The debate over U.S. tariffs has been mired in politics and is now causing market disruptions. But the real stakes are even higher: The future of American manufacturing. President Trump’s tariffs regime is not just about rebalancing trade or punishing some countries; it is very much about “reprioritizing U.S. manufacturing” and and bringing about “the day American industry was reborn,” as Trump put it.
Yet tariffs alone can’t revive an industrial base hollowed out by decades of offshoring and underinvestment. If America wants to lead again, it must go beyond trade wars and build a real manufacturing ecosystem — starting with semiconductors, the linchpin of modern industry.
Take automobile manufacturing, for example. The average car today relies on thousands of semiconductors — a number that will only grow as vehicles become electric and autonomous. If the U.S. fails to secure its own supply of these tiny but indispensable components, no tariff will shield American factories from the next shortage. Or worse, if the U.S. fails to lead in development and production of these chips, its reliance on foreign supplies could morph from a market problem to a critical national security problem.
Semiconductors aren’t just about cars, though — they’re the crude oil of the 21st century. Without them, the U.S. economy grinds to a halt, and our military loses its technological edge. These tiny microchips instruct industrial robots, power our cellphones, help fly airplanes and turn on the coffee machine. They enable increasingly sophisticated AI models and guide defense systems.
While the importance of semiconductors has long been recognized by Congress and successive presidential administrations, policies have failed to keep the U.S. at the forefront of semiconductor manufacturing. Whereas the U.S. once commanded 40 percent global semiconductor manufacturing in 1990, it accounts for just 10 percent today, and none of these are in cutting-edge chips, such as those that facilitate AI. The U.S. firm Intel, once the dominant semiconductor leader, has been in decline.
For the U.S. to regain its edge in semiconductor manufacturing and development, a robust industrial program must materialize: an initiative that mobilizes an industrial bank with the sole focus of investing in U.S. manufacturing and building up research and development, by raising a critical mass of experts to solve the most pressing technological and manufacturing problems.
The 2022 CHIPS Act was a step forward, but it treated symptoms, not the cause. While it funded Research and Development and subsidized factories like TSMC’s Arizona plants, it failed to address the root problem: America is lacking the infrastructure, talent and financial firepower to compete with Asia.
Economic history shows tariffs can have a place in industrial policy, but they won’t rebuild it. Tariffs must be accompanied by a package of policy tools for industrialization: easy financing, targeted investment in key sectors, manufacturing incentives, favorable tax policies for strategic sectors, and technological and industrial knowledge diffusion.
In the 1960s-1980s, Japan, South Korea and Taiwan used tariffs effectively to protect domestic industries. But at the same time, policy makers enhanced the industry with financing, training, technology and production incentives. These countries went on to experience some of the fastest economic growth in world history and developed large and prosperous firms such as Toyota, Samsung and TSMC.
For such reasons, we advise against repeal of CHIPS and advocate revamping it with a more comprehensive program. The U.S. should emphasize semiconductor manufacturing in the same way the Trump administration has emphasized steel and aluminum manufacturing, and more recently automobile manufacturing.
To do so will require not only high levels of investment in capital equipment, but also in research and development and technical expertise. A single EUV lithography exposure tool, for example, may cost $350 million. A typical factory will need 10 such tools, along with the highly trained personnel to operate and maintain them.
To compete, the U.S. needs a Marshall Plan for semiconductors — a wartime-level mobilization of capital, brains and industrial policy.
Foremost is funding. The price tag of semiconductor manufacturing is in the tens of billions to build a single factory, or fab. For example, TSMC recently committed $100 billion for semiconductor manufacturing in the U.S., in addition to an already committed $65 billion. American firms need ready access to that kind of capital to invest in production.
Second is research. The major technical concern in the semiconductor industry today is the leakage of electrons as transistors are shrunk to cram more functions on the same chip size. The shape of transistors are ever more complex and harder to manufacture. While TSMC and the most advanced manufacturers say they will roll out new processes later this year, it is an even more complicated production process requiring advanced equipment. What will be needed in the future is a breakthrough in physics. This can only take place with better research centers and funding.
Third is education. A critical mass of experts in semiconductor manufacturing needs to emerge. Such a body will require at least 200 advanced degree experts in math and sciences, with comparable training and skills that can be applied to manufacturing.
The choice is stark: Will America settle for begging TSMC to build factories on our soil, or will we reclaim the power to make the chips that shape the future? Tariffs can protect, but only a robust industrial program can rebuild.
The time to act is now — before the next breakthrough happens in Shanghai rather than Texas.
Macabe Keliher is an associate professor in the Clements Department of History at SMU Dallas. He is writing a book on industrialization and manufacturing in East Asia. Dev Gupta, Ph.D., is the founder and CTO of APSTL llc of Scottsdale, Ariz. His past inventions in semiconductor manufacturing at Motorola & Intel in AZ are now used widely for AI hardware. He chairs a section of the IEEE International Roadmap for Semiconductor Devices.
