Semiconductors And Taiwan's “Silicon Shield” • Stimson Center - Stimson Center - Free Websites, Share News And Posts Publicly

With very different consequences, both China and the United States depend on Taiwan’s global dominance of semiconductor fabrication for most of the computer chips their companies consume and sell. Chinese semiconductor companies can only produce about 6 percent of chips needed to feed the Asian giant’s world-leading consumer electronics industry. China depends on the Taiwan Semiconductor Manufacturing Company (TSMC) to make up 70 percent of the deficit. TSMC also fabricates—under contract—92 percent of the most advanced chips designed by U.S. semiconductor companies.
China’s dependence on Taiwan has been called the island’s “silicon shield” against a Chinese attack, a critical factor that has not been publicly acknowledged by the Biden administration. With bipartisan congressional support, the administration has expanded the scope of sweeping security sanctions on exports of leading-edge American semiconductor technology to Chinese companies and military entities and launched a significant national effort to rebuild a domestic chip fabrication capability.
This paper focuses on the semiconductor industry as a crucial part of broader US-China technological competition and considers four challenges for U.S. semiconductors and policy towards China, including the possible longer-term consequences of U.S. success in blocking China from acquiring the technology and machines needed to fabricate leading-edge chips.
For very different reasons, both the technologically laggard China and the technologically leading United States depend on Taiwan’s global dominance of semiconductor fabrication for most of the computer chips their companies consume and sell. China depends on the Taiwan Semiconductor Manufacturing Company (TSMC) to provide 70 percent of the chips needed to feed its world-leading consumer electronics industry. At the same time, TSMC also fabricates under contract 92 percent of the most advanced chips designed by U.S. semiconductor companies which collectively earn about 47 percent of total global semiconductor and other integrated circuit (IC) revenue.
House Speaker Nancy Pelosi’s arrival in Taipei on August 2, 2022, touched off a firestorm of criticism from Beijing and the largest live-fire air and sea drills in the seas surrounding the island since 1966, when China fired ballistic missiles into the air and sea approaches to Taiwan’s two main seaports of Keelung and Kaohsiung, but well in international waters.
Unlike 1996, three of six exercise zones surrounding the island that were publicized by China intrude into Taiwan’s territorial waters.¹ Questions to a Chinese academic expert about China’s other retaliatory options in a BBC broadcast on National Public Radio on August 3 revealed a crucial fact that is seldom if ever noted by American officials, members of Congress, and the foreign policy and security community—the constraints on Chinese retaliation imposed by its high dependence on Taiwan for the semiconductors needed by its consumer electronics and other important industries.
Xi Jinping, China’s leader since 2012, and other Chinese leaders have long understood that absent a serious move by Taipei to seek de jure independence a direct attack on Taiwan would be disastrous for the Chinese economy, not to mention the predictable international sanctions and opprobrium. Some industry sources and commentators have called this Chinese dependency Taiwan’s “silicon shield.”
Without a doubt, the first Chinese bomb or rocket that should fall on the island would make the supply chain impact of the COVID pandemic seem like a mere hiccup in comparison. As the world’s leading trading country, the Chinese economy would be the biggest loser, but the impact would be global.
Taiwan’s globally strategic dominance of semiconductor fabrication has altered the traditional geography-economy-technology components of major power status and the dynamics of the U.S.-China-Taiwan relationship. Biden Administration policy makers and the President himself understand that Taiwan’s technological leadership and massive capacity for fabricating both leading-edge and trailing-edge chips cannot be significantly replaced by geographically diversifying the sources of chip fabrication and government funding to reestablish a limited domestic capability.
U.S. pushback against China’s regional military assertiveness, including large scale naval exercises with Japan and other U.S. allies and strong bipartisan support for closer ties to Taiwan, have also contributed to the worst bilateral relations in recent memory. U.S. support for Taiwan’s democracy reflects traditional American values but also the social, economic, and technological benefits of intellectual freedom.
These ideas are anathema to Chinese leaders and the Communist Party and are dismissed as U.S. and western propaganda. On the other hand, Xi and other policymakers are convinced that the defensive American denial of exports of advanced semiconductors and the machinery to make them is intended to block China’s rise. This is the most serious future challenge to maintaining peace and stability in East Asia and the western Pacific.
As a self-governing island of 24 million people lying 80-100 miles off the southeastern coast of China that has been claimed by the PRC since 1949 as a renegade province, Taiwan has long been the focus of first-order U.S. military, geopolitical, technological, and economic security concerns. The island’s strategic location opposite the coast of China’s Fujian Province and its critical role in the global semiconductor supply chain inevitably make it a focal point of American and allied concern about Beijing’s rising military power, assertive disregard of international law and norms, and regional influence.
The American relationship with Taiwan has been technically unofficial since the normalization of U.S.-China relations in 1979 and carried on through the auspices of The American Institute in Taiwan (AIT) and the Taipei Economic and Cultural Representative Office (TECRO) in the United States. The United States and Taiwan enjoy very close economic, educational, and scientific and technological relations, and successive American administrations and Congress have approved the sale of billions of dollars’ worth of advanced U.S. weapons for Taiwan’s defense.
In addition to the negative political atmospherics that followed changes in power in Washington and Taipei during 2016, U.S.-China-Taiwan tensions have been rising in response to China’s military modernization and expansion, its brutal takeover of Hong Kong, and increasingly assertive moves by Beijing under Xi Jinping to displace U.S. influence in the region and weaken its alliance system. More recently, the Biden administration and many Members of Congress appear to have become equally if not more anxious about the extent to which Taiwan’s global dominance of semiconductor fabrication and the critically important American strategic industry may be dangerously interdependent.
The far-sighted vison of Taiwan’s scientific and political leaders in the late 1980s has made the privately owned TSMC the contract fabricator of more than half of the world’s semiconductors overall and 92 percent of the most advanced chips. The latter are mostly designed by a half dozen or so leading American semiconductor companies including Apple, Google, Intel, AMD, Qualcomm and Nvidia. No other semiconductor company in the world can match TSMC’s expertise in the fabrication, efficiency, and forward-looking management. A registered Taiwan corporation, its ten largest institutional investors and ten largest mutual fund investors are all American entities.²
The breakdown of commercial supply chains between the United States and Northeast Asia because of the global COVID-19 pandemic underscored Taiwan’s strategic economic and technological importance to the United States. The chip shortage was not predominantly in the advanced semiconductors that TSMC makes for U.S. companies. Rather, the global shuttering of vehicle, appliance, consumer electronics, and other assembly lines was caused by the extreme shortage of the ubiquitous “trailing-edge” chips. These chips are made by TSMC, China’s Semiconductor Manufacturing International Corporation (SMIC), South Korea’s Samsung, and Intel, the only U.S. company still fabricating a significant number of chips domestically.
While the United States and China both depend on TSMC for most of their chip production, the nature of their dependency is very different. U.S. headquartered companies dominate the design of the most advanced chips and collectively earn 47 percent of global semiconductor revenues. The main U.S. weakness is its dependence on long and vulnerable supply chains.
The Chinese problem is much more fundamental. Despite billions of dollars’ worth of subsidies, grants, and contracts, China’s only large-scale state-owned semiconductor fabrication company, the Semiconductor Manufacturing International Corporation (SMIC), cannot come close to meeting the country’s needs. The basic problem stems from China’s very unbalanced development experience and the drag imposed by inherent inefficiencies in the state-dominated industrial sector.
Paradoxically, China is both the largest producer and largest consumer of semiconductors and related integrated circuits. In 2020 China’s semiconductor imports totaled $350 billion, two times its oil imports.³ In 2021, its production of integrated circuits/semiconductors soared by 33 percent, but semiconductor imports continued to surge as well, rising by 23.6 percent to $432 billion in 2021, probably due to rebuilding inventories that had been disrupted by the pandemic impacts on supply chains.⁴
As of 2020, Chinese semiconductor companies were only able to manufacture about 6 percent of the country’s basic needs and, even more important, they still cannot fabricate the leading-edge chips that are essential to Xi’s plan to make the country a technological peer of the United States by mid-century. Despite the hundreds of billions in U.S. dollar terms that the central, provincial, and municipal governments have lavished on both China’s state-owned and private semiconductor companies, the self-sufficiency needle has barely moved.
TSMC itself produces 10 percent of China’s domestic production in its factories in Shanghai and Nanjing and exports another 60 percent from Taiwan. South Korea’s Samsung, which still designs and fabricates its own chips, supplies another 12 percent of China’s chip imports, mainly for the final assembly of Samsung consumer electronics in Chinese factories.⁵
Just as the transistor replaced the vacuum tube beginning in the 1950s, the integrated circuit, also known as the semiconductor or microchip, came into its own in the 1980s. More than any other scientific achievement of humankind, these thin, fingernail-sized pieces of silicon imprinted with minute circuits and billions of transistors have enabled and shaped the world of electronic devices over last three decades and continue to drive the world’s e-future. What they do and how they are made almost defies the understanding of a lay person.
Semiconductor chips are the key components of computers, smart phones, sophisticated gaming graphics, aircraft avionics, computer controlled industrial machinery, motor vehicles, medical devices, and household appliances. Today’s automobiles may have as many as 1,500 chips, which is why the COVID-19 breakdown of chip supply chains shuttered vehicle factories around the world. A mobile phone may have more than a hundred mature technology chips, but the “brain” will be a central processing unit (CPU) composed of one or more leading-edge logic chips.
The size, speed, and energy efficiency of semiconductors is commonly expressed in nanometers (nm) of space between the millions or even billions of transistors. By comparison, the human genome is 2.5 nanometers, and one nanometer is about the width of four molecules of water.
Fabricating the “mature” and ubiquitous commodity chips in sizes from 300nm down to 28nm requires some form of photolithography, now mostly based on passing powerful beams of ultraviolet light through a series of mirrors, lenses, and “masks” to etch the circuit on the chip. These now “mature” or “trailing-edge” chips are made by the billions with low markup pricing.
The most advanced and most profitable chips of 10, 7, 5, and 3nm are incredibly difficult and costly to fabricate and can only be made by machines designed and built by ASML Holding of Veldhoven, the Netherlands, using a process called extreme ultraviolet photolithography (EUV). Because of their complexity the company can only produce 45-55 per year.
Advanced chip foundries require a clean room 10,000 times cleaner than a hospital surgical room. The chip etching process, critical parts of which are done in a pure nitrogen environment, can take a dozen or more weeks operating 24 hours a day and as many as 700 separate steps, many of them repeated.⁶ Completing all the necessary steps in the value chain may involve multiple air flights back and forth over thousands of miles to sites with specialized capabilities.⁷
TSMC claims to own 50 percent of all the ASML machines in existence and to fabricate 60 percent of the world’s advanced chips with them. By one informal industry estimate TSMC will have about one hundred machines at the end of 2022, versus 30 for Samsung and 20 for Intel.⁸ In 2018 China ordered and partly paid for its first EUV machine but because of U.S. intervention with the Netherlands’ government the machine has never been delivered.⁹ As of early 2022 ASML reportedly continued to sell older technology non-EUV machines to China, including refurbished ones.¹⁰
That no Chinese company is even in the top ten of the world’s leading semiconductor companies begins with the fact that in 1971, when California’s Santa Clara Valley was nicknamed “Silicon Valley”, China was still in the throes of Mao Zedong’s disastrous Cultural Revolution. The forced transfer of much of the urban population to the countryside left a generation of young Chinese without a modern education.
Deng Xiaoping’s reforms beginning in 1978 made China the “world’s shop floor” for multinationals’ global brands. Aided by key infrastructure investment in special economic zones that attracted low-value-added assembly operations with imported parts and components, export earnings and GDP soared to double-digit levels. From 1993 to 2003, First Vice-Premier and subsequently Premier, Zhu Rongji steered China towards Deng’s “market socialism” and pushed sweeping economic and bureaucratic reforms that gave the country a decade of rapid growth and economic stability.
Double-digit growth also benefited significantly from China’s initial low base, a swelling and low paid work force, and fast-growing manufactured exports. The growth benefit, however, was not reflected in the central government’s investment in science and technology.
Both Zhu and President Jiang Zemin saw accession to the WTO as a way for China to find its deserved place in the international economic and institutional order and to give them the political leverage to pursue their agenda of economic opening and legal, regulatory, and bureaucratic reform. Beijing fulfilled many of the trade, economic, and regulatory reforms required by the WTO agreement. In 1998 Zhu instituted a major governmental restructuring that reduced the central bureaucracyfromeight million to four million and ruthlessly cut the number of ministries from 40 to 29.¹¹
The acceleration that followed from China’s accession to the WTO in late 2001 led to a surge of U.S., Japanese, and European investment in China, but the fast-growing exports of goods produced in foreign- or joint-venture factories caused an equally fast-rising U.S.-China trade deficit. This quickly undermined the Clinton Administration’s expectations of a new and mutually beneficial trade and economic relation. For much of the American body politic, China became the embodiment of a popular and negative view of the WTO and globalization.
Xi Jinping’s Chinese dream of “the great rejuvenation of the Chinese nation” focuses on shifting the country from an economy based on foreign technology to one based on indigenous innovation-driven growth. An “Outline of the National Innovation-Driven Development Strategy (the Outline)” proposed by the 18^th session of the Central Committee of the Chinese Communist Party (CCP) in May 2016 implicitly acknowledged that the foreign investment and technology that fueled China’s emergence as the world’s largest manufactured export country had led to the neglect of indigenous science and technology. The Outline called beginning innovation-driven an “irresistible world trend” and declared that “it is the nation’s destiny to be innovation-driven.”¹²
Semiconductors are at the core of Xi’s “Chinese Dream” of “the “great rejuvenation of the Chinese nation”, which includes achieving entry into the “front ranks” of innovative countries by 2035 and becoming a global peer of the United States in all respects by 2049, the 100^th anniversary of the founding of the PRC. Skeptics see Xi’s dream as closely tied to his effort to create a populist image for both himself and the party in the leadup to his expected appointment to an unprecedented third term as paramount leader at the 20^th five-yearly Communist Party of China (CPC) National Congress this coming November. Still, Xi’s ambition to rejuvenate China has driven his policies from the outset of his tenure in 2012, and the nationalistic concept has gained wide popularity.
Xi’s new innovation-led economic development strategy amounts to industrial policy on steroids. At the same time as the central, provincial, and municipal governments are indiscriminately throwing large amounts of money at thousands of often hastily formed high-tech electronic startups, China’s regulators have been sharply reining in online commerce and internet services tech giants like Alibaba and Tencent, which are the most innovative part of the economy. Economist Barry Naughton argues that “China is a technological super-power because it followed smart policies after 1978, pursuing marketization and investment in human and physical capital.” It is too early to know, he says, what effect its aspiration to become “the first government-steered market economy” will have.¹³
Thus far, no major aspects of Xi’s dream appear to be moving in the right direction. The production of mature technology commodity chips has been growing by double digits year-on-year, but imports are also rising. Overall economic growth has slowed markedly in the wake of continued COVID-19 lockdowns, exports falling due to supply chain breakdowns, and decreasing economic productivity.
While the raw economic nationalism of Xi Jinping’s signature “catch up” initiatives have disturbed the U.S. and other foreign governments, the most serious concerns seem likely to prove overwrought. Made in China 2025 (MIC-25), adopted in 2015 as part of the 10-year economic plan after extensive academic input, aimed at raising the domestic content of “core components and materials” to 40 percent by 2020 and 70 percent by 2025.¹⁴ To that end, Xi directed a full review of the foreign content of Chinese electronic devices, especially from U.S. companies, and their substitution with as many locally made components as possible to create “secure and controllable” domestic sources. The initiative has attracted significant public support, but with respect to semiconductors it has already fallen well short of the 2020 goal and has no realistic way of achieving the 2025 target.¹⁵
By one industry estimate, China’s domestic output of integrated circuits would only grow from about 15.9 percent of the Chinese market, of which only 5.9 percent was by Chinese companies, to about 19.4 percent in 2025, not the 70 percent target of the MIC-25. The five-year estimated amounts to only 0.7 percent per year on average, of which more than 50 percent is likely to be produced by foreign companies like TSMC, South Korea’s SK Hynix, and Samsung factories in China.¹⁶
Ironically, it was the MIC-25 program that prompted the Trump administration to release a scorching May 2018 report by the Office of the U.S. Trade Representative (USTR) on China’s intellectual property (IP) theft, its insistence on technology transfer as a condition for foreign investment, and its rigidly protectionist market. A month later the administration announced the imposition of a 25 percent, $50 billion tariff on 1,100 Chinese goods categories, including, especially the aerospace, robotics, manufacturing, and auto industries. Additional tariffs followed as U.S.-China trade conflict deepened. The Biden administration is only now reconsidering at least some of these tariffs as counterproductive, especially in a politically damaging period of historically high inflation.
MIC-25 in fact created such widespread U.S. and other international criticism that the Chinese government stopped talking about it, though elements of the program continue. A major obstacle is the still small high-technology ecosystem in China and the fact that the profit-driven private sector still prefers to purchase parts and components from established U.S. and other foreign suppliers rather than untested Chinese vendors.¹⁷
A new “dual circulation” concept announced by Xi Jinping at a May 2020 CCP Politbureau Standing Committee meeting sought to reinforce the MIC-25 initiative by reducing imports and strengthening China’s export competitiveness by raising innovation and manufacturing more sophisticated goods, partly by attracting foreign investment by companies with technology that can help Chinese partners move up the value-chain. The underlying rationale is that the production of higher quality and more sophisticated manufactured goods that would appeal to Chinese consumers would also find more foreign buyers.
The dual circulation plan has not been treated kindly by foreign experts on the Chinese economy, who see major flaws in the concept, starting with the economic reality that Chinese consumers’ share of GDP is about 55 percent compared to 70-80 percent in the West. Michael Pettis, who teaches finance at Peking University, notes that the goal of increasing consumption to support higher quality and more exportable domestic manufactured goods is economically impossible without fundamentally changing the heavy dominance of unproductive investment in economic infrastructure and real estate and subsidies to inefficient state-owned industries. Pettis also points out that technology accounts for no more than 10 percent of GDP and that the root problem for this mostly private sector, as in many countries, is not access to capital but a shortage of projects “that offer appropriate risk rewards.” All the necessary changes to break out of the failing, debt-driven development model would face major resistance from the groups who benefit from the status quo.¹⁸
Although domestic chip production has been increasing, imports are rising as much or more. Imports of integrated circuits soared by 24 percent from 2019 to 2020, and 43 percent from 2020, to a record of $434 billion in 2021.¹⁹ Some Chinese companies, especially private ones, have been making significant gains in chip design, niche technologies and “back-end” finishing operations including “packaging” and testing chips, which have allowed them to succeed in specific parts of a highly competitive market. For example, the number of Chinese-made parts in Apple iPhones has grown steadily with each new model.
Of the top ten Chinese semiconductor companies ranked by revenue, SMIC, the only foundry in the list, ranks second with $7.5 billion in earnings and a 16 percent global market share, after the United States and Taiwan.²⁰ These are almost all “mature” commercial technology chips that are in high demand but do not help China move up the technology curve. Industry sources report that while SMIC claims to have achieved production of 14nm chips, actual sales of these chips amounted to a negligible share of its overall revenue as of early 2021.²¹ Due to U.S. intervention with ASML and the Dutch government, SMIC cannot buy the machines required to make chips to the 7nm process needed for 5G capability.
Despite many iconic high technology achievements such as establishing its own space station and landing a rover on the dark side of the moon, which made China the third country in space after the United States and Russia, most of the Chinese economy is less technologically impressive. Despite its glittering cities, China remains a poor country, ranking 63^rd in nominal per capita income in 2020, just below the Seychelles and just above Malaysia. China ranks 77^thin purchasing power parity (PPP), which adjusts for differences in national costs of living, just below Thailand and Equatorial Guinea and just above Botswana. For comparison, the United States ranked 5^th in nominal per capita GDP and 7^th for PPP for 2020.²²
A study comparing national standing in technological intensity showed broad Chinese weaknesses. Despite being the world’s second largest economy and the largest exporter of goods, China only ranks 29^th out of 147 countries in an economic Intensity Index– a measure of the relative knowledge intensity of a county compared to all the others–published by the Observatory of Economic Complexity (OEC) at MIT.²³
From 2000 to 2020, China’s ranking rose significantly from 54^th in 2000 to 29^th in 2020, but Taiwan soared from 22^nd to 2^nd, South Korea from 35^th to 5^th, and Singapore from 21^st ^to 6^th. The current estimates of economic complexity put the mainland in about the same category as Taiwan, South Korea, and Singapore were two decades ago.²⁴
Far more even than Taiwan and South Korea, China also has a critical shortage of the necessary technicians and engineers to operate foundry machines to expand domestic chip production. Even the machines that fabricate mature technology chips require highly skilled technicians. China has few such highly trained personnel unless they have worked at TSMC or Samsung. By one measure China needs 300,000 more engineers and technicians to achieve its goals.²⁵
Because both Taiwan and South Korea are also short of people with the right skills, their governments have banned their electronics engineers and semiconductor technicians from going to work for Chinese companies. Still, it would be dangerous for the United States to underestimate China’s rapid rise in many aspects of science and technology and how far it may be able to exceed U.S. competitiveness with medium-level technology that may be better tailored for success in global markets.
In 2018 Congress passed the Export Control Act of 2018 (ECRA) which updated and revised a 1979 law that had expired and passed a Foreign Investment Review Modernization Act as part of the same legislation. Both Acts responded to sensitive high-level security concerns about exporting leading-edge technology, whether to Chinese entities or to U.S. invested factories in China, but focused on fourteen “Emerging and Foundational Technologies” including advanced computing technology, AI, and advanced learning technology, microprocessor technologies, and quantum information, and sensing technology.²⁶
The stiff new U.S. export controls on the transfer of leading edge technology to Chinese companies and military entities that started in the Trump administration have been both broadened and fine-tuned by the Biden administration. The cumulative effect has dealt a heavy blow to Xi Jinping’s “Chinese dream” of becoming a technological and geopolitical peer of the United States by 2049. For leading-edge semiconductors, the kind that are key to the higher reaches of AI, quantum computing, 5G, and advanced military electronics, China now faces a global technological shut-out with no ready indigenous path to catching up.
The most publicized export license denial cases have been those involving China’s largest private technology conglomerate, Huawei, the world’s fourth largest electronics company. In mid-2020 Huawei at least temporarily overtook both Apple and Samsung as the world’s largest smart phone manufacturer.
Huawei achieved a dominant global position for its 5G communication networks and base stations by combining readily available medium technology chips and Qualcomm’s leading-edge “Snapdragon” chip, fabricated by TSMC. The Qualcomm chip was also essential to the 5G capabilities of its best-selling smartphones.
In May 2019, the Trump administration announced that it would ban the domestic use of telecommunications equipment made by foreign companies that posed a security threat to the United States. No country was named in the announcement, but it was no secret that the target was Huawei. The issue of most concern to U.S. and some allied intelligence agencies was that Huawei, one of the world’s leading 5G telecommunications companies, had a “back door” in its best-selling 5G communications equipment that was fast becoming the leading brand in global markets. The threat was both one of direct national security as well as a commercial and economic one.
The U.S Commerce Department’s denial specifically applied to the export of Qualcomm 5G chips used in Huawei’s market leading 5G mobile phone and as well as updated software for Google’s Android operating system and apps. Until the license denial, TSMC had been making 98 percent of Huawei’s chips, including Qualcomm’s “Snapdragon” chip needed for 5G capability.²⁷
For a while Huawei continued to make its top-of-the-line 5G phone using stockpiled chips, but once its inventories were depleted it withdrew from the 5G market and spun off its cheapest “Honor” phone as a separate company that was not on the U.S. blacklist. Largely because of being blacklisted by the U.S. Commerce Department Huawei’s smartphone revenue dropped by an estimated 28.9 percent in 2021, the first decline ever.²⁸ Eventually the Commerce Department’s Bureau of Industry and Security (BIS) added hundreds of Chinese companies, research institutes, and military organizations to its entities list.
In addition to protecting “transformative” American technology, the use of export controls and blacklisting of Chinese entities are also having a direct impact globally, including the indirect creation of more competitive technology markets that in some cases might benefit U.S. companies. For example, while the Commerce Department has banned the export of Qualcomm chips for Huawei’s 5G telecommunications base stations, the American company reportedly is designing a new advanced chip for 5G base stations and networks that has the potential to open up mutually beneficial business for Nokia’s and Ericsson’s 5G base stations and any future entrants.²⁹ Intel has also designed its first 5G chip, the Atom P5900 5G integrated chip platform that Ericsson and China’s ZTE, two of Huawei’s few international competitors, say they will use in their base stations.³⁰
Maintaining U.S. leadership of the global semiconductor industry, particularly the most advanced chips, and China’s continuing dependence on Taiwan for the “mature” chips needed to sustain its role as the world’s largest producer of electronic devices, consumer electronics, and vehicles are critical to U.S. economic, technological, and military security. The danger of the status quo is clear:
The Biden administration, Congress, and the U.S. semiconductor industry have both been eager to find a way to reduce dependence on semiconductor supply chains from Taiwan and East Asia more generally. So far, however, the most significant governmental action has come at the state and local level with tax breaks, infrastructure, and investment in workforce education.
In February 2021, President Biden signed Executive Order (E.O.) 14017 “America’s Supply Chains,” directing the immediate conduct of a 100-day “whole-of-government approach to assessing vulnerabilities in, and strengthening the resilience of, critical supply chains.” In addition, the executive order called for a more in-depth one-year review on supply chain vulnerabilities and resiliency, including consultation with a wide array of non-government stakeholders. Regarding semiconductors, the Order directed the Secretary of Commerce to consult with other appropriate departments and agencies to “submit a report identifying risks in the semiconductor manufacturing and advanced packaging supply chains and policy recommendations to address these risks.”³³
The Final Report of the 100-day review focused on “four critical products: semiconductor manufacturing and advanced packaging; large capacity batteries, like those for electric vehicles; critical minerals and materials; and pharmaceuticals and active pharmaceutical ingredients (APIs)”³⁴ and included a broad range of concrete steps to be taken by relevant U.S. Departments and agencies ranging from the Agriculture Department to the U.S. Development Finance Corporation.³⁵
None of these initiatives alone can end U.S. dependence on Taiwan for fabricating most of the semiconductors that are critical to American technological leadership; economic and military security, and stability in the Indo-Pacific region. That said, successfully meeting four challenges can reduce supply chain risk for this critical technology and help support the interdependence between a free Taiwan and U.S. leadership.
Starting in June 2020 Congress began considering a “Pacific Deterrence Initiative” as a framework for the annual FY 2022 DOD authorization and appropriation bills that would provide up to $27 billion over six-years that would allow “Congress and the Pentagon to view the defense budget through a regional warfighting lens…”³⁶ The DoD authorization that specifically referenced the PDI has been signed into law, but among the DoD funds included in the FY2022 Consolidated Appropriations Act signed into law on March 15, 2022, some but not all of the funds were scattered across various existing programs.³⁷
The U.S. military, especially the 7^th Fleet, has significantly increased its freedom of navigation operations (FONOPS) in the South China Sea and Taiwan Strait, and has held increasingly large naval and air exercises in the Philippine Sea, east of Taiwan, and in the northern reaches of the South China Sea. The largest exercise to date took place in early October 2021, coinciding with China’s week-long National Day holiday, when two U.S. carrier strike groups led by the USS Reagan and USS Vinson, the U.K.’s Carrier Strike Group 21 led by the HMS Elizabeth, three Japanese Maritime Self-Defense forces (JSDF) helicopter carriers and two destroyers, and one frigate each from the Netherlands, Canada and New Zealand carried out joint exercises in the Philippine Sea.³⁸
The Japan Maritime Self-Defense Force (JSMDF) is now modifying two large helicopter carriers to carry the 42 F-35B Lightning II strike fighters that Japan is buying. At the request of the Japanese Maritime Self-Defense Force (JMSDF), several Marine F-35B strike fighters demonstrated the jets’ Short Take-off and Vertical Landing (STOVL) capability.³⁹ The exercise drew the largest number ever of different types of Chinese warplanes as stand-off observers.⁴⁰
Most recently, in the wake of Russia’s invasion of Ukraine, China’s growing assertiveness, and North Korea’s increasing missile and nuclear threat, Japan’s ruling Liberal Democratic Party (LDP) has called on the government to roughly double the national defense budget and consider the right of preemptive attacks on threatening missile launches as part of a review of 2013 Cabinet adopted long-term national defense guidelines.⁴¹
These proposals remain straws-in-the wind, but they appear to represent a major Japanese rethinking of the consequences if China were to control Taiwan, which is right on the doorstep of Japan’s southern islands. Japan also plans to develop new fighter aircraft and medium-range anti-aircraft missiles with the United States and the U.K. and export them to 12 countries with which it has defensive security agreements, including the U.S. and the U.K., India, Australia, Vietnam, Thailand, Indonesia, and the Philippines, both to lower production costs by increasing the volume and present an alternative to Chinese weapons exports.⁴²
The July 8, 2022 assassination of twice-former Prime Minister Shinzo Abe could cause a setback for a more active Japanese role in the U.S.-Japan alliance. The LDP increased their seats a few days later in the July 10 Lower House election, but fell short of an outright majority.
As for semiconductors, in a “Request for Solutions” dated January 15, 2021, entitled “Rapid Assured Microelectronics Prototypes – Commercial (RAMP-C)” the U.S. Naval Surface Warfare Center (NSWC) noted that “The United States currently has no onshore access to foundry technology capable of meeting the Department of Defense’s (DoD) long-term leading-edge microelectronics manufacturing needs.” The request seeks “to establish a relationship with a U.S.- located foundry that can fabricate a custom leading-edge foundry that can collaborate to modify commercial off-the-shelf semiconductors and other integrated circuits to meet critical needs.” Importantly, the project requires that the work is done within the U.S. borders, not that the company be American.⁴³
The highly competitive nature of a private sector industry that seeks to maximize profits rather than explicitly support U.S. national security policy or interests makes it very difficult to promote the reestablishment of domestic semiconductor foundries. The U.S. companies chose to offshore chip fabrication to maintain their price competitiveness, but they did so at a time when East Asia was peaceful and more conducive to regional and transpacific supply chains.
The reality is that design, not fabrication, is the most profitable part of the semiconductor market. By one calculation design creates 53 percent of the value added to a semiconductor/integrated circuit versus 24 percent for fabrication. Investors don’t like to see their capital used for inherently less profitable projects unless the resulting synergy is positive, as sometimes may be the case.
Still, knowing that China has already used leading edge American designed chips in its advanced weapons systems, the United States has no choice but to deny advanced semiconductor exports to any Chinese company working for or supplying such technology to military or internal security entities. The BIS Entities List already includes hundreds of Chinese companies and other entities that are blacklisted and a smaller number for which exports require a case-by-case review.
The COVID-19 pandemic and now the Russian attack on Ukraine have underscored that the profit motive that drove fabrication offshore in an industry as important as semiconductors is not necessarily good for the U.S. economy or national security—but at the same time, offshoring has helped maintain American supremacy in the industry. The combination of offshore contract fabrication with domestic leading-edge research and chip design has maintained and perhaps increased the 47 percent global market share for U.S. branded chips.
In the past U.S. companies contracted companies in Asia for fabrication and related services because of lower operating costs and, in Taiwan in particular, a highly trained workforce with pay rates significantly lower than in the United States. From the founding of TSMC in the late 1980s, investment and contract production have been based on a deeply trustworthy bilateral relationship. While the financials of offshore foundry services haven’t changed, the business and national security risks have demonstrably risen.
Establishing domestic semiconductor foundries, whether by American companies or their foreign competitors, is said to be a “trillion-dollar” challenge with many potential pitfalls.⁴⁴
Governmental support for infrastructure development and closing the cost differential will of necessity be a long-term commitment.
Industry experts, Intel, TSMC, and Samsung all contend that fabricating chips in the United States is significantly more expensive than in mainland China, Taiwan, or South Korea, but the actual numbers are elusive. A March 25, 2022, article in an industry publication entitled “Why subsidies are needed for U.S. domestic chip manufacturing” argues the national interest need for subsidies from the Federal government, but it never answers the question of how much is needed or why.⁴⁵
In lieu of adequate data, the estimated extra cost of reshoring chip fabrications is often based on estimates of the level of subsidies provided to foundries in the three Asian competitor countries. Apart from fabricating and testing the chips abroad, American semiconductor companies still manufacture most of their final products domestically. Semiconductor products are a major and still growing U.S. export, amounting to $55.21 billion in 2022.⁴⁶
Currently Intel, which is in a close competition with Samsung as the world’s largest semiconductor company, is the only American company producing semiconductors on a large scale domestically. In September 2021 the company broke ground on a $20 billion project to add two new foundries to four existing foundries and 12,000 employees it already has in Chandler, Arizona.⁴⁷
Intel has an even more ambitious $20 billion project to build two new foundries in New Albany, Ohio, with the intent to invest a total of $100 billion over ten years to create “the largest silicon manufacturing location on the planet.” Intel cautions, however, that any investment beyond the first two plants would depend on Federal and local subsidies and tax breaks.⁴⁸
Intel’s plan is not to produce the older chips that remain in high demand for a wide range of products, including automobile manufacturing, in these new foundries in Ohio, but rather leading-edge chips of 3nm or less, which are needed for AI, autonomous driving, the internet of things, and other advanced purposes.⁴⁹
TSMC has two multi-billion-dollar projects for leading-edge-foundries in Arizona, and Samsung is building a $17 billion plant in Taylor, Texas to make leading-edge chips for mobile phones, 5G, high-performance computing, and artificial intelligence. Samsung plans to break ground this year and start operation by 2024.
Already TSMC has said that for varying reasons, including competition with Intel’s foundry project in Phoenix for construction workers and trainable technician prospects, that the costs at its Chandler, Arizona site are higher than anticipated, but “manageable.” Senior TSMC officials have repeated their call for making Federal funding available to both domestic and foreign companies.⁵⁰ The company reportedly has not formally changed its construction schedule but has pushed the anticipated date to move in the chipmaking equipment later by six months into early 2023.⁵¹
Because offshoring foundries is the established division of labor globally, there is a basic conflict between that model and U.S. interests such as self-sufficiency, secure supply chains, and national security, narrowly defined. In the face of many variables, it is difficult to calculate the level of subsidies, infrastructure improvements and other benefits that it will be necessary for TSMC, Samsung, and Intel to operate their new or expanded U.S. facilities at a profit comparable to operations in other countries. Intel, TSMC, and Samsung all insist that it is crucial that they share the same eligibility as U.S. companies for proposed federal funding.
Several bills introduced in Congress in 2021 and 2022 sought to support the expansion of domestic semiconductor fabrication. This included a $52 billion “Creating Helpful Incentives to Produce Semiconductors (CHIPS) for America Act” as part of the FY 2021 National Defense Authorization Act, which included $45 billion in grants and loans for increasing supply chain resilience and American manufacturing.⁵² Appropriations for these purposes were included in Senate Bill 1260–United States Innovation and Competition Act of 2021, which passed the Senate on June 8, 3021,⁵³ but the necessary companion House bill was never introduced and the legislation died with the 117^th Congress.
On July 27, 2022, the Senate finally passed a long-pending $280 billion package of support to U.S. science and technology, including $52 billion “specifically to support revitalizing U.S. chip production and reducing supply chain risk”. The House of Representatives passed the bill the next day and sent it on to the President. The funding package will almost certainly be only a down payment on longer term support to the creation of a new domestic chip production.
Sanctions and blacklisting have commercial consequences and China is a very big market for U.S. semiconductor companies. Partly because it had to stop selling 5G chips to Huawei and other now-blacklisted Chinese companies, Qualcomm reported a 48.1 percent year-on-year drop in the company’s shipments to China in 2020.⁵⁴
There could come a time when U.S. allies and partners will no longer be willing to deny China leading-edge technology. However, as the United States controls about 80 percent of the relevant technology, it would be a huge step for any friendly government to violate current international law on IPR and patents. In theory, one partial solution could lie in finding a way to relax controls on exports of higher technology to China in response to a commitment by Beijing to adopt a less threatening posture towards Taiwan, while still retaining firm American control of the most important transitional technology. It is not out of the question that future Chinese leaders could return to the path of “joining the world” instead of seeking to supplant the U.S. role and influence in the Indo-Pacific region.
The Commerce Department’s handling of SMIC, by far China’s only large-scale chip maker with dual civil-military entities, has been typical. SMIC was added on to the Entities List with a flat “presumption of denial for items uniquely required for the production of semiconductors at advanced technology nodes (10 nanometers and below), including extreme ultraviolet technology.” Requests for all other items would be considered on a case-by-case basis. Exports to China’s leading designer of advanced chips, HiSilicon, whose chips have been fabricated by TSMC, are also subject to the Presumption of Denial.
For China, not being able to make chips in the 10-7-5-3-2nm notes or import the EUV machines and other semiconductor manufacturing equipment (SME) to make them will be like hitting a technological wall. Apart from possible laboratory scale achievements, there is no alternative to ASML’s EUV technology.
One of the first commercial impacts likely will be on the U.S.-China race to build and deploy autonomous driving vehicles. At present, early generation autonomous vehicles are being built, road tested, and even operated in small areas of China, the United States and five other developed countries. The technical demands of these early tests have been described as around level two on a complexity scale of one to five. Thus far, China has been moving more quickly than the United States and other competitors by using mature technology chips that are good enough for operating in a controlled environment. Leading-edge chips at much higher levels of complexity are needed to incorporate faster graphics, more demanding AI functions, multiple cameras, and other kinds of more advanced sensing devices. Because China lacks access to the more advanced chips, its competitive position will be compromised.
The fourth and most difficult challenge stems from the fact that China cannot achieve its high technology goals and “dream” of being an innovative peer of the United States under a ban on the export of the high-end EUV fabrication machinery solely made by the Dutch company, ASML. In this sense the longer-term possibility of a Chinese attack on Taiwan could become more rather than less likely, however catastrophic for China and the global economy.
The suggestion by some U.S. and allied military experts and commentators that China could somehow seize control of these machines by invading Taiwan or carrying out some more limited local military action is not realistic for many reasons:
The machines require many highly trained and experienced engineers and technicians to set up and operate them, including ASML employees and contractors.
So long as China is dependent on Taiwan to make up 70 percent or more of its domestic semiconductor shortfall, the costs of attacking Taiwan militarily will be catastrophic for China, the region, and the global economy. In theory, China-Taiwan relations could in the future move toward some level of formal instead of de facto economic and technological integration. In such an event, U.S. objections to licensing the export of leading-edge semiconductor and other transformative technology to China would not likely be relaxed, absent major improvements in China’s respect for American intellectual property and a major reduction of its aggressive behavior and hegemonic aspirations.
In a September 12, 2021 article Forbes contributor George Calhoun argued that if a war with China were to occur, it “will arise from acute economic pain, inflicted on China by actions of the United States to deprive them of the most essential physical resource of the 21^st century: semiconductors.” In support, he cited remarks by Chinese Vice-Premier Liu He to a gathering of the country’s top scientists in May 2021. “’For our country,’ Vice Premier Liu He said, “‘this technology is not just for growth. It’s a matter of survival.’”⁵⁶ (Emphasis in source.)
Vice-Premier Li has succinctly defined the problem as perceived by China’s leaders. While some Chinese companies are moving up the value chain in certain niches of the semiconductor industry, the United States controls the export of leading-edge chips and access to the machinery to manufacture them that contains American intellectual property or patents. Given the technological and human obstacles China faces, this dependency will continue into the foreseeable future, likely at a significant cost to U.S. corporate earnings and the Federal budget.
Eventually, the confrontation over high technology exports involving the world’s two largest economies and military establishments is certain to break—but in what direction? Given China’s rising power and its track record on IPR violations and outright theft, the United States has no choice but to diligently protect its most advanced technology. Under China’s current autocratic leadership, CCP controlled industrial policy, and lack of intellectual freedom, it seems very unlikely that the Made in China 2025 approach will fulfill Xi Jinping’s objectives. Ironically, the failure of Xi Jinping’s “Chinese Dream” of parity with the U.S. in technology, innovation, and global influence by the one hundredth anniversary of the PRC in 2049 could, if that leads to regional instability involving China, be an equally difficult challenge for the United States under some circumstances as China’s acquisition of advanced U.S. technology.
Notice to Visitors: Stimson’s DC office is open; visitors for events and scheduled meetings are welcome. Visitors must be fully vaccinated against COVID-19 and complete an attestation form to that effect. Because we use a hybrid work schedule, please confirm that your host will be present before you arrive.
Copyright The Henry L. Stimson Center
Privacy Policy

source

Chat read-only to anonymous users. Chat with Anyone and Anywhere. Only registered users are allowed to send messages.

Loading the chat ...

28982 Register Login

Leave a Reply Cancel reply