
Using Targeted Industrial Policy to Address National Security Implications of Chinese Chips
Last year the Federation of American Scientists (FAS), Jordan Schneider (of ChinaTalk), Chris Miller (author of Chip War) and Noah Smith (of Noahpinion) hosted a call for ideas to address the U.S. chip shortage and Chinese competition. A handful of ideas were selected based on the feasibility of the idea and its and bipartisan nature. This memo is one of them.
Challenge and Opportunity
The intelligent and autonomous functioning of physical machinery is one of the key societal developments of the 21st century, changing and assisting in the way we live our lives. In this context, semiconductors, once a niche good, now form the physical backbone of automated and intelligent systems. The supply chain disruptions of 2020 laid bare the vulnerability of the global economy in the face of a chip shortage, which created scarcity and inflation in everything from smartphones to automobiles. In an even more extreme case, a lack of chips could impact critical infrastructure, such as squeezing the supply of medical devices necessary for many modern procedures.
The deployment of partially- or fully-automated warfighting further means that the deployment of Artificial Intelligence (AI) systems now has direct and inescapable impacts on national security. With great power conflict opening on the horizon, threats toward and emanating from the semiconductor supply chain have become even more evident.
In this context, the crucial role of the People’s Republic of China (PRC) in chip production represents a clear and present danger to global security. Although the PRC currently trails in the production of cutting-edge sub-16 nm chips used for the development of AI models, the country’s market dominance in the field of so-called “trailing edge chips” of 28 nm or above has a much wider impact due to their ubiquity in all traditional use cases outside of AI.
The most important harm of this is clear: by leveraging its control of a keystone international industry, the Chinese Communist Party will be able to exert greater coercive pressure on other nations. In a hypothetical invasion of Taiwan, this could mean credibly threatening the U.S. and other democratic countries not to intervene under the threat of a semiconductor embargo. Even more dramatically, given the reliance of modern military manufacture on digital equipment, in the case of a full-scale war between the People’s Republic of China and the United States, China could produce enormous amounts of materiel while severely capping the ability of the rest of the world to meet their challenge.
A secondary, but significant risk involves the ability of China to build defects or vulnerabilities into its manufactured hardware. Control over the physical components that underlie critical infrastructure, or even military hardware, could allow targeted action to paralyze U.S. society or government in the face of a crisis. While defense and critical infrastructure supply chains represent only a small fraction of all semiconductor-reliant industrial products, mitigation of this harm represents a baseline test of the ability of the United States to screen imports relevant to national security.
Beyond Subsidies: A Blueprint for Global Manufacturing
Wresting back control of the traditional semiconductor supply chain away from China is widely recognized as a prime policy goal for the United States and allied democratic countries. The U.S. has already begun with the passage of the CHIPS and Science Act in 2022, providing subsidies and tax incentives to encourage the creation of new fabrication plants (fabs) in the United States. But a strategic industry cannot survive on subsidies alone. Preferential tax treatment and government consumption may stand up to some degree of semiconductor manufacture. But it cannot rival the size of China’s if the PRC is able to establish itself as the primary chip supplier in both its domestic market and the rest of the world.
Nascent American foundries and the multinational companies that operate them must be able to survive in a competitive international environment without relying on unpredictable future support. They must do this while fighting against PRC-backed chip manufacturers operating with both a strong domestic market and massively developed economies of scale. Given the sheer size of both the Chinese manufacturing base and its domestic market, the U.S. cannot hope to accomplish this goal alone. Only a united coalition of developed and developing countries can hope to compete.
The good news is that the United States and its partners in Europe and the Indo-Pacific have all the necessary ingredients to realize this vision. Developing countries in South and Southeast Asia and the Pacific have a vast and expanding industrial base, augmented by Special Economic Zones and technical universities. America and its developed partners bring the capital investment and intellectual property necessary to kickstart semiconductor production abroad.
The goal of a rest-of-world semiconductor alliance will be twofold: to drive down the cost of chips made in the U.S. and its allies while simultaneously pushing up the cost of purchasing legacy semiconductors produced in China to meet it. Only when these two intersect will the balance of global trade begin to tip back toward the democratic world. The first two slates of policy recommendations will focus on decreasing the cost of non-China production and increasing the cost of Chinese imports, respectively.
Finally, even in the case in which non-Chinese-influenced semiconductors become competitive with those made in the PRC, it will likely be impossible to fully exclude Chinese hardware from American and allied markets. Therefore, the final raft of policy recommendations will focus on mitigating the threat of Chinese chips in American and allied markets, including possible risks of inbuilt cyber vulnerability.
The creation of an autonomous and secure supply chain entirely outside of China is possible. The challenge will be to achieve semiconductor independence in time to prevent China from successively weaponizing chip dominance in a future war. With clashes escalating in the South China Sea and threats across the Taiwan Strait growing ever more ominous, the clock is ticking. But America’s Indo-Pacific partners are also increasingly convinced of the urgency of cooperation. The policies presented aim to make maximum use of this critical time to build strategic independence and ensure peace.
Plan of Action
Recommendation 1. Boosting non-China Manufacturing
The first and most crucial step toward semiconductor sovereignty is to build and strengthen a semiconductor supply chain outside of China. No matter the ability to protect domestic markets from Chinese competition, U.S. industrial productivity relies on cheap and reliable access to chips. Without this, it is impossible to ramp up industrial production in key industries from defense contracting to consumer electronics.
According to a report released by the CSIS Wadhwani Center for AI and Advanced Technologies, the global semiconductor value chain broadly involves three components. At the top is design, which involves creating Electronic Design Automation (EDA) software, generating IP, and producing manufacturing equipment. Next is fabrication, which entails printing and manufacturing the wafers that are the key ingredient for finished semiconductors. The final stage is assembly test, and packaging, which entails packaging wafers into fully-functioning units that are fit for sale and verifying they work as expected.
Of these three, the United States possesses the greatest competitive advantage in the field of design, where American intellectual property and research prowess drive many of the innovations of the modern semiconductor industry. Electronic Design Automation software, the software that allows engineers to design chips, is dominated by three major firms, of which two, Cadence and Synopsys are American companies. The third, Mentor Graphics, is a U.S.-based subsidiary of the German industrial company Siemens. U.S. Semiconductor Manufacturing Equipment (SME) is also an important input in the design stage, with U.S.-based companies currently comprising over 40 percent of the global market share. The United States and Japan alone account for more than two-thirds.
Meanwhile, the PRC has aggressively ramped up wafer production, aiming to make China an integral part of the global supply chain, often stealing foreign intellectual property along the way to ease its production. Recent reported efforts by the PRC to illicitly acquire U.S. SMEs underscore that China recognizes the strategic importance of both IP and SME as primary inputs to the chip making process. By stealing the products of American research, China further creates an unfair business environment in which law-abiding third countries are unable to keep up with Chinese capacity.
Semiconductor Lend-Lease: A Plan for the 21st Century
The only way to the international community is to level this playing field. In order to do so, we propose that the United States encourage and incentivize its companies to license their IP and SME to third countries looking to build wafer capacity.
Before the United States officially entered into the Second World War, the administration of President Franklin Delano Roosevelt undertook the “Lend-Lease” policy, agreeing to continue to supply allied countries including Great Britain with weapons and materials, without immediate expectation of repayment. Recently, Lend-Lease has been resurrected in the modern context of the defense of Ukraine, with the United States and European powers supplying Ukraine with armaments and munitions Ukrainian industry could not produce itself.
The crucial point of Lend-Lease is that it takes the form of immediate donations of critical outputs, rather than simple monetary donations, which require time and investment to convert into the desired goods. World War II-era Lend-Lease was not based on a long-term economic or development strategy, but rather on the short-term assessment that without American support, the United Kingdom would fall to Nazi occupation. Given the status of semiconductors as a key strategic good, the parallels with a slow-rolling crisis in the South China Sea and the Taiwan Strait become clear. While in the long term, South, East, and Southeast Asia will likely be able to level up with China in the field of semiconductors, the imminent threats of both Chinese wafer dominance and a potential invasion of Taiwan mean that this development must be accelerated. Rather than industrial and munitions production, as in 1941, the crucial ingredients the United States brought to this process were intellectual property, design tools, and SMEs. These are thus the tools that should be leased to U.S. partners and allies, particularly in the Indo-Pacific. By allowing dedicated foreign partners to take advantage of the gains of American research, we will allow them to level up with China and truly compete in the international market.
Although the economics of such a plan are complex, we present a sketch here of how one iteration might look. The United States federal government could negotiate with the “Big Three” EDA firms to purchase transferable licenses for their EDA software. The U.S. could then “Lend-Lease” licenses to major semiconductor producers in partner countries such as Singapore, Malaysia, Vietnam, the Philippines, or even in Latin America. The U.S. could license this software on the condition that products produced by such companies will be made available at discounted prices to the American market, and that companies should disavow further investment from or cooperation with Chinese entities. Partner companies in the Indo-Pacific could further agree to share any further research results produced using American IP, making further advancements available to American companies in the global market.
When growing companies attain a predetermined level of market value they can offer compensation to the United States in the form of fees or stock options, which will be collected by the United States under the terms of the treaty and awarded to the EDA firms. Similar approaches can be taken toward licensing American IP, or even physically lending SME to countries in need.
Licensing American research to designated partner countries comes with some risks and challenges. For one, it creates a greater attack surface for Chinese companies hoping to steal software and design processes created in the United States. Preventing such theft is already highly difficult, but the U.S. should extend cooperation in standardizing industrial security practices for strategic industries.
A recent surge in fab construction in countries such as Singapore and India means that the expansion of the global semiconductor industry is already in motion. The United States can leverage its expertise and research prowess to speed up the growth of wafer production in third countries, while simultaneously countering China’s foreign influence on global supply chains.
A Semiconductor Reserve?
The comparison of semiconductors to oil is frequently made and has a key strategic justification: for more than a century, oil was a key input to virtually all industrial processes, from transportation to defense production. Semiconductors now play a similar role, serving as a key ingredient in virtually all manufacturing processes.
A further ambitious policy to mitigate the harm of Chinese chips is to create a centralized reservoir of semiconductors, akin to the Strategic Petroleum Reserve. Such a reserve would be operated by the Commerce Department and maintain centralized holdings of both leading- and trailing-edge chips, obtained from free dealings on the open market. By taking advantage of bulk pricing and guaranteed, recurring contracts, the government could acquire a large number of semiconductors at reasonable prices, sourced exclusively from American and partner nation foundries.
In the event of a chip shortage, the United States could sell chips back into the market, allowing key industries to continue to function with a trusted source of secure chips. In the absolute worst case of a geopolitical crisis involving China, a strategic stockpile would create a bulwark for the American defense industry to continue producing armaments during a period of disrupted chip supply. This buffer of time would be intended for domestic and allied production to ramp up and continue to supply security functions.
However, allowing the U.S. to participate in the chip industry has a further impact on economic development. By making the U.S. a first-order purchaser of semiconductors at an industrial scale, the United States could create a reliable source of demand for fledgling businesses. The United States could serve as a transitory consumer buying up excess capacity when demand is weak, ensuring that new foundries are both capable of operation and shielded from attempts from China to smother demand. The direct participation of the U.S. in the global semiconductor market would help to kickstart industry in partner countries while providing a further incentive to build collaboration with the United States.
Recommendation 2. Fencing in Chinese Semiconductor Exports
A second step toward semiconductor independence will be in containing Chinese exports, with the goal of reducing China’s access to global markets and starving their industrial machine.
The most direct approach to reducing demand for Chinese semiconductors is the imposition of tariffs. The U.S. consumer market is a potent economic force. By squeezing Chinese manufacturers seeking to compete in the U.S. market, the United States can avoid feeding additional production capacity that might be weaponized in a future conflict. These tariffs can take a variety of forms and justifications, from increased probes into Chinese labor standards and human rights practices to dumping investigations pursued at the WTO. The deep challenges of effective tariffs is how to enforce these tariffs once they come into play and how to coordinate tariffs with international partners.
Broad Tariffs, Deep Impact
No rule works without an enforcement mechanism, and in the worst case, a strong public stance against Chinese semiconductors that is not effectively implemented may actually weaken U.S. credibility and embolden the Chinese government. Therefore, it is imperative to have unambiguous rules on trade restrictions, with a strong enforcement mechanism to match.
These measures should not just apply to chips that are bought directly from China but rather include those that are assembled and packaged in third countries to circumvent U.S. tariffs. The maximal interpretation of the tariffs mandate would further include a calculated tariff on products that make use of Chinese semiconductors as an intermediate input.
In the case of semiconductors made in China but assembled, packaged, or tested in other countries, we suggest an expansion of the Biden Administration’s 50% tariff on Chinese semiconductors to include all chips, consumer, or industrial products that include a wafer manufactured in the People’s Republic of China, based on their international market rate. That is, if an Indonesian car manufacturer purchases a wafer manufactured in China with a market value of $3,000, and uses it to manufacture a $35,000 car, importing this vehicle to the United States would be subject to an additional tax of $1,500.
While fears abound of the inflationary effects of additional tariffs, they are necessary for the creation of an incentive structure that properly contains Chinese manufacturing. In the absence of proportional tariffs on chips and products assembled outside China, Chinese fabs will be able to circumvent U.S. trade restrictions by boosting wafer production that then takes advantage of Assembly, Testing, and Packaging (ATP) in third countries. Further, it is imperative for the United States to not only restrict Chinese chip growth but to encourage the development of domestic and foreign non-China chip manufacturers. Imposing tariffs on Chinese chips as an intermediate ingredient is necessary to create a proper competitive environment. Ultimately, the goal is to ensure a diversification of fab locations beyond China that will create lower prices for consumers overall.
How would tariffs on final goods containing Chinese chips be enforced? The policy issue of sanctioning and restricting an intermediate product is, unfortunately, not new. It is well known that Chinese precursor chemicals, often imported into Mexico, form much of the raw inputs for deadly fentanyl that is driving the United States opioid epidemic. Taking a cue from this example, we further suggest the creation of an internationally maintained database of products manufactured using Chinese semiconductors. As inspiration, the National Institutes of Health / NCATS maintains the Global Substance Registration System, a database that categorizes chemical substances, along with their commonly used names, regulatory classification, and relationships with other related chemicals. Such a database could be administered by the Commerce Department’s Bureau of Industry and Security, allowing the personnel who enforce the tariffs to also collect all relevant information in one place.
Companies importing products into the U.S. would be required to register the make and model of all Chinese chips used in each of their products so that the United States and participating countries could impose corresponding sanctions. Products imported to the U.S. would be subject to random checks involving disassembly in Commerce Department workshops, with failure to report a sanctioned semiconductor component making a company subject to additional tariffs and fines. Manual disassembly is painstaking and difficult, but regular, randomized inspections of imported products are the only way to truly verify their content.
The maintenance of such a database would bring follow-on national security benefits, in that the disclosure of any future vulnerability in a Chinese electronic component would allow quick diagnosis of what systems, including critical infrastructure, might be immediately vulnerable. We believe that creating an enforcement infrastructure that coordinates information between the U.S. and partner countries is a necessary first step to ensuring that tariffs are effective.
Zone Defense: International Cooperation in the Semiconductor Tariffs
At first glance, tariff action by the United States on Chinese-produced goods would appear to be a difficult coordination problem. By voluntarily declining an influx of cheaply-priced goods, American consumers exacerbate an existing trade glut in world semiconductor markets, allowing and incentivizing other nations to purchase these chips in greater volume and at a lower price.
However, rather than dissuading further sanctions in world markets, tariffs may actually spur further coordination in blocking Chinese imports. The Biden Administration’s imposition of tariffs on Chinese electric vehicles coincided with parallel sanctions imposed by the European Union, India, and Brazil. As Chinese overcapacity in EVs is rejected by U.S. markets, other countries face intensified concerns about the potential for below-price “dumping” of products that could harm domestic industry.
However, this ad-hoc international cooperation is still in a fragile and tentative stage and must be encouraged in order to create an “everywhere but China” semiconductor supply chain. Further, while countries impose tariffs to protect existing automotive and steel industries, global semiconductor manufacturing is currently concentrated in the Indo-Pacific. Thus, coordinating against China calls on countries to not just impose tariffs to protect existing industries, but to impose “nursery” tariffs that will protect nascent semiconductor development, even in places where an existing industry does not yet exist.
A commonsense first step to building an atmosphere of trust is to take actions protecting partner countries from retaliation in the form of Chinese trade restrictions. In response to EU tariffs on Chinese EVs, Beijing has already threatened retaliatory restrictions on chicken feet, pork, and brandy. For a bloc as large as the European Union, these restrictive sanctions can irritate an important political constituency. For a smaller or less economically powerful country, these measures might be decisive in sending the message that semiconductor tariffs are not worth the risk.
The United States should negotiate bilateral treaties with partner nations to erect tariffs against Chinese manufacturing, with the agreement that, in the case of Chinese retaliation against predetermined fundamental national industries, the United States government will buy up excess capacity at slightly discounted prices and release it to the American market. This preemptive protection of allied trade will blunt China’s ability to harm U.S. partners and allies. Raising tariffs on imported goods also imposes costs on the Chinese consumer, meaning that in the best case, the decreased effectiveness of these tools will deter the PRC from attempting such measures in the first place.
Recommendation 3. Mitigating the Threat of Existing Chips
No matter the success of the previous measures mentioned, it will be impossible to keep Chinese products entirely outside the U.S. market. Therefore, a strategy is required for managing the operational risks posed by Chinese chips that have and will exist inside the U.S. domestic sphere.
Precisely defining the scope of the threat is very important. A narrow definition of threats might allow threats to pass through, while an overly wide definition may expend time and resources over nothing. A recent British effort to exclude Chinese-made cap badges presents a cautionary tale. By choosing a British supplier over an existing Chinese one after the acquisition process was already underway, the UK incurred an additional delay in its military pipeline, not to mention the additional confusion caused by such an administrative pivot. Implanting GPS-tracking or listening devices within small pieces of metal by one company within the Chinese supply chain seems both impractical and far-fetched– though the PRC surely enjoys the chaos and expense such a panic can cause.
We consider it analogously unlikely that China is currently aiming to insert intentional defects into its semiconductor manufacturing. First, individual wafers are optimized for their extremely low cost of production, meaning that inserting a carefully designed (and hidden) flaw would introduce additional costs that could compromise the goal of low-cost manufacturing. Any kind of remotely activated “kill switch” would require some kind of wireless receiver–and a receiver of any reasonable strength could not be effectively hidden on a large scale. Second, such a vulnerability would have to be inserted only into wafers that are eventually purchased by the U.S. and its allies. If not, then any attempt to activate a remote exploit could risk compromising uninvolved countries or even the Chinese domestic market, either by accidentally triggering unintended chips or by providing a hardware vulnerability that could be re-used by Western cyber operations. Deliberately planting such vulnerabilities would thus require not just extreme technical precision, but a careful accounting of where vulnerable chips arrive in the supply chain.
Nonetheless, the existence of Chinese chips in the American market can accomplish much without explicitly-designed defects or “kill switches”. Here, a simple lack of transparency may be enough. China currently requires that all software vulnerabilities be reported to the Ministry of Industry and Information Technology, but does not have any corresponding public reporting requirement. This raises the fear that the Chinese government may be ‘stockpiling’ vulnerabilities in Chinese-produced products, which may be used in future cyber operations. Here, China does not need to explicitly build backdoors into its own hardware but may simply decline to publicly disclose vulnerabilities in software in order to attack the rest of the world.
Shining a Light on Untrusted Hardware
The likelihood of cooperation between Chinese industry and the CCP exposes a potentially important risk. Chinese software is often deployed atop or alongside Chinese semiconductors and is particularly dangerous in the form of hardware drivers, the “glue code” that ties together software with the low-level hardware components. These drivers by default operate with high privileges and are typically closed-source and thus difficult to examine. We believe that vulnerable drivers may be a key vector of Chinese espionage or cyber threats. In 2019, Microsoft disclosed the existence of a privilege escalation vulnerability found in a Huawei driver. Although Huawei cooperated with Microsoft, it is unclear under the current legal regime whether the discovery of similar vulnerabilities by Huawei would be reported and patched, or if it would be kept as an asset by the Chinese government. The promulgation of Chinese drivers packaged with cheap hardware thus means that the Chinese Communist Party will have access to a broad, and potentially pre-mapped, attack surface with which to exploit U.S. government services.
The first policy step here is obvious: banning the use of Chinese chips in U.S. federal government acquisitions. This has already been proposed as a Defense Department regulation set to take effect in 2027. If possible, this date should be moved up to 2026 or earlier. In order to enforce this ban, continuous research should be undertaken to map supply chains that produce U.S. government semiconductors. How to accelerate and enforce this ban is an ongoing policy question that is beyond the scope of this paper.
However, a deeper question is how to protect the myriad components of critical infrastructure, both formal and informal. The Cybersecurity and Infrastructure Security Agency (CISA) has defined 16 sectors of critical infrastructure whose failure could materially disrupt or endanger the lives of American citizens. The recent discovery of the Volt Typhoon threat group revealed the willingness of the Chinese government to penetrate U.S. critical infrastructure using vulnerable components.
While some of the 16 CISA sectors, such as Government Services and the Defense Industrial Base are within the purview of the federal government, many others, such as Healthcare, Food and Agriculture, and Information Technology, are run via complex partnerships between State, Local, Tribal, and Territorial (SLTT) governments and private industry. Although the best effort should be made to insulate these sectors from over-reliance on China, fully quarantining them from Chinese chips is simply unrealistic. Therefore we should explore proactive efforts at mitigation in the case of disruption.
A first step would be to establish a team at CISA to decompile or reverse-engineer the drivers for Chinese hardware that is known to operate within U.S. critical infrastructure. Like manual disassembly, this is an expensive and arduous process, but it has the advantage of reducing an unknown or otherwise intractable problem to an issue of engineering. In this case, special care should be taken to catalog and prioritize pieces of Chinese hardware that impact the most critical infrastructure systems, such as Programmable Logic Controllers (PLCs) in energy infrastructure and processors in hospital databases. This approach can be coordinated with the threat database described in the previous section to disassemble and profile the drivers of the highest-impact semiconductor products first. If any vulnerabilities are found, warnings can be issued to critical infrastructure providers, and patches issued to the relevant parties.
Brace for Impact: Building Infrastructure Resiliency
Even in the case that neither the reduction of Chinese hardware nor the proactive search for driver vulnerabilities is able to prevent a Chinese attack, the United States should be prepared to mitigate the harms of a cyber crisis.
A further step toward this goal would be the institution of resiliency protocols and drills for designated critical infrastructure providers. The 2017 WannaCry ransomware attack substantially incapacitated the UK National Health Service by locking providers out of Electronic Medical Record (EMR) systems. Mandating routine paper backups of digital medical records is one example of a resiliency strategy that could be deployed to ensure emergency functioning even in the case of a major service disruption.
A further step to protect pieces of critical infrastructure is to mandate regular cyber training for critical infrastructure providers. CISA could work in cooperation with State, Local, Tribal, and Territorial regulatory bodies to identify critical pieces of infrastructure. CISA could develop hypothetical scenarios involving outages of critical Information Technology services, and work with local infrastructure providers, such as hospitals, municipal water services, transit providers, and the like, to create plans for how to continue to operate in the event of a crisis. CISA could also prepare baseline strategies, such as having non-internet connected control systems or offline backups of critical information. Such strategies could be adapted by individual infrastructure providers to best protect their services in the event of an attack. These plans could then be carried out in mock ‘cyber drills’ to exercise preparedness in the event of an incident.
Ultimately, plans of this kind only prepare for service disruptions and do not address the longer-reaching impacts of breaches of confidentiality or the targeted manipulation of sensitive data. However, as we believe that the likelihood of targeted or sophisticated vulnerabilities in Chinese chips is relatively low, these kinds of brute force attacks are the most likely threat models. Preparing for the most basic and unsophisticated service disruptions is a good first step toward mitigating the harm of any potential cyber attack, including those not directly facilitated by Chinese hardware. This cyber-resiliency planning is therefore a strong general recommendation for protecting Americans from future threats.
Conclusion
We have presented the issue of international semiconductor competition along three major axes: increasing production outside of China, containing an oversupply of Chinese semiconductors, and mitigating the risks of remaining Chinese chips in the U.S. market. We have proposed three slates of policies corresponding to each challenge with some guidance on how to proceed, divided into three complementary categories:
Boosting non-China semiconductor production
- Institute a “Lend-Lease” policy for semiconductor IP, design software, and Semiconductor Manufacturing Equipment (SME) in allied and partner nations
- Create a strategic reserve of semiconductors held by the U.S. government, in order to insulate against chip shortages and shore up demand in fledgling markets
Containing Chinese exports
- Institute tariffs on all imported products making use of Chinese semiconductors and components, with a product database to catalog products containing Chinese chips and ensure enforcement
- Negotiate bilateral treaties to protect partner countries from Chinese trade retaliation by buying up excess trade capacity in the event of Chinese tariffs
Mitigating the threat of chips in the U.S. market
- Disassemble and reverse-engineer drivers in imported Chinese hardware to identify vulnerabilities
- Build resiliency strategies and institute training drills for critical infrastructure to function without computer infrastructure
We hope that this contribution will advance future discussions on the semiconductor trade and make a measurable impact on bolstering U.S. national security.
In recent years, China has heavily subsidized its legacy chip manufacturing capabilities. Although U.S. sanctions have restricted China’s access to and ability to develop advanced AI chips, they have done nothing to undermine China’s production of “legacy chips,” which are semiconductors built on process nodes 28nm or larger. It is important to clarify that the “22 nm” “20nm” “28nm” or “32nm” lithography process is simply a commercial name for a generation of a certain size and its technology that has no correlation to the actual manufacturing specification, such as the gate length or half pitch. Furthermore, it is important to note that different firms have different specifications when it comes to manufacturing. For instance, Intel’s 22nm lithography process uses a 193nm wavelength argon fluoride laser (ArF laser) with a 90nm Gate Pitch and a 34 nm Fin height. These specifications vary between fab plats such as Intel and TSMC. The prominence of these chips makes them a critical technological component in applications as diverse as medical devices, fighter jets, computers, and industrial equipment. Since 2014, state-run funds in China have invested more than $40 billion into legacy chip production to meet their goal of 70% chip sufficiency by 2030. Chinese legacy chip dominance—made possible only through the government’s extensive and unfair support—will undermine the position of Western firms and render them less competitive against distorted market dynamics.
Challenge and Opportunity
Growing Chinese capacity and “dumping” will deprive non-Chinese chipmakers of substantial revenue, making it more difficult for these firms to maintain a comparative advantage. China’s profligate industrial policy has damaged global trade equity and threatens to create an asymmetrical market. The ramifications of this economic problem will be most felt in America’s national security, as opposed to from the lens of consumers, who will benefit from the low costs of Chinese dumping programs until a hostile monopoly is established. Investors—anticipating an impending global supply glut—are already encouraging U.S. firms to reduce capital expenditures by canceling semiconductor fabs, undermining the nation’s supply chain and self-sufficiency. In some cases, firms have decided to cease manufacturing particular types of chips outright due to profitability concerns and pricing pressures. Granted, the design of chip markets is intentionally opaque, so pricing data is insufficient to fully define the extent of this phenomenon; however, instances such as Taiwan’s withdrawal from certain chip segments shortly after a price war between China and its competitors in late 2023 indicate the severity of this issue. If they continue, similar price disputes are capable of severely subverting the competitiveness of non-Chinese firms, especially considering how Chinese firms are not subject to the same fiscal constraints as their unsubsidized counterparts. In an industry with such high fixed costs, the Chinese state’s subsidization gives such firms a great advantage and imperils U.S. competitiveness and national security.
Were the U.S. to engage in armed conflict with China, reduced industrial capacity could quickly impede the military’s ability to manufacture weapons and other materials. Critical supply chain disruptions during the COVID-19 pandemic illustrate how the absence of a single chip can hold hostage entire manufacturing processes; if China gains absolute legacy chip manufacturing dominance, these concerns would be further amplified as Chinese firms become able to outright deny American access to critical chips, impose harsh costs through price hikes, or impose diplomatic compromises and quid-pro-quo.Furthermore, decreased Chinese reliance on Taiwanese semiconductors reduces their economic incentive to pursue a diplomatic solution in the Taiwan Strait—making armed conflict in the region more likely. This weakened posture endangers global norms and the balance of power in Asia—undermining American economic and military hegemony in the region.
China’s legacy chip manufacturing is fundamentally an economic problem with national security consequences. The state ought to interfere in the economy only when markets do not operate efficiently and in cases where the conduct of foreign adversaries creates market distortion. While the authors of this brief do not support carte blanche industrial policy to advance the position of American firms, it is the belief of these authors that the Chinese government’s efforts to promote legacy chip manufacturing warrant American interference to ameliorate harms that they have invented. U.S. regulators have forced American companies to grapple with the sourcing problems surrounding Chinese chips; however, the issue with chip control is largely epistemic. It is not clear which firms do and do not use Chinese chips, and even if U.S. regulators knew, there is little political appetite to ban them as corporations would then have to pass higher costs onto consumers and exacerbate headline inflation. Traditional policy tools for achieving economic objectives—such as sanctions—are therefore largely ineffectual in this circumstance. More innovative solutions are required.
If its government fully commits to the policy, there is little U.S. domestic or foreign policy can do to prevent China from developing chip independence. While American firms can be incentivized to outcompete their Chinese counterparts, America cannot usurp Chinese political directives to source chips locally. This is true because China lacks the political restraints of Western countries in financially incentivizing production, but also because in the past—under lighter sanctions regimes—China’s Semiconductor Manufacturing International Corporation (SMIC) acquired multiple Advanced Semiconductor Materials Lithography (ASML) DUV (Deep Ultraviolet Light) machines. Consequently, any policy that seeks to mitigate the perverse impact of Chinese dominance of the legacy chip market must a) boon the competitiveness of American and allied firms in “third markets” such as Indonesia, Vietnam, and Brazil and b) de-risk America’s supply chain from market distortions and the overreliance that Chinese policies have affected. China’s growing global share of legacy chip manufacturing threatens to recreate the global chip landscape in a way that will displace U.S. commercial and security interests. Consequently, the United States must undertake both defensive and offensive measures to ensure a coordinated response to Chinese disruption.
Plan of Action
Considering the above, we propose the United States enact a policy mutually predicated on innovative technological reform and targeted industrial policy to onshore manufacturing capabilities.
Recommendation 1. Weaponizing electronic design automation
Policymakers must understand that from a lithography perspective, the United States controls all essential technologies when it comes to the design and manufacturing of integrated circuits. This is a critically overlooked dimension in contemporary policy debates because electronic design automation (EDA) software closes the gap between high-level chip design in software and the lithography system itself. Good design simulates a proposed circuit before manufacturing, plans large integrated circuits (IC) by “bundling” small subcomponents together, and verifies the design is connected correctly and will deliver the required performance. Although often overlooked, the photolithography process, as well as the steps required before it, is a process as complex as coming up with the design of the chip itself.
No profit-maximizing manufacturer would print a chip “as designed” because it would suffer certain distortions and degradations throughout the printing process; therefore, EDA software is imperative to mitigate imperfections throughout the photolithography process. In much the same way that software within a home-use printer automatically screens for paper material (printer paper vs glossy photo paper) and automatically adjusts the mixture of solvent, resins, and additives to display properly, EDA software learns design kinks and responds dynamically. In the absence of such software, the yield of usable chips would be much lower, making these products less commercially viable. Contemporary public policy discourse focuses only on chips as a commodified product, without recognizing the software ecosystem that is imperative in their design and use.
Today, there exist only two major suppliers of EDA software for semiconductor manufacturing: Synopsys and Cadence Design Systems. This reality presents a great opportunity for the United States to assert dominance in the legacy chips space. In hosting all EDA in a U.S.-based cloud—for instance, a data center located in Las Vegas or another secure location—America can force China to purchase computing power needed for simulation and verification for each chip they design. This policy would mandate Chinese reliance on U.S. cloud services to run electromagnetic simulations and validate chip design. Under this proposal, China would only be able to use the latest EDA software if such software is hosted in the U.S., allowing American firms to a) cut off access at will, rendering their technology useless and b) gain insight into homegrown Chinese designs built on this platform. Since such software would be hosted on a U.S.-based cloud, Chinese users would not download the software which would greatly mitigate the risk of foreign hacking or intellectual property theft. While the United States cannot control chips outright considering Chinese production, it can control where they are integrated. A machine without instructions is inoperable, and the United States can make China’s semiconductors obsolete.
The emergence of machine learning has introduced substantial design innovation in older lithography technologies. For instance, Synopsis has used new technologies to discern the optimal route for wires that link chip circuits, which can factor in all the environmental variables to simulate the patterns a photo mask design would project throughout the lithography process. While the 22nm process is not cutting edge, it is legacy only in the sense of its architecture. Advancements in hardware design and software illustrate the dynamism of this facet in the semiconductor supply chain. In extraordinary circumferences, the United States could also curtail usage of such software in the event of a total trade war. Weaponizing this proprietary software could compel China to divulge all source code for auditing purposes since hardware cannot work without a software element.
The United States must also utilize its allied partnerships to restrict critical replacement components from enabling injurious competition from the Chinese. Software notwithstanding, China currently has the capability to produce 14nm nodes because SMIC acquired multiple ASML DUV machines under lighter Department of Commerce restrictions; however, SMIC heavily relies on chip-making equipment imported from the Netherlands and Japan. While the United States cannot alter the fact of possession, it has the capacity to take limited action against the realization of these tools’ potential by restricting China’s ability to import replacement parts to service these machines, such as the lenses they require to operate. Only the German firm Zeiss has the capability to produce such lenses that ArF lasers require to focus—illustrating the importance of adopting a regulatory outlook that encompasses all verticals within the supply chain. The utility of controlling critical components is further amplified by the fact that American and European firms have limited efficacy in enforcing copyright laws against Chinese entities. For instance, while different ICs are manufactured within the 22nm instruction set, not all run on a common instruction set such as ARM. However, even if such designs run on a copyrighted instruction set, the United States has no power to enforce domestic copyright law in a Chinese jurisdiction. China’s capability to reverse engineer and replicate Western-designed chips further underscores the importance of controlling 1) the EDA landscape and 2) ancillary components in the chip manufacturing process. This reality presents a tremendous yet overlooked opportunity for the United States to reassert control over China’s legacy chip market.
Recommendation 2. Targeted industrial policy
In the policy discourse surrounding semiconductor manufacturing, this paper contends that too much emphasis has been placed on the chips themselves. It is important to note that there are some areas in which the United States is not commercially competitive with China, such as in the NAND flash memory space. China’s Yangtze Memory Technologies has become a world leader in flash storage and can now manufacture a 232-layer 3D NAND on par with the most sophisticated American and Korean firms, such as Western Digital and Samsung, at a lower cost. However, these shortcomings do not preclude America from asserting dominance over the semiconductor market as a whole by leveraging its dynamic random-access memory (DRAM) dominance, bolstering nearshore NAND manufacturing, and developing critical mineral processing capabilities. Both DRAM and NAND are essential components for any computationally integrated technology.
While the U.S. cannot compete on rote manufacturing prowess because of high labor costs, it would be strategically beneficial to allow supply chain redundancies with regard to NAND and rare earth metal processing. China currently processes upwards of 90% of the world’s rare earth metals, which are critical to any type of semiconductor chips. While the U.S. possesses strategic reserves for commodities such as oil, it does not have any meaningful reserve when it comes to rare earth metals—making this a critical national security threat. Should China stop processing rare earth metals for the U.S., the price of any type of semiconductor—in any form factor—would increase dramatically. Furthermore, as a strategic matter, the United States would not have accomplished its national security objectives if it built manufacturing capabilities yet lacked critical inputs to supply this potential. Therefore, any legacy chips proposal must first establish sufficient rare earth metal processing capabilities or a strategic reserve of these critical resources.
Furthermore, given the advanced status of U.S. technological manufacturing prowess, it makes little sense to outright onshore legacy chip manufacturing capabilities—especially considering high U.S. costs and the substantial diversion of intellectual capital that such efforts would require. Each manufacturer must develop their own manufacturing process from scratch. A modern fab runs 24×7 and has a complicated workflow, with its own technique and software when it comes to lithography. For instance, since their technicians and scientists are highly skilled, TSMC no longer focuses on older generation lithography (i.e., 22nm) because it would be unprofitable for them to do so when they cannot fulfill their demand for 3nm or 4nm. The United States is better off developing its comparative advantage by specializing in cutting-edge chip manufacturing capabilities, as well as research and development initiatives; however, while American expertise remains expensive, America has wholly neglected the potential utility of its southern neighbors in shoring up rare earth metals processing. Developing Latin American metals processing—and legacy chip production—capabilities can mitigate national security threats. Hard drive manufacturers have employed a similar nearshoring approach with great success.
To address both rare earth metals and onshoring concerns, the United States should pursue an economic integration framework with nations in Latin America’s Southern Cone, targeting a partialized (or multi-sectoral) free trade agreement with the Southern Common Market (MERCOSUR) bloc. The United States should pursue this policy along two industry fronts, 1) waiving the Common External Tariff for United States’ petroleum and other fuel exports, which currently represent the largest import group for Latin American members of the bloc, and 2) simultaneously eliminating all trade barriers on American importation of critical minerals––namely arsenic, germanium, and gallium––which are necessary for legacy chip manufacturing. Enacting such an agreement and committing substantial capital to the project over a long-term time horizon would radically increase semiconductor manufacturing capabilities across all verticals of the supply chain. Two mutually inclusive premises underpin this policy’s efficacy:
Firstly, the production of economic interdependence with a bloc of Latin American states (as opposed to a single nation) serves to diversify risk in the United States; each nation provides different sets and volumes of critical minerals and has competing foreign policy agendas. This reduces the capacity of states to exert meaningful and organized diplomatic pressure on the United States, as supply lines can be swiftly re-adjusted within the bloc. Moreover, MERCOSUR countries are major energy importers, specifically with regard to Bolivian natural gas and American petroleum. Under an energy-friendly U.S. administration, the effects of this policy would be especially pronounced: low petroleum costs enable the U.S. to subtly reassert its geopolitical sway within its regional sphere of influence, notably in light of newly politically friendly Argentinian and Paraguayan governments. China has been struggling to ratify its own trade accords with the bloc given industry vulnerability, this initiative would further undermine its geopolitical influence in the region. Refocusing critical mineral production within this regional geography would decrease American reliance on Chinese production.
Secondly, nearshoring the semiconductor supply chain would reduce transport costs, decrease American vulnerability to intercontinental disruptions, and mitigate geopolitical reliance on China. Reduced extraction costs in Latin America, minimized transportation expenses, and reduced labor costs in especially Midwestern and Southern U.S. states enable America to maintain export competitiveness as a supplier to ASEAN’s booming technology industry in adjacent sectors, which indicates that China will not automatically fill market distortions. Furthermore, establishing investment arbitration procedures compliant with the General Agreement on Tariffs and Trade’s Dispute Settlement Understanding should accompany the establishment of transcontinental commerce initiatives, and therefore designate the the World Trade Organization as the exclusive forum for dispute settlement.
This policy is necessary to avoid the involvement of corrupt states’ backpaddling on established systems, which has historically impeded corporate involvement in the Southern Cone. This international legal security mechanism serves to assure entrepreneurial inputs that will render cooperation with American enterprises mutually attractive. However, partial free trade accords for primary sector materials are not sufficient to revitalize American industry and shift supply lines. To address the demand side, the exertion of downward pressure on pricing, alongside the reduction of geopolitical risk, should be accompanied by the institution of a state-subsidized low-interest loan, with available rate reset for approved legacy chip manufacturers, and a special-tier visa for hired personnel working in legacy chip manufacturing. Considering the sensitive national security interests at stake, the U.S. Federal Contractor Registration ought to employ the same awarding mechanisms and security filtering criteria used for federal arms contracts in its company auditing mechanisms. Under this scheme, vetted and historically capable legacy chip manufacturing firms will be exclusively able to take advantage of significant subventions and exceptional ‘wartime’ loans. Two reasons underpin the need for this martial, yet market-oriented industrial policy.
Firstly, legacy chip production requires highly specialized labor and immensely expensive fixed costs given the nature of accompanying machinery. Without targeted low-interest loans, the significant capital investment required for upgrading and expanding chip manufacturing facilities would be prohibitively high–potentially eroding the competitiveness of American and allied industries in markets that are heavily saturated with Chinese subsidies. Such mechanisms for increased and cheap liquidity also render it easier to import highly specialized talent from China, Taiwan, Germany, the Netherlands, etc., by offering more competitive compensation packages and playing onto the attractiveness of the United States lifestyle. This approach would mimic the Second World War’s “Operation Paperclip,” executed on a piecemeal basis at the purview of approved legacy chip suppliers.
Secondly, the investment fluidity that accompanies significant amounts of accessible capital serves to reduce stasis in the research and development of sixth-generation automotive, multi-use legacy chips (in both autonomous and semi-autonomous systems). Much of this improvement a priori occurs through trial-and-error processes within state-of-the-art facilities under the long-term commitment of manufacturing, research, and operations teams.
Acknowledging the strategic importance of centralizing, de-risking, and reducing reliance on foreign suppliers will safeguard the economic stability, national defense capabilities, and the innovative flair of the United States––restoring the national will and capacity to produce on its own shores. The national security ramifications of Chinese legacy chip manufacturing are predominantly downstream of their economic consequences, particularly vis-à-vis the integrity of American defense manufacturing supply chains. In implementing the aforementioned solutions and moving chip manufacturing to closer and friendlier locales, American firms can be well positioned to compete globally against Chinese counterparts and supply the U.S. military with ample chips in the event of armed conflict.
In 2023, the Wall Street Journal exposed the fragility of American supply chain resilience when they profiled how one manufacturing accident took offline 100% of the United States’ production capability for black powder—a critical component of mortar shells, artillery rounds, and Tomahawk missiles. This incident illustrates how critical a consolidated supply chain can be for national security and the importance of mitigating overreliance on China for critical components. As firms desire lower prices for their chips, ensuring adequate capacity is a significant component of a successful strategy to address China’s growing global share of legacy chip manufacturing. However, there are additional national security concerns for legacy chip manufacturing that supersede their economic significance–mitigating supply chain vulnerabilities is among the most consequential of these considerations.
Lastly, when there are substantial national security objectives at stake, the state is justified in acting independently of economic considerations; markets are sustained only by the imposition of binding and common rules. Some have argued that the possibility of cyber sabotage and espionage through military applications of Chinese chip technology warrants accelerating the timeline of procurement restrictions. The National Defense Authorization Act for Fiscal Year 2023’s Section 5949 prohibits the procurement of China-sourced chips from 2027 onwards. Furthermore, the Federal Communications Commission has the power to restrict China-linked semiconductors in U.S. critical infrastructure under the Secure Networks Act and the U.S. Department of Commerce reserves the right to restrict China-sourced semiconductors if they pose a threat to critical communications and information technology infrastructure.
However, Matt Blaze’s 1994 article “Protocol Failure in the Escrowed Encryption Standard” exposed the shortcomings of supposed hardware backdoors, such as the NSA’s “clipper chip” that they designed in the 1990s to surveil users. In the absence of functional software, a Chinese-designed hardware backdoor into sensitive applications could not function. This scenario would be much like a printer trying to operate without an ink cartridge. Therefore, instead of outright banning inexpensive Chinese chips and putting American firms at a competitive disadvantage, the federal government should require Chinese firms to release source code to firmware and supporting software for the chips they sell to Western companies. This would allow these technologies to be independently built and verified without undermining the competitive position of American industry. The U.S. imposed sanctions against Huawei in 2019 on suspicion of the potential espionage risks that reliance on Chinese hardware poses. While tighter regulation of Chinese semiconductors in sensitive areas seems to be a natural and pragmatic extension of this logic, it is unnecessary and undermines American dynamism.
Conclusion
Considering China’s growing global share of legacy chip manufacturing as a predominantly economic problem with substantial national security consequences, the American foreign policy establishment ought to pursue 1) a new technological outlook that exploits all facets of the integrated chip supply chain—including EDA software and allied replacement component suppliers—and 2) a partial free-trade agreement with MERCOSUR to further industrial policy objectives.
To curtail Chinese legacy chip dominance, the United States should weaponize its monopoly on electronic design automation software. By effectively forcing Chinese firms to purchase computing services from a U.S.-based cloud, American EDA software firms can audit and monitor Chinese innovations while reserving the ability to deny them service during armed conflict. Restricting allied firms’ ability to supply Chinese manufacturers with ancillary components can likewise slow the pace of Chinese legacy chip ascendence.
Furthermore, although China no longer relies on the United States or allied countries for NAND manufacturing, the United States and its allies maintain DRAM superiority. The United States must leverage capabilities to maintain Chinese reliance on its DRAM prowess and sustain its competitive edge while considering restricting the export of this technology for Chinese defense applications under extraordinary circumstances. Simultaneously, efforts to nearshore NAND technologies in South America can delay the pace of Chinese legacy chip ascendence, especially if implemented alongside a strategic decision to reduce reliance on Chinese rare earth metals processing.
In nearshoring critical mineral inputs to the end of preserving national security and reducing costs, the United States should adopt a market-oriented industrial policy of rate-reset, and state-subsidized low-interest loans for vetted legacy chip manufacturing firms. Synergy between greater competitiveness, capital solvency, and de-risked supply chains would enable U.S. firms to compete against Chinese counterparts in critical “third markets,” and reduce supply chain vulnerabilities that undermine national security. As subsidy-induced Chinese market distortions weigh less on the commercial landscape, the integrity of American defense capabilities will simultaneously improve, especially if bureaucratic agencies move to further insulate critical U.S. infrastructure against potential cyber espionage.
Familiar semiconductor policy approaches – export controls and subsidies – are inadequate alone to prevent reliance on Chinese-made legacy chips. Washington and its allies will instead have to turn to the old-fashioned, disruptive tools of trade defense in the face of a challenge of this scale.
In an industry with such high fixed costs, the Chinese state’s subsidization gives such firms a great advantage and imperils U.S. competitiveness and national security. To curtail Chinese legacy chip dominance, the United States should weaponize its monopoly on electronic design automation software.
The technical advances fueled by leading-edge nodes are vital to our long-term competitiveness, but they too rely on legacy devices.
To tackle AI risks in grant spending, grant-making agencies should adopt trustworthy AI practices in their grant competitions and start enforcing them against reckless grantees.