Revive Taiwan's 'Silicon Shield'
Speaking at the Shangri-La Dialogue in Singapore in June, Japanese prime minister Fumio Kishida expressed what many have come to fear: “Ukraine today may be east Asia tomorrow.”
“East Asia” is euphemistic for a Chinese invasion of Taiwan, where the latest polling indicates that Taiwanese no longer believe the U.S. would fight a war with China on behalf of Taiwan, and only a third of Taiwanese people believe Taiwan has made adequate preparations to defend against a Chinese invasion.
Given China’s rising military power and longstanding ambitions to “reunify” Taiwan to the mainland, such fears are not unreasonable. President Biden’s recent remarks in Japan on American resolve to take military action were China to invade Taiwan, later clarified by aides to emphasize acts short of war, illustrates Taiwan’s need for an autonomous deterrent capability.
Taiwan has something Ukraine did not have: a “Silicon Shield.” The shield, at the center of which is Taiwan Semiconductor Manufacturing Company (TSMC), which produces 90% of the world’s leading-edge semiconductors including the latest M2 Apple chip, is intended to function like oil did for Kuwait in 1990: as something so important to the global economy that the United States would be compelled to protect it.
Comments last month from a leading economist and government advisor in China, Chen Wenling, however, indicate that the Silicon Shield is becoming a Silicon Magnet that attracts Chinese ambitions — as Kuwait’s oil attracted Saddam’s — rather than repels them. Speaking at a forum at Renmin University, Chen stated that the need for China to “seize TSMC,” a key to “the reconstruction of the industrial chain and supply chain,” is growing as the company builds semiconductor fabrication plants (“fabs”) in the United States. Chen feared that the transfer of leading-edge technologies to the U.S. would position the United States to more effectively sanction China, like it has Russia.
Although TSMC is building a fab in Arizona, it does not appear to have any plans to systematically diversify its leading-edge production or research and development facilities. But Ms. Chen is surely right that it would not be in China’s interests were TSMC to geographically diversify. Onshoring TSMC research facilities and leading-edge fabs in Japan, Europe, and most importantly the United States should be seen as a national security imperative not just for each of these countries, but for Taiwan as well. But to the contrary, with the announcement of multiple new $10 billion fabs in Taiwan, TMSC and Taiwan are doubling down on their current strategy.
Ms. Chen’s comments assume that the PRC could successfully invade Taiwan without TSMC and the wider semiconductor industry being destroyed, and that thereby the PRC could gain a near-monopoly on the cutting-edge semiconductor fabrication and research capability that would position China to achieve the leap-frog development its grand strategy calls for.
Turning the Silicon Magnet back into a Silicon Shield requires two coordinated efforts. First, Taiwan must convince itself and China that an invasion would likely severely damage or destroy TSMC’s physical and intellectual capital, meaning the Chinese mainland would gain nothing in this regard and Taiwan would experience an existential destruction of its economic engine. Second, to lower the attractiveness of its Silicon Magnet, Taiwan should initiate an intentional strategy to move elements of TSMC’s key intellectual asset, its research and development capability, out of Taiwan. What this means is that if China invaded, TSMC’s leading-edge capabilities could still operate — but not in service to China, but to China’s rivals in Japan, Europe, and the United States (as well as TSMC shareholders). Rebuilding Taiwan’s extensive semiconductor capacity without western inputs would be a very time consuming and costly affair. Even a “successful” Chinese invasion or coercion of Taiwan, hence, would relegate China to the backwaters of technological development, not hurdle it into the front ranks.
It is sometimes said that economic costs would already be “baked in” to any PRC assessment of the relative benefits of invasion compared with the likely costs, and so the semiconductor dynamic should be seen as not subject to manipulation as part of an overall tailored deterrence package. The responses to Russia’s invasion of Ukraine, and especially the freezing of its international currency reserves, do lend credibility to prospects of economic mutually assured destruction were similar techniques to be applied to China. However, much deeper entanglement between China and the rest of the world ensures that any economic harm would be widely distributed, and it is surely plausible that Chinese leaders might bet that deep entanglement means a Chinese invasion would not be treated in an equivalent manner.
In strategic discourse, Taiwan’s semiconductor manufacturing capabilities have been treated as something that China desires to take (viz. Ms. Chen’s comments) or as something it might be denied from taking via accidental or intentional destruction, which would also deprive the world of TSMC’s technological leadership. The third option — that the PRC would then face the prospect of losing leading-node production while all other major actors maintained access to these overseas capabilities — would be the worst possible outcome for the PRC. Credibly threatening such an outcome would contribute additional heft to the “no invasion” side of the cost-benefit equation.
TSMC plans to commence operations at its Arizona fab in 2024 with technology it was delivering last year (5 nanometer chips). The fab is small, representing only about 2.5% of TSMC capacity. This effort does not create much useful capacity for the US, which represents 65% of TSMC revenue, if something were to happen to interrupt production in Taiwan.
TSMC’s shareholders should be concerned with their dependence on the one geographical location called “the most dangerous place on Earth” by The Economist. Not building leading-edge capacity in the U.S. ignores the risk that as Intel builds its foundry business and Samsung expands U.S. capacity, TSMC’s customers may seek to lower their own risk of Taiwan dependence and shift to TSMC’s competitors. Chinese leaders have repeatedly expressed a goal of “reunifying” with Taiwan by 2049. Every year closer to that mid-century date comes with significant additional risk for companies dependent on the Taiwan fabs.
If TSMC were to increasingly diversify its operations, would this create an additional short-term incentive for the PRC to invade Taiwan? Quite possibly. However, sooner is better than later, as the relative military balance is likely to continue to grow in China’s favor as the PLA marches towards its 2035 modernization goals. As the military balance shifts in one direction, having a countervailing reinvigorated Silicon Shield can help maintain the deterrence equilibrium.
If the U.S. is going to attract TSMC to onshore R&D and a significant amount of leading-edge capacity, it will likely need to supplement the strategic argument with economic incentives. Today, tax benefits, strong workforce training, and a favorable regulatory environment are facilitating the building of over 40 baseball stadiums worth of new semiconductor factories in Taiwan. Estimates suggest that 40-70% of cost advantages for production in Asia are generated by government support. The U.S. Government will have to provide significant financial, workforce, tax and ESG incentives if it is going to attract substantial onshoring. These are over and above existing CHIPS Act proposals.
Onshoring TSMC research and development to the U.S. is the right way to recast Taiwan’s Silicon Shield as a deterrent — a way to make the relative costs of a Chinese invasion so imposing that the question is judged by China’s leaders to be better deferred to another day. At the same time, it would make American and allied economies more resilient in an era of growing uncertainty and deglobalization. Scholars, policymakers, and ordinary people alike recognize that deterrence in the Taiwan Straits is decaying. The decay can be reversed in the short to medium term by putting semiconductors at the center of the equation.
Dr. Jared Morgan McKinney is an assistant professor of international security studies at Air War College. Mark Rosenblatt runs Rationalwave Capital Partners, a technology focused investment fund. Opinions, conclusions, and recommendations expressed or implied within are solely those of the authors and do not necessarily represent the views of the Air University, the United States Air Force, the Department of Defense, or any other US government agency.
