Risks in the Semiconductor Manufacturing and Advanced Packaging Supply Chain

This commentary is a lightly edited version of a comment submitted to the U.S. Department of Commerce.

Western companies benefited for decades from inserting China into their supply chains and by selling to the Chinese market. This occurred during a long period of good relations. But that period ended with the arrival of President Xi Jinping. The United States now finds itself in competition with China because of decisions by Xi’s government.

This September 2019 CSIS report discussed broad policy for managing technology transfer to China. Semiconductors are a technology focal point for this competition. The United States has sought for years to maintain superiority in semiconductor technology over China. China, in turn, has repeatedly launched major investment programs (accompanied by the usual industry espionage) to develop an indigenous industry. Another CSIS report details those efforts.

China’s Technological Dependence

While China has repeatedly failed to develop an indigenous semiconductor industry, it is persistent, well resourced, and continues to try. Until recently, China was following the same path as other semiconductor producers such as Japan and South Korea in building a domestic industry, starting in less advanced sectors (usually memory chips) and progressing to more advanced chips. U.S. sanctions slowed China’s progress, and it will likely take a decade for China to create an indigenous industry. For now, China is dependent on Western suppliers for “fabless” production services, most chips, and semiconductor manufacturing equipment (SME).

China will eventually succeed in building an indigenous industry, but U.S. decisions can affect the pace of this. China is very unlikely to end predatory commercial policies or adopt a less confrontational international stance. This argues for greater restriction, but restriction must be balanced against the need to protect (as far as possible without damaging national security) the robustness of the U.S. semiconductor industry and that of its allies. As an aside, China is unlikely to observe any agreements on trade or cybersecurity absent strong verification and compliance mechanisms and multilateral support, so they are not worth pursuing at this time.

Given China’s behavior and intent, further decoupling is inevitable, and semiconductors are a good place to start. This makes for a complex policy decision, balancing hampering China versus avoiding harm to U.S. industry. We should be clear that without changes in China’s behavior, the long-term effect of a failure to impose export restrictions will be to damage the U.S. semiconductor industry and national security. The general direction for policy should be increased export restrictions.  

Technology Transfer Risk

Since China is still largely dependent on Western sources of supply (although it is striving to end this dependence), managing technology transfer to China is an important strategic task. National security risk created by trade with China is easy to overstate given that the United States has already banned sales of arms and munitions to that country. Dual-use exports to the Chinese military are also restricted. Companies have responded to the political risk created by the contest with China by reducing their exposure while attempting to preserve access to the China market. China’s consumer market and its official programs create immense demand. The United States can increase perceptions of political risk to incentivize companies to diversify out of China by using additional restrictions and subsidies. If the United States pursues complete decoupling, other countries will be strongly tempted to substitute their producers when possible. This points to the need for multilateral agreement as a pacing factor in deploying further restrictions. 

A counterfactual argument would ask what would happen if the United States does not restrict exports to China. In this, Huawei becomes the poster child, a company supported by the Chinese state for strategic benefit, one result of which would be to destroy Western competitors. China’s economic goals are to displace foreign competitors and achieve market dominance for its companies. Its failure to comply with its World Trade Organization commitments for more than two decades suggests that, absent penalties for noncompliance, any agreements are of limited value. China’s technological base has improved rapidly, but it is still dependent on Western technology and know-how. Removing restrictions would accelerate China’s pursuit of its commercial and political ambitions and harm the United States. China uses technological and commercial espionage to support its progress and removing restrictions would increase China’s ability to acquire intelligence. Additional restrictions make strategic and economic sense.

However, a complete and immediate embargo on trade and technology would damage the United States as well as China. For commodity chips (as defined in the export administration regulations), there is little risk in allowing these sales to continue. Denying more advanced chips, such as those necessary for 5G infrastructure, hampers China’s ability to build a global signal collection system and demonstrate the falseness of the claims of Chinese companies that they alone can make 5G. The only way to justify continued exports is to cushion the effect of reduced revenues on U.S. companies.

Chips

This CSIS report from May 2020 discusses semiconductor exports to China. It cannot yet make advanced semiconductors and is spending billion to build its own industry. The U.S. decision about whether or not to allow sales is irrelevant to this. Continued sales will not disincentivize China. The debate over Huawei’s indigenous capabilities of a few years ago has largely ended; Huawei is dependent on U.S. sources of supply. A clear breakpoint would be to allow continued chip exports to Huawei for earlier generations of telecom equipment—3G and below. Export of chips that support 5G should not be allowed. Chips for 4G are an area of decreasing risk as the technology matures and is replaced by 5G, but concessions on allowing continued export of Chinese 4G products should be made in exchange of other considerations, such as agreements with allies (like Germany, which is still deploying 4G) on a broader China policy.

In practical terms, the divisions among generations of technology are not as clear. But U.S. policy can be based on considerations of the level of sophistication of the chips in questions, what China itself can make, and what it can obtain from other suppliers, all factors the Bureau of Industry and Security (BIS) is well placed to evaluate. Less advanced commodity chips can still be safely transferred to China with the goal of allowing U.S. companies to continue to earn revenues for as long as possible. The Department of Commerce should be able to distinguish in its regulation between those chips that can safely be sold and those that should be denied.

SME and Materials Are a Chokepoint

SME is a more difficult problem. Denying China access to SME will greatly slow its chip effort, but the loss of revenue for U.S. companies will be painful and, according to the companies, reduce their ability to conduct necessary R&D. Barring further deterioration in the relationship, the United States should allow only limited transfers of less advanced SME to China to continue. This will not satisfy all producers, but since China’s intent is to displace the United States and its companies, U.S. national interest points in the direction of increasing restrictions undertaken in cooperation with key allies.

Chips and SME are dual-use goods, but the nature of the conflict with China means that SME exports should be treated with the same scrutiny and restraint we apply to aerospace or other advanced and emerging technologies. China’s willingness to spend billions of dollars in building its own semiconductor industry makes it the primary market for SME.

The strategic questions are (1) whether the United States, in a situation where China is not an honest competitor, should let its companies help China build a competing industry, and (2) how to minimize the harm to U.S. companies if the China market is closed before China is ready to do so on its own. SME is a logical chokepoint, but it would be useful to find ways to mitigate the pain for U.S. suppliers and compensate them in some way for their losses. 

A rough breakdown of the SME market would show the United States with a 50 percent market share, Japan with a 25 percent market share, and various other suppliers (such as the Netherlands’ ASML) making up the rest. SME is scientifically intensive, as the most advanced equipment is at the cutting edge of physics. There are few alternative sources of supply for cutting-edge SME (although China has in the past taken advantage of a “second-hand” market for used SME made available as companies upgrade their production). This means that denying access to SME hampers the ability of China to build a competing industry.

SME is the centerpiece of a complex, globally distributed supply chain that also includes producers of the materials and gasses needed for chip manufacturing; researchers developing new materials, new processing techniques, and new chip architectures; and the actors involved in each stage of the manufacturing process: chip design, silicon wafer production, fabrication (where a chip design is drawn on the wafers), packaging, and testing. It is globally distributed because that offers economic efficiency, and there is an advantage for producers in being located close to customers. Giving its complexity, any interference with this supply chain should be done cautiously and counterintuitively. Premature or badly designed export restrictions could undercut resilience. 

China, given its multibillion-dollar government programs to develop a chip industry, has become the leading market for SME producers. China cannot make SME itself. A more restrictive policy makes sense, tempered by the question of whether the harm to the U.S. economy and technological leadership from denying a sale outweighs the security benefits to the United States. The strategic implications of this are not black and white. Cutting off SME exports to China would greatly slow its development, but it would also damage the U.S. SME industry if there was no counterbalancing compensation. If this would damage SME’s ability to invest in R&D, federal programs must compensate for this, using subsidies, workforce expansion, and cooperative research programs with allies.  

A decision to block exports of SME to China would be supported by Japan but would face reluctance in Europe. Europe is not yet ready for this step but is moving in the direction of increased scrutiny given its greater concern over China’s predatory behavior. It would also harm U.S. SME makers, as China’s massive investment surge to build indigenous chip production capacity has made it by far the largest market, one where companies are eager to sell. An intermediate position between unrestricted sales and embargo would be to block exports of the most advanced SME while allowing earlier generations of equipment to go and negotiating with Europe and Japan on adopting much stricter controls. 

Reshoring and Fabless Production

The importance of the semiconductor supply chain and the current temporary shortages of chips have led to calls for “reshoring” the chip industry and an expansion of domestic capacity. In this, some numbers on market share give a misleading perception. It is true that the U.S. share of semiconductor fabs has fallen to 10 percent of the global total. On the other hand, U.S. companies have a 45 percent share of the global chip market, almost double the share of South Korean companies (the next largest producer). Many of these U.S. companies rely on Taiwanese “fabless” producers as pat of a globally distributed supply chain that still includes China. This made commercial sense but no longer makes strategic sense.

Deliberate interference with key parts of the supply chain that are located in Taiwan could be tempting for China, since Taiwan is easily within range of Chinese forces and China asserts that its political status is ambiguous. Interfering with South Korea or Japan (a very risky prospect) would face both operational and political constraints. This makes Taiwan a central concern, and U.S. semiconductor policy should, without undercutting Taiwanese industry, seek to diversify and encourage Taiwanese companies to open facilities in the United States. In doing so, the United States could study China’s effort to bring Taiwanese companies to locate some facilities in China using a combination of inducements and threats.  

In the previous era of globalization, it made sense to see production migrate to low-cost centers. China’s provocative behavior calls for a reassessment of risk, but this risk should not be generalized beyond China. Suppliers located in North America or in allies like Japan, Israel, and the Netherlands (or neutral countries like Ireland) face less risk of interference of any kind. “Vaccine nationalism” demonstrates the limits of open globalization, but this is an unusual instance of scarcity, and Covid-19 has challenged the political management of many countries in ways that are not applicable to capital goods or commodities. Reshoring and renationalizing only make economic and strategic sense as part of a larger strategy to diversify the semiconductor supply chain and reduce reliance on China. 

Resilience from a Globally Distributed Supply Chain

While discussion has focused on the risks of a globally distributed supply chain, it also provides considerable resilience. The United States would be best served by reinforcing a supply chain distributed among allies where relocation is decided by commercial factors. The United States may find it useful to strengthen domestic companies’ fab capabilities (this requires subsidies) or to encourage Taiwanese companies to move facilities to the United States to better manage risk, but a supply chain distributed among allies is the best way to achieve resilience. To summarize, the United States would be best served if it:

• Allows less advanced chips (using thresholds defined by BIS) to continue to go to China for now;


• Blocks exports or re-exports of advanced technologies (again using thresholds defined by BIS). This will affect SME more than chips;


• Relies on a commercially driven, global supply chain that excludes China rather than renationalize or reshore production;


• Uses subsidies and other incentives to increase supply chain diversity, strengthen R&D (in light of potential revenue losses as exports to China are cut), and expand domestic production as part of a globally distributed supply chain independent of China; and
• Rethinks the reliance on Taiwanese fabless companies and finds ways to diversify the chip supply chain, including persuading companies to build in the United States, as a part of a larger Taiwan policy.

James Andrew Lewis is a senior vice president and director of the Strategic Technologies Program at the Center for Strategic and International Studies (CSIS) in Washington, D.C.

Commentary is produced by the Center for Strategic and International Studies (CSIS), a private, tax-exempt institution focusing on international public policy issues. Its research is nonpartisan and nonproprietary. CSIS does not take specific policy positions. Accordingly, all views, positions, and conclusions expressed in this publication should be understood to be solely those of the author(s).

© 2021 by the Center for Strategic and International Studies. All rights reserved.