An Allied Approach to Semiconductor Leadership

Stephen Ezell September 17, 2020
September 17, 2020
Many countries rightly seek to maximize their value added in the global semiconductor industry. But like-minded allied nations can also advance their leadership collectively by collaborating on technology and ecosystem development, intellectual property, and trade liberalization.
An Allied Approach to Semiconductor Leadership

The Global Semiconductor Industry

Global Semiconductor Value Chains

Countries’ Semiconductor Competitiveness Strategies

Collaborating to Collectively Enhance Semiconductor Competitiveness

Coordinated Technology Development

Coordinated Semiconductor Ecosystem Development

Coordinated Technology Protection

Advance Supportive Trade Rules, Regimes, and Practices

Conclusion

Endnotes

Semiconductors represent one of the world’s most important industries, the core technology that powers the modern digital world and empowers innovation and productivity growth across every industry. In turn, the evolution of global value chains has enabled the industry to sustain its relentless, multi-decades drive to produce ever-more powerful integrated circuits at ever-lower costs—a dynamic captured in Moore’s Law: the notion that the number of transistors on a microchip doubles about every two years, effectively meaning a semiconductor’s capability in terms of speed and processing is doubled, even though its cost is halved. However, recognizing semiconductors’ foundational role in the modern global economy, and their importance to national security, an increasing number of nations are seeking to capture as much value as possible from the industry, whether from semiconductor research and development (R&D); design; fabrication; or assembly, test, and packaging (ATP).

While national, including U.S., policies to spur semiconductor R&D and production are important, it’s also important to recognize that self-sufficiency cannot and should not be the goal. The increasing expense, complexity, and scale required to innovate and manufacture semiconductors means that no single nation can afford to go it alone. However, there exists an opportunity and a need for a like-minded set of nations committed to open trade and fair economic competition to collaborate in ways that collectively empower the competitiveness of their semiconductor industries.

This report begins by examining the global semiconductor industry, including the rise of semiconductor global value chains, and by examining nations’ semiconductor competitiveness strategies. It then examines how a community of allied nations can work together across four areas—by working collaboratively on semiconductor technology development, ecosystem support, and technology protection, as well as by developing supportive trade rules and regimes—to collectively enhance the competitiveness and innovation potential of their respective semiconductor industries and the industry globally.

The report makes the following policy recommendations:

Coordinated Technology Development

  • Establish Manufacturing USA Institute(s) supporting semiconductor industry innovation—in activities including R&D, manufacturing, and packaging—and invite participation by semiconductor enterprises headquartered in like-minded nations.
  • Expand international cooperation in semiconductor sector public-private partnerships.
  • The United States and like-minded nations should increase funding for collaborative, pre-competitive R&D efforts, and ensure that there is reciprocal opportunity for semiconductor enterprises from like-minded nations to participate in such consortia.
  • The U.S. government should work to more effectively coordinate the semiconductor R&D programs being conducted across various government agencies.
  • The U.S. government should explore authorizing more-flexible federal contracting guidelines, such as a relaxation of Federal Acquisition Regulations, or allowing greater use of other transactional authority vehicles, in order to increase the commercialization potential of federally funded semiconductor R&D research programs.
  • The U.S. government should invite other allied nations to co-invest in semiconductor moonshots, with resulting intellectual property (IP) and technical discoveries shared at levels proportionate to national mutual investment.
  • The United States should explore additional opportunities to enroll peers from allied nations in the trusted foundries programs, with allied nations acting reciprocally for their related programs.
  • Like-minded nations should amend their procurement guidelines by adding a fourth key pillar—security—in addition to the traditional standards of price, cost, and quality.

Coordinated Semiconductor Ecosystem Development

  • Like-minded nations should continue to advocate for open standards-development processes, both as they relate to semiconductors specifically and to the vast panoply of downstream digital technologies fundamentally predicated on semiconductors, such as 5G, artificial intelligence (AI), the Internet of Things, and autonomous vehicles.
  • Like-minded nations and enterprises therein should collaborate to develop a fundamentally more-secure computing infrastructure.

Coordinated Technology Protection

Export Controls

  • U.S. export controls must be regularly updated to reflect the global state of play in semiconductor industries, such that controls do not preclude U.S. enterprises’ ability to sell goods that are on a technical par with commercially available goods and services from foreign competitors.
  • Any emerging technologies that are ultimately deemed to meet the statutory standards for export controls should be designated as such only in cases of exclusive development and availability within the U.S. market—and the controls should be removed if and when that exclusivity no longer exists.
  • The United States should eschew the application of unilateral export controls and seek to develop a more ambitious and effective plurilateral approach to promulgate export controls among like-minded nations that have indigenous semiconductor
    production capacity.
  • Congress should expand the remit and funding for the Export Control and Related Border Security (EXBS) Program at the U.S. Department of State.
  • At the 2020 Multilateral Action on Sensitive Technologies (MAST) conference, scheduled for September 2020, the United States should consider introducing a plurilateral approach to advanced-technology export controls.

Foreign Investment Screening

  • The United States should work with like-minded nations to align foreign investment screening practices and to exchange information when it appears other nations are trying to use unfair practices in making foreign investments, such as heavily state-subsidized, state-owned enterprises (SOEs) attempting to purchase foreign enterprises in advanced-technology industries.
  • The United States should continue to work with like-minded nations to coordinate investment screening procedures, and it should consider expanding its list of “excepted foreign states” to include countries such as France, Germany, the Netherlands, Italy, Japan, and South Korea (among others).

Cataloging and Combatting Foreign Technology and Intellectual Property Theft

  • Like-minded nations should develop a comprehensive list of enterprises and individuals who have attempted or affected IP theft, and develop mechanisms to restrict such firms and individuals from competing in like-minded nations’ markets.
  • Like-minded nations should enhance information-sharing efforts to combat foreign economic espionage and IP/technology/trade secret theft.
  • The United States should lead like-minded nations in developing stronger information-sharing mechanisms focused on combatting state-sponsored economic espionage in advanced-technology industries.
  • The United States should continue to work with like-minded nations to strengthen their trade secret protection regimes.
  • The United States and other like-minded nations should continue to include robust trade secret protections, and penalties for willful large-scale commercial trade secret theft, in trade agreements they pursue.

Supportive Trade Policies, Regimes, and Practices

  • Elevate the imprimatur and stature of the World Semiconductor Council (WSC).
  • Expand the Information Technology Agreement (ITA).
  • Maintain the World Trade Organization (WTO) e-commerce customs duty moratorium.
  • With like-minded nations, join and expand the Comprehensive and Progressive Trans-Pacific Partnership (CPTPP) Agreement.
  • Expand subsidies disciplines at the WTO.
  • Insist on market access reciprocity in digital government procurement activity.
  • Consider forming a Global Strategic Supply Chain Alliance (GSSCA).
  • Develop an allied approach to expand market-based trade approaches in the
    Indo-Pacific region.

The Global Semiconductor Industry

The term “semiconductor” actually refers to a solid substance—such as silicon or geranium—which has electrical conductivity properties allowing it to be used either as a conductor or an insulator. In 1956, Bell Labs’ John Bardeen, Walter Brattain, and William Shockley won a Nobel Prize for their 1947 invention of the transistor, a semiconductor device used to amplify or switch electronic signals and electrical power. In the mid-1950s, Jack Kilby at Texas Instruments and Robert Noyce and a team of researchers at Fairchild Semiconductor pioneered the integrated circuit, placing multiple transistors on a single flat piece of semiconductor material, giving rise to the modern visage of a “semiconductor chip.” But modern semiconductors are a far cry from those invented by the early pioneers; today, they contain billions of transistors on a chip the size of a square centimeter, circuits are measured at the nanoscale (“nm,” a unit of length equal to one millionth of a meter), and the very newest semiconductor fabrication facilities are producing semiconductors at 5 nm and 3 nm scales.[1] Leading-edge semiconductors contain transistors that are 10,000 times thinner than a human hair.

The increasing miniaturization of semiconductors alongside performance enhancements in both processing capacity and speeds as well as power efficiency lie at the core of every single information and communications technology (ICT) product. It’s principally the evolution of semiconductors that explains the ever-increasing capability at an ever-decreasing relative price of digital products—everything from cells phone costing $4,000 in 1983 to just a few hundred dollars today to the cost of personal genome sequencing dropping from $2.7 billion to $300 over the past 20 years to the emergence of new “G’s” in wireless communications about every decade.[2]

Semiconductors represent one of the world’s most important industries, the core technology that powers the modern digital world and empowers innovation and productivity growth across every sector of every economy.

Harvard economist Jon Samuels estimated that total factor productivity in the U.S. semiconductor sector grew at close to 9 percent over the period from 1960 to 2007 (25 times the rate for the overall economy) and to have accounted for nearly 30 percent of the United States’ aggregate economic innovation over this period.[3] In terms of industry-specific contributions, from 1960 to 2007, semiconductors accounted for about 37 percent of the growth in the U.S. communications equipment manufacturing industry, 14 percent of the expansion of the electrical equipment and appliances sector, and 24 percent of the growth in output among other electronic products.[4] Oxford Economics estimated that the semiconductor industry helps create $7 trillion in global economic activity and is directly responsible for $2.7 trillion in total annual global gross domestic product (GDP).[5] And with the digital economy now accounting for nearly one-quarter of global GDP, semiconductors power the future of digitalization, underpinning everything from AI, cloud computing, and the Internet of Things to advanced wireless networks, smart grids, smart buildings, and smart cities, and even the next generation of quantum computing.[6]

The semiconductor industry itself represents a $470 billion highly globalized industry (expected to become a $730 billion industry by 2026) that shipped over 1 trillion semiconductors for the first time ever in 2019, with some of these processors containing over 30 billion transistors.[7] In 2019, U.S.-headquartered semiconductor enterprises held a 47 percent market share of global semiconductor industry sales, followed by South Korean firms with 19 percent, Japanese and European firms each with 10 percent, Taiwanese firms with 6 percent, and Chinese enterprises with 5 percent. (See figure 1.)

Figure 1: 2019 Global semiconductor industry sales market share[8]

 

However, the picture is very different when it comes to value added (the value of actual production in a nation). In 2016 (the most-recent year for which data is available), China produced $120 billion in value added, compared with $83 billion for the United States, $55.6 billion for Taiwan, and $39 billion for South Korea. (See figure 2.) U.S. value added in the sector peaked at $91.3 billion in 2011 (values in nominal dollars). China’s value added in the sector increased three-fold from 2007 to 2016. In terms of share of global value added in the semiconductor industry, from 2001 to 2016, China’s grew almost four-fold, from 8 to 31 percent, while the United States’ share fell from 28 to 22 percent, and Japan’s share fell by over two-thirds, from 30 to 8 percent. Taiwan and South Korea both saw their shares double or almost so, with Taiwan’s share growing from 8 to 15 percent and South Korea’s growing from 5 to 10 percent. Germany and Malaysia maintained shares of 2 percent each. (See figure 3.)

Figure 2: Value added ($ billions) of semiconductor industry by economy, 2001–2016[9]

Figure 3: Country share of value added in global semiconductor industry, 2001 and 2016[10]

According to Organization for Economic Cooperation and Development (OECD) data, in 2018, China exported $138 billion in semiconductors, Taiwan $111 billion, South Korea $92 billion, Singapore $87 billion, the United States $53 billion, the EU-27 and United Kingdom combined $53 billion, and Japan $48 billion. (See figure 4.) In terms of trade balances, in 2019 (or the most recent year in which data is available for that country), Taiwan recorded a trade surplus of $54 billion, South Korea $47 billion, Japan $21 billion, Singapore $18 billion, and the United States $2 billion.[11] Conversely, in 2018, India recorded a $14 billion semiconductors trade deficit, the EU-27 countries and the United Kingdom an $18 billion one, and China a $235 billion deficit. (See figure 5.) However, it’s important to note that while China’s semiconductor trade deficit might appear quite substantial, the reality is that about half of these semiconductor imports were re-exported—with value added during assembly and manufacturing—from China as part of global production networks for cell phones, tablets, and other electronic products (one reason why China’s semiconductors trade balance is no justification for it seeking autarky in semiconductor production).[12] China’s trade deficit in semiconductors grew significantly during a time when its trade surplus in electronics goods (e.g., computers, cell phones, etc.) also grew significantly, accounting for 58 percent of the value of total exports.[13]

Figure 4: Semiconductor exports by country ($ billions), 2019 or most recent year available[14]

Figure 5: Semiconductor trade balances by nation, 2000–2019 ($ billions)[15]

Semiconductors represent the world’s second-most R&D-intensive industry, after biopharmaceuticals. In 2018, U.S.-headquartered semiconductor companies invested 16.4 percent of their sales in R&D, compared with 15.3 percent on average for European-headquartered companies, 10.3 percent for Taiwanese firms, 8.4 percent for Japanese firms, 8.3 percent for Chinese companies, 7.7 percent for Korean firms, and 5.6 percent on average for semiconductor companies from all other nations.[16] (See figure 6.) Of the 12 most R&D-intensive semiconductor companies in the “2019 EU Industrial R&D Investment Scoreboard” report, half hail from the United States, and the top three most R&D-intensive companies are Qualcomm, which invests one-quarter of its revenues back into R&D annually, followed by Taiwan’s MediaTek with 24.2 percent, and America’s Advanced Micro Devices (AMD) with 22.1 percent. (See table 1.) In terms of actual investment, Samsung leads with €14.8 billion (approximately $17.6 billion) invested in R&D in 2019, followed by Intel with €11.8 billion ($13.7 billion).[17]

Figure 6: National semiconductor industry R&D intensity, 2019[18]

Table 1: Leading semiconductor investors on the 2019 EU Industrial R&D Investment Scoreboard[19]

The industry is also highly capital intensive; in fact, the U.S. semiconductor industry’s global gross capital expenditures (CapEx) reached $31.9 billion in 2019, making the industry’s capital expenditures as a percentage of sales, at 12.5 percent, second only to America’s alternative-energy sector.[20] In 2019, Korean-headquartered enterprises invested 31 percent of global CapEx in the sector, followed by U.S. companies with 28 percent, Taiwanese firms with 17 percent, Chinese firms with 10 percent, Japanese firms with 5 percent, and European firms with 4 percent. (See figure 7.)

Figure 7: Countries’ headquartered-enterprise’s percent share of global semiconductor industry capital expenditures, 2019[21]

The industry must be so R&D- and capital-intensive because innovating in the semiconductor industry requires increasingly complex chip designs at ever-smaller scales, especially if the industry is to keep up with the vaunted Moore’s Law. And while some believed at the start of the last decade that the 28 nm threshold would herald the limit of Moore’s Law, materials-engineering breakthroughs over the past decade in extreme ultraviolet lithography (EUV), etching, and thin-film deposition have brought the current industry frontier to 5 nm, with fairly clear visibility into the processes needed to get to 3 nm, 2 nm, and even 1 nm sizes.[22] As Dan Hutcheson, CEO and Chairman of VLSI Logic explains, even at the 7 nm level, today’s semiconductors pack over 20 billion transistors on a single chip, working at tolerances that are 1/10th or smaller the size of the coronavirus.[23] Contributions from enterprises and researchers from a wide range of nations—including China, Germany, France, Japan, Taiwan, the Netherlands, South Korea, the United States, and the United Kingdom, among others—have been responsible for advances in improving device performance, lowering power consumption, and shrinking size, reflecting the truly global nature of the industry.

However, whereas innovation in the sector historically was largely about doubling the processing power of chips while reducing or maintaining costs, the locus of innovation today is shifting and expanding, moving beyond mere processing speed to include energy consumption, “systems on a chip” functionality, and entirely new forms of technology and computing architectures. For instance, Silicon Valley-based Tachyum is working on a new “universal processor” microchip that would consolidate three types of microprocessors—the central processing unit (CPU), graphics processing using (GPU), and a tensor processing unit (TPU)—into a single chip, potentially delivering significant processing speed and power-consumption benefits.[24]

Yet the expertise, capital, and scale needed to develop a new semiconductor design, or build a new semiconductor fab, is extremely high, and increasing. For instance, an April 2020 study finds that the number of researchers required to achieve Moore’s Law (i.e., doubling of computer chip density) today is more than 18 times larger than the number required in the early 1970s.[25] This is one reason costs are increasing. In 2019, Taiwanese-manufacturer TSMC announced it would build a 5 nm fab in Arizona at a cost of $12 billion; in 2017, it had announced it was making plans to build a 3 nm fab in Taiwan at an anticipated cost of $20 billion.[26] As of 2020, it’s estimated that building a new 14–16 nm fabs costs, on average, $13 billion; a 10 nm fab $15 billion; a 7 nm fab $18 billion; and a 5 nm fab $20 billion.[27] Reflective of the increasing cost of competing in the sector, whereas almost 30 companies manufactured integrated circuits at the leading-edge of technology 20 years ago, only 5 do so today (Intel, Samsung, TSMC, Micron, and SK Hynix).[28]

The number of researchers required to achieve Moore’s Law (i.e., doubling of computer chip density) today is more than 18 times larger than the number required in the early 1970s.

Thus, the semiconductor industry represents a classic innovation-based industry characterized by extremely high fixed upfront costs of R&D and design, yet incremental costs of production (i.e., an individual chip comes off the production line at marginal cost). Moreover, the industry depends on one generation of innovation to finance investment in the next, so profits from the 10 nm fab beget the revenues to invest in the 7 nm fab, which make possible the 5 nm and 3 nm fabs of the future. As such, the ability of the global industry to sustain itself depends on several conditions attaining across the global economy. First, semiconductor companies need access to large global markets so they can amortize and recoup their costs across a single large global marketplace. Given the significant growth in fixed costs of R&D and capital equipment, the ability to access global markets is more important than it has ever been.

Second, semiconductor companies cannot face excess, non-market-based competition, such as governments pumping in hundreds of billions of dollars in subsidies, which unfairly disadvantages enterprises that are attempting to compete on genuinely market-based terms. In other words, if leading-edge companies cannot be assured that they can earn a reasonable, risk-adjusted rate of return on their investments—which is put in doubt by some governments such as China investing massive amounts of money to create a domestic semiconductor industry—then the leading companies will cut R&D and capital expenditures.

Third, because the industry fundamentally depends on knowledge, technology, and know-how, the international system must feature robust IP rights—including patents, trade secrets, and trademarks—for an extraordinary amount of the value is knowledge-based.[29] Again, if companies cannot retain that expensive IP, and it goes to competitors illegally and illicitly, their margins will go down, thereby reducing investment.

Finally, the industry relies on open and smoothly flowing global semiconductor value chains, as the following section elaborates.

Global Semiconductor Value Chains

The semiconductor industry has perhaps the most complex and geographically dispersed value chain of any industry in the world. In one stylized example, provided in the report “Beyond Borders: The Global Semiconductor Industry Value Chain,” large silicon ingots might be produced and cut into silicon wafers (the material used for producing semiconductors) in Japan; those bare wafers shipped to the United States to be transformed into fab wafers and cut into dies, on which the functional integrated circuit is etched to make a semiconductor; those semiconductor chips then being shipped to a country, such as Malaysia or Vietnam, where the semiconductor chips go through the ATP process; those chips then being exported to a country, such as China, South Korea, or the United States, to be integrated into end products such as tablets, mobile phones, or servers; and then those final consumer end products exported to the world.[30] (See figure 8.) In fact, the typical production process toward a final electronics product can see the underlying semiconductors within it cross international borders 70 or more times in a process that takes over 100 days and includes 3 full trips around the world.[31] One reason for this globalized supply chain is that unlike some industries such as cement, or even automobiles, with a high weight (and volume) to value ratio, semiconductors are small and light—and the costs of moving them around the globe is minimal compared with their actual value.

Figure 8: Stylized example of semiconductor value chain[32]