In 1990, the US produced 37% of the world’s semiconductors and Europe 44%. Now, those numbers are down to 12 and 9%, respectively, and in addition Taiwan has a near-monopoly on advanced node logic chips, of which it produces 92% of global supply (Financial Times). This dominance came about because TSMC’s early strategy was to produce chips at a loss in order to capture market share, banking on later using economies of scale to maintain profit margins at prices below what manufacturers in any other country were capable of.
Why do semiconductor chips matter?
These advanced chips are necessary for supercomputer development, AI research, quantum computing, cryptography, and other military and intelligence applications. In short, they play a central role in US-China competition, meaning both countries are begrudgingly reliant on Taiwan for their national security.
Taiwan’s 5nm chips (not actually 5 nanometers, just a marketing term) are also integral to the US economy. Examples of their uses in high-end consumer electronic components include Apple’s M1 and A16 chips, as well as in Nvidia and AMD’s latest GPUs.
Can the United States (or China) catch up to Taiwan?
In the short term, no other country in the world can replicate Taiwan’s semiconductor industry. The process of setting up a new manufacturing foundry capable of producing cutting-edge chips requires an iterative and highly capital-intensive process, and only Taiwanese firms have access to the engineering talent and experience to do it in a commercially-viable timeframe.
Both the US and China are now heavily subsidizing their domestic semiconductor industries in an attempt to bolster supply chain security, but they are experiencing mixed results. In the US, the new CHIPS Act led Intel to announce $70 billion in foundry investments, but even so it will be more than a generation behind TSMC, and its production will only meet around 5% of US market needs. Meanwhile, ramping US restrictions on Chinese access to both technology and expert personnel are making it harder for PRC tech firms to survive, let alone catch up to Taiwan. Its semiconductor industry is plagued by bankruptcies and malfunctioning chips, and a gray market has developed which has led to significant quality-control and reliability issues.
What should the US goal be instead of trying to catch up?
The US should play to its strengths and develop its tech ecosystem as a whole, leveraging investment to sidestep the race against TSMC and instead research technologies that could replace the current silicon-based semiconductor chip manufacturing process. This may come through new materials science or new chip architecture, but either way it would require substantial financing for startups, firms, labs, and STEM students. The CHIPS Act does provide funding for all of this, but not as its primary aim.
In the meantime, Taiwan’s chip foundries are irreplaceable, and both the US and China are reliant on them for both military and economic reasons.
So that means China won’t invade Taiwan, right?
It is extremely improbable taking into account the situation as it is now. China is economically and geopolitically disincentivized from invading Taiwan, and an independent Taiwan has rhetorical value to the CCP because they can use propaganda about the US meddling in so-called domestic affairs as a scapegoat to distract from domestic grievances. US allegiances and relationship-building in the Pacific also act as a strong deterrent. However, just because China itself would suffer tremendously from a war with the US does not necessarily mean that some set of unknown future conditions may change their calculus for military action.
If a Chinese invasion did occur, would the US defend Taiwan?
The answer is unequivocally that yes, the US would defend Taiwan, although the scale, scope, speed, and nature of that defense cannot be predicted without making extensive assumptions. Nonetheless, here are two broad conclusions that can be stated with high confidence:
First, if there have been no significant breakthroughs in semiconductor production or technology by Intel or Silicon Valley startups, then an attack on Taiwan would be tantamount to a declaration of war on the United States. It would not only be a severe blow to the American economy, but would also have dire implications for future cybersecurity and AI balance of power were China to successfully capture Taiwanese foundries intact. The US response would need to be significant enough to prevent that, which may lead the government to defend the island in its totality.
Second, even putting semiconductor chip production aside, Taiwan is nevertheless critical to US security in the Asia-Pacific, and an invasion would necessitate action. Countries such as South Korea or the Philippines that are either allies or active regional partners may be forced to reformat their relationships with China to be more conciliatory if they feel the US does not demonstrate a commitment to regional defense. Even larger countries like Japan and India may feel that the US is withdrawing from the Pacific if they did nothing after a Chinese invasion of Taiwan. In response, they would likely adjust their policies to put up with Chinese statecraft, else be excluded from regional agreements and trade. In effect, a failure on the part of the US to defend Taiwan may lead to de facto Chinese control of the Pacific region.
Although it may be tempting to point to declining domestic US support for the war in Ukraine and generalize that there is reduced American commitment to stand up to authoritarian countries abroad, the risk of an aggressive Russia to US security interests is far lower than an aggressive China. The commercial and political impact of an invasion of Taiwan would be so severe that there would be near-universal alignment of interests in favor of taking action.
The United States needs to take actions that raise the cost of an invasion of Taiwan for China.
A number of steps could and should be taken to improve the likelihood of deterrence being successful. The US should better arm and train Taiwan, as well as pressure them to reform their military. It should strengthen Pacific alliances, and encourage more military cooperation in the region with and without US forces. Congress should also be more proactive about passing legislation regulating social media companies, because misinformation and disinformation are contributing to instability and corruption in India, the Philippines, and other countries in the region. Steps should also be taken toward institutionalizing US foreign policy so it changes less from one presidential administration to the next, with the aim of allowing US allies to more confidently stand up to China and other authoritarian states without worrying about mercurial Washington politics.
In addition, any action that limits Chinese economic or technological advancement without substantively impacting that of the US should be taken. The new chip sanctions are an example, but the same principle can be extended to many more industries to slow down China’s acquisition of new weaponizable technology. It should be noted that these types of restrictions may in fact make invading Taiwan a more attractive option in the short term. However, they are nevertheless important because an invasion in the short term is unlikely, so those restrictions’ value is in limiting future military capability. The probability of an invasion in the future is plausibly higher than it is now, so the trade-off is worth it.
What about negotiating with China?
With regard to negotiated agreements, there can be zero trust or assumption of good-faith intention on the part of China as it exists today, especially in light of its actions in Hong Kong. Updated negotiated agreements are also impossible because China has for decades considered Taiwan a domestic issue. Were there to be new discussions, the expectation on the part of the Chinese public would be for Taiwan (and the US) to give ground rather than the reverse. The CCP would lose credibility and legitimacy otherwise, because the notion that Taiwan belongs to China is too deeply ingrained in their education and media.
Taiwan’s semiconductor chip industry is irreplaceable at this point in time. The United States has a compelling reason to defend Taiwan in case of Chinese invasion just on this basis.
However, Taiwan’s continued technological dominance may eventually come to an end if breakthroughs in chip technology occur. The United States should continue relationship-building and should help bolster political stability in the region to ensure deterrence is both effective and lasting.
Selected interviews and articles:
A semiconductor strategy for the United States, by Christopher A. Thomas for Brookings.
And finally, a special thanks to my good friend Bobby Shore, who helped guide my research in preparation for this blog post. Bobby is pursuing a master’s degree at Stanford focusing on China’s political economy.