Chip War: The Fight for the World’s Most Critical Technology
The chip supply chain is so complex that for all the talk of offshored manufacturing and dependence on China, the U.S. and her client states control critical choke points and will continue to do so.
I’m usually wary of new releases covering the day’s hot-button issues, but Chip War is a highly readable and informative history about one of the most complicated and expensive technologies humanity has ever built. Miller traces the history of the microprocessor from the early Silicon Valley days of the Traitorous Eight’s Fairchild Semiconductors to China’s SMIC. In contrast to the title, the book left me with the feeling that things are not as dire for the US as the media presents, and this is not because Miller is some naïve neoliberal or a kumbaya globalist. Rather, the chip supply chain is so complex that for all the talk of offshored manufacturing and dependence on China, the US and her client states control critical choke points and will continue to do so for the foreseeable future.
“Few [countries] truly have full sovereignty, as most contract out their defense to the US or (nowadays) China” writes Balaji Srinivasin in The Network State. Indeed, the US as Global Policeman has many downsides, but an upside is we de facto control significant portions of the chip supply chain via Taiwan, South Korea, and the Netherlands. Taiwan’s TSMC and South Korea’s Samsung fabricate almost all of the most advanced logic chips (the remainder are produced by Intel). The advanced processors require extreme ultraviolet lithography (EUV) machines produced by only one company, the Netherlands’ ASML, and the EUV’s irreplaceable light sources are manufactured in San Diego.
The US has a significant contribution besides Intel: “nearly every chip in the world uses software from at least one of three US-based companies: Cadence, Synopsys, and Mentor (pg 315).” If there was a question of if these companies would fall into line when faced with US allegiance or profits, it was answered when TSMC enthusiastically supported the US’s 2019 crusade to ban Huawei, despite Huawei being TSMC’s second-largest customer.
Miller doesn’t paint an overly rosy picture, and the long tail of possible risks is very, very bad: the invasion of Taiwan and the TSMC takeover, if successful, would disrupt the global economy by eliminating access to the most advanced chips. And no, this is not the good kind of Unicorn-led, Silicon Valley disruption. What’s worse is China would potentially still receive what it needed from TSMC. I say “potentially” because we have to imagine there would be some drop in production from recalcitrant Taiwanese who would sooner give up their life (or at least craftily introduce production delays) than work for China.
The less extreme scenarios are also bad for the US. While China doesn’t produce the most advanced chips, it does have 15% of the fabrication market, mostly for low-grade chips, and this is projected to increase to 24% by 2030. We rely on low-grade chips for everything from cars to coffeemakers. But we can at least rest assured that China is stuck in a supply chain of “weaponized interdependence” for advanced chips. “Establishing a cutting-edge, all-domestic supply chain would take over a decade and cost well over a trillion dollars in that period (pg 323).” The lithography machines alone cost $150mm. And it’s not just about having the money, but also the expertise to natively support such an advanced supply chain.
One of the biggest questions I had going into this book was how Taiwan birthed TSMC, one of the most valuable companies today, and, short of having nukes, the best deterrent to a Chinese invasion. The answer is Operation Paperclip — only there were no Nazis, and instead of Werner von Braun, Taiwan got Morris Chang. He moved to Taiwan in 1987 to start TSMC after over twenty years at Texas Instruments, where he was passed over for CEO. He brought key American-trained execs along with him, including from TI, Motorola, and Intel.
Transplanting this know-how into Taiwan proved to be a key differentiator as to why TSMC is in Taiwan and not Hong Kong, Singapore, South Korea, or Malaysia, which were all investing heavily in advanced memory chips. (China was a competitive threat with the low wages it offered for chip assembly, but the aftermath of the Cultural Revolution proved a real hammer to the knees). Chang going to Taiwan to start TSMC and explicitly bringing US know-how and customers with him is an interesting quasi-controlled natural experiment where you could feasibly imagine him starting TSMC in a Singapore or a Hong Kong. This is also what makes the recent Biden ban on American execs working for Chinese chip companies so powerful — China potentially just lost access to the next Morris Chang.
Chip War provides key nuance and critical historical information entirely lost in media coverage of this issue over the last two years. During peak Covid in 2021, there was a chip shortage, yes, but it was also the year with the most semiconductors ever produced (1.1 trillion with a 13% increase from 2020). Miller christens the shortage “mostly a story of demand growth rather than supply issues.” For example, the automotive shortage was caused by car companies cutting chip orders in the early days of the pandemic and then unsuccessfully trying to claw back chip capacity that had been reallocated to other customers.
In finance, a “poison pill” is a defensive strategy used by public companies to prevent activist investors from a takeover. The takeover becomes self-defeating by diluting the activist’s stake in the company, therefore making it harder to extract outsized value. Is it too optimistic to view TSMC as Taiwan’s poison pill? Surely China’s dependency on TSMC would give it pause before an invasion, and, as I speculated above, we don’t know how production would be impacted by an invasion, even if it were a successful one.
Here’s to hoping advanced fabrication plants are the new nukes.