When the Chips Are Down: Policy Priorities for Sustaining U.S. Semiconductor Leadership
Throughout 2021, stories have dominated headlines of a chip, or semiconductor, shortage generated by the demand spike due to the COVID-19 pandemic. Yet the short-term commercial pain from a semiconductor shortage pales in comparison to the long-term economic and national security damage of failing to invest in U.S. semiconductor leadership. Semiconductors are foundational to the information and communications technologies underpinning the modern global economy — powering AI and machine learning, biotech, quantum computing, self-driving vehicles, and the Internet of Things, all linked through high-connectivity 5G — and as a result, represent the keys to innovation, productivity and economic growth, and national security.
If the United States intends to retain its longstanding advantage in critical technologies essential for national defense and economic growth, it needs to surmount coming policy challenges tied to semiconductors. The Biden administration will need to prioritize responding effectively to major policy challenges in industrial policy, defense applications, human capital, export controls, and allied cooperation. Calibrating semiconductor policy responses through legislation, interagency deliberation, consultation with allies, and stakeholder engagement, especially with industry leaders, is fundamental to ensure that the United States maintains national security advantages and drives private enterprise-led, market-based, and rules-governed global commerce. Failing to do so will hand a distinct advantage to U.S. competitors for decades to come.
Semiconductors and National Interest
Amid a bipartisan recalculation that the U.S. approach toward China needs to take a stricter line, how the United States manages the evolving competition for advanced technology development and leadership will be critical. The semiconductor industry is one area where nascent U.S.-Chinese technology competition and rising global trade frictions have already made a noticeable impact. Semiconductors are foundational to U.S. economic prosperity and national security because they power virtually every defense platform the United States operates, and because they are an essential input to nearly every modern product or technology produced in the U.S. economy, as the repercussions of the current semiconductor shortage on the U.S. auto industry show. The industry also represents a vital source of U.S. exports — semiconductors are currently America’s fifth-largest export — and generator of high-skill, high-paying jobs with significant employment multiplier effects across regional economies. The United States and a handful of partners and allies (notably Germany, Japan, the Netherlands, South Korea, and Taiwan) remain global leaders across essential facets of the semiconductor ecosystem, notably semiconductor research and development, design, and manufacturing and in semiconductor manufacturing equipment. However, challenges to U.S. leadership are mounting, while its advantages are eroding.
International competition for leadership in this critical sector is fiercely intensifying. For instance, from 2001 to 2016, China’s share of total value added in the global semiconductor industry grew almost fourfold, from 8 to 31 percent and Taiwan’s share doubled, while America’s share fell from 28 to 22 percent, and Japan’s share fell by over two-thirds, from 30 to 8 percent. In 2016, China produced $120 billion in value added, compared with $83 billion for the United States. These developments underscore the high-stakes competition that the United States faces in this crucial area of strategic technologies, with supply chain security and resiliency a prime consideration.
Key Industry Dynamics
The most important trends impacting the U.S. semiconductor industry include restoring U.S. leadership at the cutting edge of semiconductor production and increasing domestic semiconductor production. Additional concerns include foreign barriers to market access, Chinese government efforts to pump money into inferior competitors regardless of market performance, supply chain disruption, shortfalls in human capital, insufficient public investment in pre-competitive research and development, and gaps in resiliency for advanced manufacturing, particularly for defense applications. These dynamics are critical for national security because secure access to leading-edge semiconductors will drive new technologies in defense applications, including faster networks, cloud computing, unmanned systems, cyber security, and AI applications — all components of U.S. future defense planning.
In 2019, U.S.-headquartered enterprises’ market share of this $470 billion industry stood at 47 percent, according to the Semiconductor Industry Association’s annual assessment. Yet making semiconductors is complex and expensive. The semiconductor industry is among the most capital intensive in the world. A new foundry can cost upwards of $15 billion to $20 billion, and require extensive manufacturing know-how and robust infrastructure to operate. Contemporary back-end facilities for assembly, packaging, and testing can run to $5 billion to $7 billion. In addition, firms invest a significant portion of their revenue on research and development. This permits them to stay ahead of the competition by developing and iterating technologies in a global pipeline, introducing new products every 18 to 24 months. In 2019 alone, for example, Intel invested over $29 billion in capital expenditure and research and development. Overall, the industry is America’s second-most capital intensive, beyond only alternative energy.
The production process is truly global, with some advanced semiconductor products transiting over 25,000 miles before completion. The global semiconductor value chain is primarily centered in the United States and East Asia. On average, each of the major facets of the semiconductor production process (research and development, design, manufacturing, and outsourced packaging, assembly, and testing) has firms from about 25 countries directly involved and firms from 23 countries in supporting roles. Linkages between governments, between government and industry, and within industry itself all have essential roles in shaping these dynamics. Yet the geopolitical aspects of the industry have begun to take on significant importance, largely in response to China’s actions.
The Chinese government has redoubled its efforts to develop an indigenous semiconductor capability, despite past failures. Beijing aims through its Made in China 2025 strategy and massive national semiconductor development investments to halve its dependency on U.S. semiconductors by 2025, and to achieve independence from U.S. supply entirely by 2030 through domestic development and alternative sources. One reason China’s government wishes to do so is that the country spends more money importing semiconductors ($160 billion annually) than any other product, even oil — although this ignores that these imported semiconductors are often re-exported, for instance, as part of the $681 billion of Chinese exports of final information and communications technology goods in 2018. China remains relatively low on the value chain in most areas, but is spending tremendous amounts to close the gap, with nearly $1.6 trillion allocated to the sector since 2014, with particular focus on the design of memory and AI chips. Chinese government actions, including immense subsidies and planned construction of over 60 new fabrication facilities, threaten the stability of the global supply chain for semiconductors and are likely to disproportionately impact other global companies attempting to compete on market-based terms. Fundamentally, excessive Chinese state support, technology/trade secret theft, and limiting U.S. companies’ market access will create an environment where uncompetitive Chinese companies remain in the market, damaging prospects for global competitors, including for U.S. firms.
Why should U.S. policymakers take action? Ensuring U.S. leadership in an industry that contributed $7 trillion in economic activity and $2.7 trillion in total gross domestic product underscores its vital economic position. Yet future scenarios where the United States loses access to advanced fabrication facilities and supply chains; cannot reliably develop or procure secure, leading-edge chips for commercial or military applications; lacks the human talent to continue to innovate semiconductor products; or allows transfer of advanced chips to China’s military are all feasible. Even a scenario where the United States fails to manage the trade-off between U.S. companies needing to sell to China to fund research and development and limiting the Chinese ability to close the technology gap in the sector is problematic. How should the United States respond? There are five high-level debates brewing with regard to securing the technology, know-how, and facilities for advanced production of semiconductors.
Toward a Smart Industrial Policy
One emerging concern is that, despite its strengths in semiconductor design, the United States lacks pure-play foundries at the cutting edge of the spectrum, particularly to meet military requirements for advanced systems and for overall resiliency. Policy intervention will be required to ensure a level playing field for U.S. companies on the global stage. (Currently, Intel, Samsung, and Micron all operate advanced foundries, but control other aspects of their chip production as well.) As a strategic industry, U.S. primacy in semiconductor manufacturing may dissipate if there are inadequate new facilities constructed to manufacture chips. For instance, whereas China held barely 1 percent of global semiconductor manufacturing capacity in 2000, this had grown to 15 percent by year-end 2020, and is forecast to increase to 24 percent by 2030. The U.S. share is now just 12 percent, which is expected to fall to 10 percent or less by the end of this decade without effective policy intervention. From 2007 to 2021, China’s share of semiconductor fabrication capacity increased by 13.4 percent, while America’s fell by 6 percent.
The United States still has many advantages to offer manufacturers, including high quality infrastructure and logistics, innovation clusters, excellent universities, and a history of leadership in the field. However, challenges abound — in recent decades the political will to retain advanced manufacturing has dwindled, federal government tax incentives have diminished compared to competitors, and federal support for pre-competitive research and development has languished. This is an area where Taiwan, principally through Taiwan Semiconductor Manufacturing Co., has emerged as a global leader in semiconductor fabrication. Taiwanese companies now account for 78 percent of global foundry revenues, and in turn Apple alone now accounts for 25 percent of Taiwan Semiconductor Manufacturing Co.’s revenue. Fortunately, there is a high degree of complementarity between U.S. and Taiwanese strengths, with Taiwan Semiconductor Manufacturing Co. producing chips designed by leading U.S. technology companies (e.g., AMD, NVIDIA, and Apple) and Taiwan’s firms being leading customers of American-made semiconductor manufacturing equipment. Preserving access to Taiwan’s foundry ecosystem is a fundamental condition for continued U.S. leadership in semiconductors. Without even taking into account security and diplomatic commitments to Taiwan, calibrating the U.S. approach to the island in light of its strategic importance to the technology industry will be crucial.
There have been two promising developments in terms of increasing U.S. semiconductor manufacturing capacity. First, the U.S. government has reached an agreement with Taiwan Semiconductor Manufacturing Co. to build a new advanced facility in the continental United States with the company announcing plans for a new facility that will target five-nanometer production in Arizona. However, by the time of completion this plant will be well behind leading foundries in Taiwan, and likely insufficient to provide full coverage for commercial and Department of Defense needs in microelectronics and semiconductors in the future. Second, in March 2021, Intel announced its intent to move into the foundry business to serve external customers, with plans to build two additional fabs in Arizona for advanced semiconductor manufacturing, representing an evolution in its business model. While these steps are encouraging and will bolster supply chain resiliency, more needs to be done to ensure continued U.S. leadership.
Congress has awoken to the situation. Representatives from both parties put forward the Creating Helpful Incentives to Produce Semiconductors for America Act, better known as the CHIPS Act, and the American Foundries Act in 2020, which were consolidated into the Fiscal Year 2021 National Defense Authorization Act under Title XCIX and passed into law. The bills enjoyed broad, bipartisan support in Congress. While Congress still needs to appropriate funding for the legislation’s initiatives, the authorizations represent important steps in foundational investment to restore U.S. fabrication capacity and research and development advantage in semiconductors. Critics argue that a sluggish and potentially miserly U.S. response hinders the benefits of the bill in offsetting the significant domestic incentives offered by U.S. competitors, especially China. More investment, grants, and incentives (such as a proposed tax credit) will likely be required to offset the costs of building additional advanced facilities in the United States, but ideological concerns about government involvement in industrial policy could trigger a reduction in appropriations.
Here the European Union has taken what serves as an instructional step. Brussels is making a push toward an advanced two-nanometer production node using up to 145 billion euros ($175 billion) from the EU pandemic recovery fund (Next Generation EU) set aside for “digital transition.” Likewise, the Biden administration and Congress should allocate a portion of the proposed Endless Frontier legislative package and/or the national infrastructure bill to fund CHIPS Act provisions. Finally, Congress and the Biden administration might also consider expanding incentives to support firms’ efforts to diversify operations out of China toward South and Southeast Asia, in line with similar efforts from countries like Australia, Japan, South Korea, and Taiwan.
Expanding the Trusted Foundries Program with “Zero Trust”
To ensure reliable access to semiconductors for the defense industrial base, the United States needs to revise and expand its Trusted Foundry program, which accredits firms for contributing to the ecosystem of semiconductors utilized in weapons systems, sensors, operations, and communications. The challenge in procurement has long been that while the capabilities that semiconductors provide are essential to advanced U.S. defense capabilities, the Department of Defense is not such a large customer that larger U.S. firms could justify restructuring their production to accommodate such a relatively small set of orders. Efforts to revise the Trusted Foundries program to craft a next-generation global process of procurement and review to procure secure, off-the-shelf technology are welcome but should be sustained and fully fleshed out to protect U.S. security interests.
While many current defense systems leverage older semiconductors and microelectronics, the United States faces a challenge in investing in a strategic industry like semiconductors to ensure trusted supply for defense applications. Over recent decades, in several industries connected to defense such as telecommunications technology and commercial satellites, the United States has squandered long-held advantages, which has implications for access to critical technology. Without genuine effort now, America could face a similar challenge in the semiconductor sector.
As part of its incorporation into the National Defense Authorization Act, the CHIPS Act addresses some of these areas by directing the Department of Defense to form a group of U.S. firms to build advanced, secure microelectronics, but additional steps are required. In 2020, the Department of Defense announced an effort to move to a “zero trust” approach in microelectronics and cyber security, requiring hardware to be validated after receipt to ensure reliability and security. A zero trust approach for semiconductors, requiring verification and validation after procurement, would allow for the flexibility to incorporate products from allies and partners, including South Korea, Japan, and Taiwan. Steps need to be taken to ensure that foreign products from trusted partners are eligible for procurement and that U.S. firms have corresponding reciprocal access among allied countries. How the zero trust policy will be implemented by the Biden administration is still unclear, but shifting the department’s acquisition cycle to modern chips will be an essential benchmark. Given the lag between policy implementation and secure chips coming off a fabrication line, expanding the Pentagon’s secure semiconductor options will be paramount.
Ensuring a Robust Semiconductor Sector Workforce
Training and developing human talent to drive the semiconductor industry forward remains a significant hurdle for the United States and its allies. This problem is indicative of the broader shortfall the United States faces in producing enough trained science, technology, engineering, and mathematics personnel to support continued U.S. leadership in commercial and defense technology fields. For instance, 81 percent of full-time graduate students in U.S. university electrical engineering programs, and 79 percent in computer science, are international students. Nested within the overarching problem, the United States confronts choices over how to expand and incentivize training for engineers, materials scientists, and computer scientists essential to the field. On the training side, potential building blocks include public-private partnership training programs, such as a new initiative by SEMI, the global industry association representing the electronics manufacturing and design supply chain, and the State University of New York Polytechnic Institute (funded by the National Science Foundation) to establish a new semiconductor manufacturing curriculum. The government should consider investing seed money for additional public-private partnerships through higher-learning institutions, as well as redoubled efforts to attract students to science, technology, engineering, and mathematics disciplines.
As many computer science graduates hail from overseas, shaping U.S. immigration policy to continue to accommodate international students and facilitate their living and working in the United States will be a crucial bellwether for long-term success. According to the Center for Security and Emerging Technology, 40 percent of high-skilled semiconductor workers in the United States were born abroad. Failing to retain highly skilled engineers and computer scientists would severely hamper U.S. competitiveness in the industry, but these efforts face obstacles and concerns over the volume of U.S. immigration. Fashioning durable legal, education, legislative, and immigration pathways to secure top-tier talent will be imperative if the United States wishes to maintain its advantage in the industry.
Creating Effective Export Control Policies
Three countries — the United States, Japan, and the Netherlands — control the narrowest chokepoint in the supply chain in terms of tools (etching, deposition, and extreme ultraviolet lithography equipment) and design software used to manufacture advanced semiconductors, particularly for seven-nanometer and smaller nodes. In extreme ultraviolet lithography technology, Dutch firm ASML is currently the global leader, with only two integrated device manufacturers — Intel and Samsung — and one pure-play foundry, Taiwan Semiconductor Manufacturing Co., operating extreme ultraviolet lithography machines, which cost about $150 million per machine.
China lags far behind in semiconductor manufacturing equipment. The Chinese state-backed foundry Semiconductor Manufacturing International Corp. is trying to purchase extreme ultraviolet lithography equipment from ASML, and the Dutch company and government has faced pressure from the United States to block the sale. To formalize a multilateral approach that blocks transfer of advanced semiconductor manufacturing equipment will require the United States and its allies to frame a new minilateral export control arrangement. Some of the relevant parties could be persuaded to add the equipment to the Wassenaar Arrangement, but this may be more challenging and does not include most of the relevant partner countries. Fashioning a new regime for semiconductors and other emerging technologies will require the expenditure of significant diplomatic capital and resources from the United States.
Blocking the sale of the equipment wouldn’t necessarily prevent Chinese fabless companies (companies that design chips, but contract their manufacture to dedicated foundry companies) from developing advanced products for manufacture on others’ hardware or making breakthroughs in miniaturization. It would, however, severely curtail Chinese efforts to enhance the capabilities of domestic foundries (such as Semiconductor Manufacturing International Corp., which the Trump administration added to a list of companies controlled by the Chinese military) to scale advanced products quickly and cost effectively. On the other hand, it would negatively impact the sales of equipment from U.S., Japanese, and European firms, and the tools themselves are less effective without know-how and a suite of support from those companies.
Another crucial area of consideration is the potential for the U.S. executive branch to leverage U.S. Department of Commerce de minimis and foreign direct product rules under the expanded authority from the Foreign Investment Risk Review Modernization Act and Export Control Reform Act. The focus in this area so far has been to curtail Huawei’s access to leading chips due to espionage concerns and how to appropriately limit the People’s Liberation Army’s access to advanced semiconductors. The Trump administration leveraged the foreign direct product rule and entity listing to implement rules to block transactions (independent of a U.S. company) to Huawei and its fabless subsidiary, HiSilicon, that incorporate U.S. semiconductor technology. Limiting Huawei’s access to these chips (and foundries that manufacture them) by requiring licenses for any transaction represented a decisive step from the Trump administration.
However, such actions blocking the sale to or transactions with many Chinese companies negatively impact revenue for U.S. firms and other global firms, like Taiwan Semiconductor Manufacturing Co., that make chips for HiSilicon and Huawei. Firms, including U.S. companies, have argued they need the revenue from the Chinese market to invest in research and development to maintain their industry position. Given the importance of AI chips, there is also discussion of specifically targeting a certain subset of chips, but there is a lack of specificity about what type of chips (i.e., edge/cloud, application-specific integrated circuit, graphics processing unit, field-programmable gate array) or what threshold to follow beyond blocking transfers to the Chinese military. Moreover, Intel and AMD both successfully applied for licenses to continue selling chips to Huawei in September 2020. This complicates matters, particularly for Korean, Japanese, and Taiwanese firms, which are required to abide by the U.S. ruling but may lack influence with the U.S. government necessary to receive a new license. (In the final days of the Trump administration some of these licenses were canceled once again by the Department of Commerce.) In addition, overly broad, unilateral export controls are among the most disruptive instruments to technology supply chains and can constrain innovation in the sector. Biden administration officials, and specifically the Bureau of Industry and Security in the Department of Commerce, need to weigh these competing considerations carefully in deciding on future U.S. export controls for semiconductor technology.
Expanding Cooperation with Allies
The Trump administration made strides in drawing attention to the challenge that China raises and imposed export controls to mitigate the effect of China’s mercantilist policies, but it failed to effectively leverage U.S. allies. All of the aforementioned issues — including export controls, the advanced foundry deficit, human capital, research and development, supply chain resiliency, and expanding sources for procuring military hardware — represent areas where the Information Technology and Innovation Foundation has argued there is potential to collaborate in concert with allied partners. On the commercial side, incorporating allies has the benefit of insulating U.S. companies from negative market pressures, and on the security side, it ensures that ecosystems for intelligence and defense applications are compatible. For export controls, allied buy-in makes controls more likely to be durable and effective. Pooling research and development funding and promoting collaboration among allies and partners, such as the T-12 democracies (the United States, United Kingdom, France, Germany, Canada, Australia, Japan, South Korea, Finland, Sweden, India, and Israel), also provides clear benefits to nurturing the semiconductor innovation ecosystem. Finally, the United States should exercise leadership and cooperate with partners at the World Trade Organization negotiations on electronic commerce and on another expansion to the Information Technology Agreement, or ITA-3, to include additional new semiconductor and microelectronic products, in order to ensure market access and low tariffs for U.S. and global semiconductor products.
Maintaining its leadership in both semiconductor design and production represents an essential imperative for continued U.S. economic preeminence and national security. The Biden administration will need to effectively navigate the opportunities and challenges briefly explored here if the United States is to remain a world-leading semiconductor player and if competition across the broader global industry is to unfold on private enterprise-led, market-based, rules-governed terms.
Jeffrey D. Bean focuses on the nexus of geopolitics, trade, technology, security, and nontraditional threats, with regional expertise in the Indo-Pacific and Europe. He was visiting fellow at East-West Center in Washington in 2020, researching supply chain disruption for emerging technologies. He was previously editor of the Center for Strategic and International Studies’ Asia Policy blog and podcast.
Stephen Ezell is vice president, global innovation policy, at the Information Technology and Innovation Foundation. He focuses on science and technology policy, international competitiveness, trade, manufacturing, and services issues. Stephen is the author of two recent studies on the semiconductor industry, An Allied Approach to Semiconductor Leadership and Moore’s Law Under Attack: The Impact of China’s Policies on Global Semiconductor Innovation.
Image: U.S. Department of Energy