When the world's at stake,
go beyond the headlines.

The Commerce Department’s imminent Section 232 investigation — launched in April and expected to conclude soon — may fundamentally shift how the United States acquires semiconductors. The chips America imports range from commodity devices embedded in household appliances to the expensive, high-performance AI processors that power the AI boom, designed in America by Nvidia, but manufactured in Asia.

Taiwan and Korea sit at the center of this challenge. Together, they produce the majority of semiconductors used by the United States across nearly every category. This concentration represents a major national security risk, given China’s preparations and repeated threats to use force against Taiwan and antagonism toward South Korea.

Commerce Secretary Howard Lutnick phrased the situation thus in a recent interview: “If you can’t make your own chips, how can you defend yourself?”

Tariffs are one tool to address this exposure, designed to boost demand for chips manufactured in the United States. But they are not enough. Lutnick offers a second measure, called “chip for chip.” It would tie tariff waivers directly to verifiable domestic production milestones, incentivizing U.S. firms to act more in the national interest.

However, Lutnick’s ‘chip-for-chip’ framework could succeed only if Washington pairs it with enforceable production benchmarks and demand-side incentives. Otherwise, it risks becoming another half-measure in America’s decades-long struggle to rebuild semiconductor sovereignty.

From Offshore Expansion to “Dragon” Dependence

America is in this bind partly because of deliberate state engineering, dating back to the 1960s, that created the export-led “East Asian Miracle.” Favorable government subsidies and credit, structural reforms such as free trade zones, and pro-growth regulations fostered the most successful export-oriented firms in the “dragons” (aka: “tigers”): Taiwan, South Korea, Singapore, and Hong Kong, and a prominent “little dragon,” Malaysia.

As the “East Asian Miracle” developed, American chipmakers saw opportunities to lower costs and expand markets. Intel, founded in 1968, kept its leading-edge semiconductor fabrication technology at home but opened an assembly and test plant in Penang, Malaysia, in 1972. That “back-end” factory, which used less advanced technology to package finished chips, was followed by others in the Philippines, Costa Rica, China, and Vietnam. Meanwhile, Intel experimented in taking the higher tech “front-end” fabrication lines overseas: Fab 8 in Jerusalem (1985) and Fab 14 in Ireland (1990). These moves sought global efficiency while retaining the most advanced technology and processes at American fabs located in Arizona, Ohio, Oregon, and New Mexico.

As the “Miracle” evolved into the 21st century, it was exemplified in the electronics sector by Taiwan’s TSMC and Korea’s Samsung. Taiwan’s policies tightly regulate TSMC’s global moves while rewarding domestic investment, and South Korea used its own policies to foster and sustain Samsung’s scale as one of the country’s chaebol, the family-controlled conglomerates with close ties to the government that were central to building the nation’s modern economy.

Enter the Bigger Dragon

But China, as ever, was different. While foreign companies never admitted to trading technology for market access, Beijing’s implicit demands loomed large. Foreign-established fabs and accompanying supply chains sprouted up like bamboo shoots after a spring rain: Motorola MOS 17 (Tianjin, 2000, later sold to China’s SMIC), Texas Instruments (Chengdu, 2010), Intel Fab 68 and 68A (Dalian, 2010 and 2018), UMC (Xiamen, 2016), TSMC Fab 16 (Nanjing, 2018), and Samsung’s Xi’an NAND Fab (2020).

The Few, the Fabless, and the Fab-Lite

Intel is now a rare bird — the only major American “integrated device manufacturer” left standing, meaning that they run a fully vertical operation, from chip design, to fabrication, to assembly and test, to marketing and sales. Soaring front-end manufacturing costs prompted other companies to exit the  manufacturing space. For example, AMD spun off its factories in 2008 and 2009, creating Global Foundries. Six years later, IBM paid Global Foundries $1.5 billion to take over its fabs. AMD and IBM can now be characterized as “fabless.” Texas Instruments tried going “fab-lite” but decided to retain and expand its fabs. It is the other rare bird alongside Intel.

Meanwhile, companies that had always been fabless thrived. Nvidia, Qualcomm, and Broadcom focused on design, sales, and marketing, while relying on TSMC and Samsung to manufacture their products. The result: The world’s most advanced semiconductor capacity, along with much of the mature-node production, became concentrated in Taiwan and South Korea.

Today, TSMC alone produces roughly 90 percent of the world’s most advanced chips. Taiwan does not wish to give up that bounty or the silicon shield protection it perceives to come with it. As a result, the island presents a stark single point of failure. A Taiwan Strait crisis, natural disaster, or geopolitical rupture could shut down supply virtually overnight, crippling American and allied manufacturing in everything from automobiles to defense systems.

Washington’s Response

Both the Trump and Biden administrations recognized the danger. They reframed semiconductor manufacturing as a matter of national security, dangling incentives for TSMC and Samsung to establish advanced fabs on U.S. soil.

But progress has been uneven. Though TSMC is building facilities in Arizona, and Samsung has broken ground in Texas, both companies and their governments remain reluctant to export bleeding-edge processes to America. TSMC’s 3-4 nanometer lines are scheduled for Arizona, but the chip giant is prohibited by Taiwan law from producing that node abroad. Samsung has also been conservative, keeping its 2-nanometer production in South Korea.

The Demand Problem

Building a fab is only half the battle — keeping it full of orders is the harder part. The challenge is particularly acute when U.S. fabs will initially produce at 3-4 nm, one or two generations behind Taiwan’s best. Tech giants like Apple, Dell, and HP will likely have no incentive to pay more for chips produced domestically if cheaper or more advanced versions are available abroad.

Tariffs are one lever to address this imbalance. However, as currently structured, they are blunt instruments, punishing some firms while sparing others and creating loopholes that clever corporate lawyers can exploit. Investors may see risk without certainty. Companies could see costs without clear incentives.

Administering “Chip for Chip”

The premise of “chip for chip” is simple. Tie tariff relief directly to verifiable domestic production milestones. For every unit of certified U.S. fab capacity under construction or ramp-up, a company would be allowed to import an equivalent unit tariff-free. Once that U.S. fab reached volume production, the waiver would expire.

This promises to create three immediate effects. First, it raises the likelihood that buyers will be there once U.S. capacity comes online, reassuring investors and mitigating risk for new construction. Second, it would compel global firms like Apple and Dell to begin shifting their sourcing strategies, knowing their tariff relief depends on supporting U.S. fabs. Third, it would close loopholes, replacing open-ended waivers with measurable milestones backed by retroactive duties and “claw-back” penalties if companies fail to meet agreements. Chipmakers would need to match every imported chip with one made in the United States. If they fail to do so over time, penalties in the form of import duties — potentially reaching 100 percent — would be imposed.

There Will Be Pushback

A chip-for-chip framework will not be universally welcomed, and Lutnick’s plan may require tweaking to make it acceptable to major stakeholders. Multinationals that have long optimized their supply chains for cost and convenience will resist reallocating production at a higher expense. End-users operating on thin margins, like auto and electronics firms, may pass costs along to disenchanted consumers. If tariffs raise the cost of importing advanced chips, America’s allies may complain to the World Trade Organization that the United States is distorting markets. And Taiwan has not agreed to Lutnick’s proposal that the two sides aspire to a “50-50” solution, with half of advanced semiconductors made in the United States, and half in Taiwan.

Moreover, the complex semiconductor supply chain that services a fab’s equipment, supplies chemicals, and handles sensitive cargo would have to react to the chip-for-ship scenario without significant disruption through the construction, ramp-up, and full production stages. On the positive side, many key semiconductor equipment, chemical, and clean room suppliers are U.S.-based, and the “pull” of supplier response to manufacturer demand has always been strong. On the other hand, it is not frictionless or instantaneous, and is no longer U.S.-centric. Technical and other barriers are inevitable in the wholesale shift envisioned under chip-for-chip.

Meanwhile, Lutnick is sticking to his 50-50 goal, and Washington has legal cover: national security exceptions to trade rules. The precedent exists in other strategic sectors, and semiconductors are arguably today’s most strategic commodity. More importantly, the risks of inaction outweigh the costs of friction. If supply from Taiwan is disrupted, no amount of World Trade Organization litigation will help Detroit restart auto production or the Pentagon procure advanced weapons. And Beijing knows it.

The Strategic Stakes

China is not standing still. Beijing is pouring tens of billions into the electronics industry in general and new fabs in particular, subsidizing firms like SMIC, and tightening controls over critical materials such as gallium and germanium. Its long-term goal is technological self-sufficiency, freeing itself from U.S. export controls while increasing global leverage.

The U.S. semiconductor industry stands at an inflection point. Though propelled forward by developments in AI, decades of offshoring and complacency have left the world’s most advanced production clustered in Taiwan and South Korea — right next to the country that most threatens American security.

The CHIPS Act was a start. Tariffs are a blunt but necessary step. But the missing pieces are how the economic pain will be borne across the supply chain when major changes are in progress, and the details about enforcement: a framework that ties market access to domestic investment, ensuring that new capacity is built and sustained.

If successful, Lutnick’s chip-for-chip proposal compels companies to move closer to national interest. By aligning economic incentives with national security, Washington might finally break the cycle of dependence, just as Huawei advances by leaps and bounds and the PRC builds fab after fab. Tariffs alone will not secure America’s semiconductor future. Chip-for-chip might.

Matt Brazil was a corporate investigator and fab security manager in China for Intel. He is the co-author of Chinese Communist Espionage, An Intelligence Primer. A fellow at the Jamestown Foundation and an analyst with BluePath Labs, he previously served in U.S. Army Intelligence and as a commercial officer at the American Embassy in Beijing. His views do not represent those of his current or previous employers.

Image: www.dorox.com via Flickr