.
Both the Chinese and the US are dependent on complex, international supply chains with no easy way route to strategic independence. The position of the world’s most advanced foundry on the west coast of Taiwan, within easy reach of Chinese missiles and naval blockade is a profound source of worry for all parties who are dependent on this indispensable commodity.
China too, is deeply dependent on imported chips for its huge consumer electronics and defence industries. Each year, China spends more on importing foreign microchips than oil and as such is extremely sensitive to being cut off from global supply chains (which are dominated by the US). Thus far, China has built up its capacities through a mixture of espionage, coercion and legitimate technology transfer from foreign companies eager to gain access to China’s vast internal market. Nonetheless, the US remains dominant over many of the critical chokepoints in the semiconductor supply chain and could easily cut off China’s semiconductor industry in the event of a conflict over Taiwan. Recent US administrations, under both Trump and Biden, have taken a much more muscular, nationalistic approach to competition with China, and semiconductors in general and have taken aggressive steps to cut off the growth of China’s independent semiconductor industry.
The book also brings home just how much the world has shifted away from the free trade consensus that was the dominant narrative from the ’90s to the early ’10s. Back then, it was considered common knowledge among respectable economists that it didn’t matter where an item was manufactured, as the consumer will ultimately benefit from lower costs, even if an advanced economy loses out on manufacturing jobs. In retrospect this was obviously nonsense as the economy (which is made of real people doing real things) cannot be reduced to dollar values. Leaving aside the obvious social harms of unemployment and deskilling that go along with deindustrialisation, expertise and human capital matter. A dollar (or euro) from manufacturing is not equivalent to a dollar from finance, and a country which retreats from the leading edge of technology will find it difficult to regain that edge again. This view now animates global (or at least US) thinking on trade, where a renewed focus on strategic resources and industries now dominates at the expense of pure economic efficiency maximisation.
All that aside, it is tremendously refreshing to read a book on the ‘tech’ industry that reminds you that there is more to it than software. It feels timely to read such a book as the 10-year nonsense bubble of ‘Silicon Valley’ tech has just finished popping. The decade of silly, money losing, tech-enabled startups grabbing insane valuations and stealing the oxygen of headlines had blinded us to the real work going on underneath that supports our digital world. In retrospect, the NFT/Metaverse silliness of last year should have warned us all that a crash was coming – the world cannot sustain such a level of stupidity for an extended period of time.
2022 feels like a year where our attention is re-focused on the real substrate that allows our society to function, in particular on energy, without which everything else is impossible. This book feels like part of that refocusing on the essentials. As a European however, I cannot help but notice the commonality that Europe will remain just as dependent on others for our semiconductors as we are for our gas.
