Chipping Away at Competitiveness: Why Tariffs Won’t Save U.S. Semiconductor Manufacturing
Bipartisan agreement remains rare in DC these days. However, one area where both the outgoing Biden administration, the incoming Trump administration, and congressional policymakers on both sides of the aisle remain in heated agreement is the pressing need to revitalize U.S. semiconductor manufacturing. The 70-percent decline in U.S. share of global semiconductor manufacturing from 1990 to 2020 animated passage of the $52 billion CHIPS Act in 2022, showing how effective U.S. competitiveness policy can support the sector. And that should remain the guiding path forward for the Trump administration, for while substantially heightened tariffs can have an effect at the margin of inducing locational shift in manufacturing, tariffs as the principle focus of a Trump 2.0 semiconductor policy would be confounded by the vast supply chains underpinning the industry and the reality that tariffs would likely contribute to price increases that would make U.S. semiconductor manufacturing, and exports thereof, uncompetitive, depriving U.S. firms of the scale they need to effectively compete globally.
The president-elect contends that aggressive tariffs on manufactured goods coming into the United States will stimulate greater levels of domestic manufacturing by making U.S. manufacturing relatively more competitive and thus either inducing the shift of existing manufacturing plants to the United States, or enticing manufacturers to build or expand facilities in the United States as opposed to elsewhere. While details are unclear, Trump has proposed a 20 percent tariff on every item entering the United States, 60 percent tariffs on all Chinese goods, and also 100 percent tariffs on products coming from the BRICS nations (Brazil, China, India, Russia, and South Africa), should these nations try to develop financial alternatives to the dollar. Trump is correct that tariffs can be impactful, as long as other nations do not respond tit-for-tat. Assuming that they do not, tariffs on manufactured goods (as a means to induce production shift) tend to work best in low- and moderate-tech, price-driven industries—like automobile tires or microwave ovens—and are much less likely to be as effective in highly technically complex industries, especially industries that rely on global value chains that depend on best-of-breed inputs from vast networks of suppliers.
Manufacturing semiconductors is perhaps the most complex, expensive engineering task humanity undertakes. The newest, most-sophisticated semiconductor fabs can cost over $30 billion to construct. Because semiconductor manufacturers must make the right choice in where to site a $30 billion-plus investment, they consider as many as 500 discrete factors—ranging from countries’ and states’ talent, tax, trade, and technology policies to their regulatory, environmental, and labor-market policies. To be sure, a country’s tariff regime represents one of these factors. But more often than not, companies are looking for zero tariffs on all the inputs they need to make chips. Moreover, tariffs are not nearly such a significant factor that they could convince a company to relocate a multibillion-dollar fab to the United States (or to choose to cite a new facility in America as opposed to somewhere else).
In fact, on the contrary, high tariffs are much more likely to engender the precise opposite effect. Consider semiconductor manufacturing equipment such as the extreme ultraviolet lithography (EUV) machines that print semiconductors: The laser alone in an EUV machine has 457,329 component parts, and the laser itself is just one of more than 100,000 parts in an EUV machine. And the EUV machine itself is just one component among thousands in a multi-billion-dollar semiconductor fab. The key point is that for nations to be competitive at semiconductor manufacturing, they need to ensure that companies have cost-efficient access to the best-of-breed, most-sophisticated components and inputs—such as chemicals, substrates, photomasks, and other materials—that are sourced from more than 50 countries worldwide.
Indeed, the globalization of the semiconductor industry has enabled specialized suppliers of key inputs to emerge in certain corners around the world. For instance, Japanese suppliers provide 90 percent of the world’s photoresists and over half the photomasks. These are the materials that contain and imprint the circuit pattern on the wafers. The specialization that pervades the global semiconductor industry has played a key role in advancing the industry’s rapid innovation while decreasing the unit costs of computer processing. In other words, it’s what’s enabled Moore’s Law—the principle that the speed and capability of computers can be expected to double roughly every two years. America simply does not have the technical or manufacturing capacity by itself to produce the myriad thousands of inputs and components that underpin advanced semiconductor manufacturing.
Thus, the tariffs the incoming Trump administration has proposed would only raise the costs of thousands of inputs into U.S. semiconductor production, and thus make it less globally competitive. This would also have a deleterious impact on U.S. semiconductor exports. Semiconductors were America’s sixth-largest export industry in 2023, tallying $52.7 billion in sales. And U.S. semiconductor exports matter greatly to the health of the U.S. industry, especially considering that over 80 percent of global semiconductor consumption occurs outside the United States. Like in other high-tech industries, from aerospace to biopharmaceuticals, U.S. semiconductor companies need access to markets at global scale to recoup their expensive R&D investments and earn returns that finance future generations of innovation.
As noted, by raising the relative cost of other countries’ manufactured goods to U.S. consumers, the theory is that tariffs will make U.S. manufacturers relatively more competitive—at least in the U.S. markets. And again, while higher tariffs might make U.S. manufacturers more competitive in lower-tech, price-driven industries with far narrower and fully domestically substitutable supply chains for key inputs, it’s not going to work that way in the semiconductor industry.
Indeed, historical evidence has shown that high tariffs provide a one-way ticket for excising nations from participating in value chains for the production of high-tech goods such as information and communications technology (ICT) products. In 1996, 82 countries banded together to sign the Information Technology Agreement (ITA), a plurilateral World Trade Organization agreement that eliminates tariffs among participating countries in hundreds of ICT products. In 2015, 53 countries joined together in completing an ITA expansion (ITA-2) that eliminated tariffs on an additional 201 ICT parts, components, and final products.
Countries imposing high tariffs on ICT components and products only make themselves unattractive to multinational enterprises wishing to seamlessly integrate their global production networks. This explains why the Organization for Economic Cooperation and Development has found that countries not participating in the ITA saw their participation in global ICT value chains decline by more than 60 percent from 1995 (two years before the ITA went into effect) to 2009. In contrast, countries participating in the ITA saw their participation rise after 1995. In fact, from 2005 to 2015, ITA member nations enjoyed nearly one-third greater participation in ICT global value chains (GVCs) than did non-ITA member nations. The message is clear: Countries that don’t provide a zero-in/zero-out tariff environment for the production of high-value-added ICT products such as semiconductors simply get circumvented and excised from ICT GVC participation. It helps explain why South America has virtually no electronics or ICT goods-producing industries left. And it shows that high tariffs are unlikely to stimulate a significant net increase in U.S. semiconductor manufacturing.
Two additional points should be noted. The first is that U.S. tariffs on China are already quite high. In fact, in May 2024, the Biden administration extended the Trump administration’s initial set of tariffs, increasing the rate on semiconductors from 25 percent to 50 percent by 2025. Moreover, the level of U.S. imports of semiconductors from China has already decreased by half from $24 billion in 2018 to $12 billion in 2023, so China-focused semiconductor tariffs aren’t going to have a tremendous impact in attracting more U.S. semiconductor manufacturing.
Lastly, tariffs represent a far-inferior mechanism to encourage U.S. semiconductor manufacturing compared to the instruments already introduced through the CHIPS Act. The $39 billion in grants the CHIPS Act provides serves to directly offset the U.S. semiconductor-manufacturing industry’s cost-competitiveness gap, where it costs some 30 percent more to build and operate a U.S. semiconductor fab over 10 years in America than it does in Asian competitors’ markets. The cash directly counteracts this cost differential and is immediately additive to making U.S. semiconductor manufacturing more attractive. Similarly, the 25 percent investment tax credit (ITC) implemented through companion legislation to the CHIPS Act further addresses the cost differential and becomes immediately available once manufacturers put a shovel in the ground toward new facilities.
In contrast, tariffs represent a diffuse, nonspecific, nontargeted instrument that might vaguely make U.S. manufacturing seem more cost-competitive, but in reality will usually have the opposite effect, especially in extraordinarily complex manufacturing industries like semiconductors and in particular on intermediate as opposed to final goods. In these cases, tariffs might lower the cost on inputs, but raise the price of the final good, limiting U.S. exports. For example, modern vehicles use 1,000 to 3,000 semiconductors, meaning that raising the cost of semiconductors would increase the cost of the final vehicle. That is only one of the hundreds of industries where this dynamic would apply.
Ultimately, tariffs aren’t going to engender the U.S. semiconductor manufacturing renaissance the incoming Trump administration rightfully seeks. Far better to focus on increasing the genuine competitiveness of American industries, including semiconductors. Indeed, the CHIPS Act has already contributed to over $450 billion in announced semiconductor and electronics sector investments across 20 U.S. states in projects that will employ some 125,000 workers in the years ahead. Trump should continue to focus on policies that will genuinely increase U.S. semiconductor manufacturing competitiveness, which could include focusing on activities such as enhancing workforce skills—which matters especially when the U.S. semiconductor industry faces an expected shortfall of 67,000 trained workers by the end of this decade—and turbocharging public-private partnerships, such as the recently announced semiconductor digital twin Manufacturing USA Institute. The Trump administration could also extend the 25 percent investment tax credit for semiconductor investments through 2030. (It’s currently set to expire the first day of 2027.) Indeed, a much better idea would be to get that list of 500 factors and make sure the United States offers the most attractive and supportive environment for semiconductor manufacturing in the world. Indeed, that’s exactly what many U.S. states—and America’s competitors around the globe—are doing.
