Tensions between Washington and Beijing have disrupted global semiconductor supply chains. The US seeks to promote research and development, manufacturing systems and export rules independent of the Chinese, squeezing them out in the medium term. In the long run, argues Chen Xi of Xi’an Jiaotong University and Peking University, the impact on China will be limited.
Credit: Kritsana Maimeetook / Shutterstock.com
China-US geopolitical and technological (geo-tech) competition has focused on the information technology sector. In particular, Washington has sought to starve China of essential semiconductors it needs to install in information and communication technology (ICT) equipment and systems. Meanwhile, a glut of orders due to booming ICT demand (including for a wide range of products from appliances to automobiles) and supply-chain disruption caused by pandemic have exacerbated a global chip shortage. All this has inspired both the US and China to become more independent in their own core supply chains.
In the US, for example, the Semiconductors in America Coalition (SIAC) was formed in May 2021 to advocate for policies that will promote semiconductor research and manufacturing in the US, easing reliance on imported chips. China, for its part, is focusing on boosting the quality of its semiconductors and its production capacity. This will require surmounting challenges including the need for basic sciences breakthroughs, technological innovation, and market opening.
Taiwan: Lynchpin of the global supply chain
Wafer fabrication accounts for 64 percent of the total capital expenditure (capex) in the semiconductor industry chain and requires the largest investment. Taiwan accounted for 20 percent of global wafer fabrication capacity in 2019, just ahead of South Korea (19 percent), Japan (17 percent), mainland China (16 percent), and the US (13 percent).
Logic chips for computing functions account for about 42 percent of all chip production capacity. Taiwan has obvious advantages in this segment, accounting for 92 percent of the production of those with individual transistors measuring 10 nanometers (nm) or less – the most advanced available technology, accounting for 2 percent of the global wafer fabrication capacity. These will support higher levels of autonomous driving, communication base stations, and high-performance computing. Mainland China and the US have no capacity for these 10 nm chips.
Taiwan is also ahead of its rivals in other logic-chip lines. It produces 28 percent of the 10-22 nm (8 percent of global wafer fabrication capacity), 47 percent of the 28-45 nm (9 percent), and 31 percent of the >45 nm (22 percent) logic chips, which support consumer and automotive electronics. Taiwan's production capacity is obviously critical for the global semiconductor industry chain, but its share of total world sales is very small – US$412 billion, or just 1 percent, in 2019.
South Korea is in the game
Expansion of semiconductor production capacity requires massive state subsidies and a significant pool of skilled talent, both of which the South Korea have. Memory chips for storage account for about 26 percent of all semiconductor production capacity and have a wide range of uses including in mobile phones, automotive electronics, computers and servers. South Korean firms (mainly Samsung and SK Hynix) account for 44 percent of the world's production capacity of these chips. South Korea also has global influence in logic-chip production capacity, with 8 percent of <10 nm, 5 percent of the 10-22 nm, 6 percent of 28-45 nm, and 10 percent of >45 nm. Samsung will invest about US$150 billion in non-memory chips over the next decade.
Production capacity is key for the US
American chip production capacity is not weak, contributing 5 percent of global memory chip capacity and 43 percent of the 10-22 nm range. But for the US, Taiwan and South Korean semiconductor production capacities must be a concern, given the political risks in the Taiwan Strait and the Korean Peninsula.
The US government hopes to see the transfer of manufacturing capacities to the homeland, which would directly promote job creation and tax revenue related to high-performance computing, automotive electronics, and consumer electronics. For example, South Korea’s Hyundai, Samsung, LG and SK will invest US$39.4 billion in the US to build R&D and manufacturing bases for electric vehicles and charging piles, wafer fabrication, batteries, and other projects. Semiconductor entrepreneurs in Taiwan are cooperating with the US government to hedge geo-tech competition risks. For them, the US market is as attractive as mainland China (North America accounts for 62 percent of TSMC's total sales). TSMC has invested in a high-end 5 nm wafer fabrication facility in Arizona, which will start production in 2023. United Microelectronics Corporation (UMC), which racked up US$6.2 billion in sales in 2020, sources 21.9 percent of its non-Taiwan revenue from the US and 20.2 percent from mainland China.
The US may learn from mainland China as it pursues a strategy to develop its semiconductor industry. In a report, the Washington-based Information Technology and Innovation Foundation concluded that “to the extent China’s mercantilist policies erode market share and reduce revenue for innovation leaders, they also reduce investment in the next round of innovation” and deter global capital from entering the market. Yet, it called on the US government to wield state power to ramp up investment in information technology infrastructure. The “Endless Frontier Act”, legislation introduced in the US Congress, could result in more than US$100 billion in investment in the development of artificial intelligence, high-performance computing, quantum information, robotics, communications, energy, cyber security, materials, biology and other areas. While the motivation behind the bill is the geo-technological competition with China, this race could end up driving significant scientific and technological progress in both countries.
Ramping up production capacity is not a challenge for China
For China, building production capacity is not the core challenge. Manufacturing to scale is not a problem due to its large pool of talented engineers. Mainland China's share of global semiconductor production capacity is expected to rise from about 15 percent by 2020 to 24 percent in 2030, while the shares of other producers are expected to be flat or to decline.
The core challenges for China lie along several parts of a long supply chain – for example, chip design (13 percent of total industry capex), materials (6 percent), equipment (3 percent), and Electronic Design Automation, or EDA, and Intellectual Property, or IP (less than 1 percent).
Taiwan companies will invest more in the mainland Chinese market to counterbalance moves by mainland competitors. The world’s third-biggest chip maker, Taiwan Semiconductor Manufacturing Company (TSMC), with sales of more than US$45 billion in 2020, 17 percent of which was to Chinese mainland, announced in April 2021 the expansion of its 28 nm Nanjing foundry with an investment of nearly US$3 billion to meet the growing demands of the local market. This is TSMC’s answer to the investment of US$7.6 billion for new 28 nm chip production capacity in Beijing in 2020 and US$2.3 billion in Shenzhen this year by rival Shanghai-based Semiconductor Manufacturing International Corporation (SMIC), which is partially state-owned.
Mainland China is clearly aware that it needs to boost innovation to build a semiconductor supply chain to match that of the US. The overarching obstacle to developing its domestic semiconductor industry is the lack of basic scientific talent. In most practical technology and engineering fields of higher education such as communications, transportation, energy and machinery, China leads the world. But higher education in mathematics, physics, chemistry, electrical engineering, computers and materials science, China still lags behind the US.
But mainland China is focused on remedying these knowledge deficiencies. A key part of its efforts to drive innovation are smart-city projects planned in Beijing, Shanghai and Shenzhen which will open up opportunities for solution developers across a wide range are sectors including energy, health and culture. This will accelerate the development of new technologies, materials and equipment including 6G, edge servers, satellite internet, vehicles powered with new sources of energy, and wearable devices – the meshing of digital technologies with green tech that will accelerate progress towards a low-carbon future. This is part of Beijing’s unrelenting efforts to raise per-capita national income and promote rural prosperity to increase total social consumption power. The aim is to provide high-quality public-service products for citizens and respond more quickly to social and natural crises at home and around the world.
Decoupling – and then a recoupling?
Manufacturing plants that supply wafers for logic and memory chips typically require US$20 billion in capital expenditure. Relying on the US market is not enough to make an acceptable return on investments. If lithography system manufacturers such as ASML halt exports to China, they will lose their competitive edge over the next decade due to lower R&D investing and the emergence of alternatives. Local businesses as well as companies in Europe or Asia will rush to meet the market opportunities left by companies exiting China.
Geopolitically, China has no desire to replace the US in global leadership. Beijing will engage modestly in global governance as a complement to the US-led global system. Economically, China will continue to innovate, open up and reform to benefit its own economy and the world. With China accelerating the expansion of its domestic production and a parallel supply chain gradually forming, the global semiconductor industry will reach the stage where two choices are on offer, two bets can be made. This would essentially be a decoupling. By around 2028, the size of the Chinese economy will surpass that of the US and more and more global enterprises will bet on China. The capital-rich semiconductor industry chain will once again link and merge with China’s – and the global supply chain will gradually stabilize.
That at least is the hope: that decades from now, this period of strategic competition between the US and China especially their decoupling in the technology sector including semiconductors was but a temporary phase in a reordering of the world. Eventually, the major powers will come to realize that global challenges such as climate and ecosystem change, public health, and inequality are the real issues – and that a chip war, supply-chain competition, and other trade spats were mere money- and time-wasting distractions.
Chen, Xi. (May 28, 2020) “Resolving the China-US Competition and the Greater Global Challenge”, AsiaGlobal Online, Asia Global Institute, The University of Hong Kong.
Chen Xi. (September 24, 2020) “What’s Next for the US-China Tech War”, AsiaGlobal Online, Asia Global Institute, The University of Hong Kong.
FP Analytics. (February 16, 2021) “Semiconductors and the US-China Innovation Race: Geopolitics of the supply chain and the central role of Taiwan”, FP Insider Special Reports, Foreign Policy, Washington, DC, USA.