The Transnational Origins of Taiwan’s Semiconductor Industry

Nowhere is the logic of U.S. President Donald Trump’s trade policy more obviously flawed than his attempt to reshore semiconductor manufacturing to the United States.

Following Trump’s March claim that Taiwan “stole” its chip industry from the U.S. and his suggestion that Taipei should pay for American protection, Taiwan Semiconductor Manufacturing Company announced a further $100 billion investment in its Arizona plant.

Flanked by TSMC Chairman C.C. Wei at a White House meeting, Trump touted the investment – which will reportedly add new wafer fabs, advanced packaging facilities, and an R&D center to existing infrastructure – as ensuring the “most advanced AI chips” would now be manufactured in the United States. Commerce Secretary Howard Lutnick said it was proof that Trump’s tariff policy was bringing manufacturing and investment back to America.

Yet neither of these statements is tenable.

At best, the Arizona plant will turn out 3-nanometer chips by 2028, a level Taiwan reached in 2022. For each incremental improvement in the United States and elsewhere, you can be sure Taiwan will have moved commensurately several steps ahead.

Echoing the views of industry experts, the Danish policy analyst Jonas Parello-Plesner, a former diplomat stationed in Washington during Trump’s first term, has written of his confidence that “the holy grail” of cutting-edge chips will remain in Taiwan. Based on his conversations with policymakers in Taiwan, there is no way Taipei is relinquishing its silicon shield.

As for the idea that the chips industry will “return” to U.S. shores at any substantive level, few see this as feasible. First there’s the obvious issue of cost. “Making chips at scale in the U.S. is obviously going to be more expensive,” says Honghong Tinn, author of “Island Tinkerers,” a social history of Taiwan’s computer and semiconductor industry. “So, they would probably need huge subsidies,” continues Tinn, who is also an assistant professor in the Program in the History of Science, Technology, and Medicine and the Department of Electrical and Computer Engineering at the University of Minnesota, Twin Cities.

In fact, a subsidy of at least $6.6 billion was already in place under the CHIPS and Science Act (often shortened to the CHIPS Act), a bipartisan law passed under the Biden administration in 2022. The act oversaw the first stage of TSMC’s Arizona plant, with $65 billion invested to establish the initial fabs at the site. While the precise details are murky, many analysts have argued that further investments in and development of the site were covered by the initial agreement reached under the law.

Having railed against the CHIPS Act as “a horrible, horrible thing” and vowed to repeal it during his 99-minute address to Congress in March, Trump unveiled an “alternative” U.S. Investment Accelerator the following month.

Rather than replacing the CHIPS Act, however, the initiative establishes an office that, aside from some vague references to speeding various deals up, appears to add little to the existing legislation. The office seems largely to be in place to rubber stamp the deals already approved under the CHIPS Act and to attach Trump’s name to them.

Speaking at an event in Taipei in March, just days before the Wei-Trump meeting, Parello-Plesner made that exact point, predicting “a repackaging of the TSMC presence in the U.S.”

Referring to Trump’s “transactional side,” and his obsession with putting his “personal stamp” on things, Parello-Plesner anticipated the rebranding of the deal as “the President Donald J. Trump initiative.”

Absent any details to differentiate it from the provisions and deals that were already in place under the CHIPS Act, there is no indication of how the Trump administration intends to bring back chip manufacturing to the United States.

“Trump’s decisions show not only that he doesn’t trust foreign countries, including Taiwan, but also that he doesn’t understand his own Silicon Valley that well,” says Chang Kuo-hui, a professor and research fellow at National Taiwan University’s Graduate Institute of National Development. “That’s not how it works. Do they really want IC [integrated circuit] manufacturing in the U.S.? It’s not in the spirit of Silicon Valley.”  

There have also been problems with local labor unions in Arizona, which demanded guarantees from TSMC in areas such as worker safety, a commitment to rely more on local labor – which, again calls into question the issue of cost – and support for worker training programs. This dispute began in 2023 during the construction of Fab 21 – described as “the heart” of the Arizona operation – and pushed back completion of that phase by almost a year. 

“If TSMC wants to scale up production, they’ll have to work with the local unions, who were against the way they managed their facilities,” says Bono Shih, a postdoctoral researcher in engineering ethics at Penn State University. Shih, along with Chang and Virginia Tech scholar Gary Downey, coauthored “Engineers and the Two Taiwans,” which – like Tinn’s book – demonstrates the inextricable early links between Taiwanese and American innovators.

The “feud,” as it was referred to by one tech publication, erupted after delays to the installation of infrastructure, equipment, and in particular cleanroom tools. Union leaders bristled at implications by TSMC contractors that local workers were not up to the task. “It is challenging and costly for Taiwanese cleanroom builders to communicate with foreign construction workers in an unfamiliar environment,” one contractor was quoted as saying.

On the surface, these issues were resolved, following an agreement in December 2023. However, Shih suggests the wrangle is indicative of deep-rooted divergences on working conditions and management practices. “It was obvious there are cultural differences,” says Shih. “But the U.S. still wants this, so I think it helps us understand the bargaining power of TSMC, as much as Trump.”

In this regard, Shih – like many other Taiwanese observers – believes TSMC and Taiwan will have got as much out of the deal as Washington. In contrast to Trump’s favored depiction of every negotiation as a zero-sum game, Shih and others frequently refer to a “win-win” scenario.

On top of the subsidies guaranteed by the CHIPS Act, Shih suspects there were other incentives built into the investment agreement, including tax breaks, cheaper electricity, and options on “large strips of land.” With no substantive changes announced to the deal as laid out under the Biden-era act, Trump’s claims of a victory based on TSMC’s capitulation to his tariff threats appear hollow. 

“It allows him to make a big spectacle and a public speech to say, ‘I've got what I wanted. I told you Taiwan would have to pay,’” says Shih.

The standoff over labor rights is not the only workforce-related obstacle, with Chang suggesting that enticing Taiwanese engineers to Arizona may prove tricky. Unlike the problem of local labor, this is not because of insufficient financial incentives but due to the perception that working on less cutting-edge technology could hinder the professional standing and career prospects of those employed at the plant. Referring to a section of “Engineers and the Two Taiwans” that touches on this, Chang highlights the phenomenon of returning Taiwanese professionals who “lamented that they were not receiving enough respect from local engineers.”

This was influenced partly by “a lack of understanding” between U.S.-based engineers and their Taiwan counterparts and the perception that the former are less skilled or qualified. “For the latter, they have grown into a very large community and are seen as successful since Taiwan’s high-tech industry has developed well in the past few decades,” says Chang. “As TSMC’s American fabs are one or two generations behind those in Taiwan, its engineers there might be looked down on.”

This curious role reversal points to the origins of semiconductor production as a transnational enterprise, based on a spirit of cooperation, mutual benefit, and security considerations. This history is lost on Trump. Perhaps this should not be surprising  – even long-time Taiwan watchers and experts have missed or dismissed important elements in this equation, including the foresight that some of the tech trailblazers and officials in Taiwan showed.

“It would be tempting to say that this was all in accordance with a deliberate strategic plan on Taiwan’s part, ” writes Kerry Brown in “The Taiwan Story,” a 2024 work that emphasizes Taiwan’s importance to the West, not just through global supply chains but also in terms of shared democratic values.  “But in the early phases of Taiwan’s economic development… there were no economic blueprints for how economic development would have important national security implications.”

In “Island Tinkerers,” Tinn paints a different picture, arguing that security concerns were leveraged by engineers and academics from the get-go as a way of securing state sponsorship. One means of doing this was by impressing upon officials the connection between advancements in electrical engineering and the development of nuclear power and, thus, security for Taiwan.

Key to this drive were alumni of Chiao Tung University (now National Yang Ming Chiao Tung University), which had been established in Shanghai as Nanyang Public School in 1896, at the tail-end of imperial rule by China’s last dynasty, the Qing.

For decades, this institution had been at the forefront of technical innovation, producing the engineering talent that helped modernize China across fields as diverse as copper mining and telecommunications. Many of these individuals had trained and worked for periods in the United States, Europe, and Japan. 

Following the relocation of the ruling Kuomintang (KMT) and their Republic of China (ROC) government to Taiwan, after years of lobbying, Chiao Tung alumni convinced the KMT to support the re-establishment of the university on the island. Initially lukewarm to the idea, the KMT’s leaders were won over by appeals to security concerns and “Chiang Kai-shek’s favorite topic, atomic energy,” writes Tinn.

These efforts were exemplified in an article by Chiao Tung alumnus Tsao Tsen-cha who had studied, interned, and worked in Germany and the United Kingdom, before relocating permanently to the United States. There, he formed a Chiao Tung alumni U.S. chapter before participating in research on guided missiles for a project contracted by the U.S. Army. In 1958, just as Tsao was working on this research – and, thanks to his and others’ fundraising efforts, Chiao Tung was re-established – the Second Taiwan Strait Crisis erupted.

While months of shelling and attempts to take the ROC-held islands of Kinmen and Matsu off the coast of China by People’s Liberation Army forces ended in stalemate, the conflagration pressed home the urgency of bolstering defense capabilities, with development of a nuclear deterrent considered vital.

As early as 1956, the editor of the Chiao Tung alumni magazine had echoed Tsao’s view that “electronic devices were the basis for utilizing atomic energy and atomic bombs.” The following year, the link was made even more explicit in an article titled “Atoms and Electronics,” written by then-Vice Minister of Transportation and  Communications Chien Chi-chen. Chien’s article recognized not only “the centrality of electronic instruments to atomic energy research and nuclear reactors” but also, conversely, that “electronics science researchers should profit from the widely recognized importance of existing nuclear research.” Crucially, Chien highlighted recent developments in the United States involving the use of computers and transistors in missile launch and detection, aircraft navigation and even data processing.

As a marginalized state, birthed into turmoil following the end of World War II, Taiwan was under threat from the beginning. Add to this the KMT’s decades-long fixation with retaking the “mainland,” and it is clear that the desire to leverage technology and talent in the service of defense was there from the beginning. The Americans were obviously aware of this and, likewise, knew that their assistance, through the dispatch of experts and educators to Taiwan, and the training of Taiwanese engineers in the U.S. could and would be turned to these ends.

It is true that Washington eventually ended Taiwan’s atomic ambitions in the late 1980s by demanding that Taipei cease its nuclear program, after Chang Hsien-yi, deputy director of the Institute of Nuclear Energy Research turned CIA informant, revealed that a bomb was close to completion. But the United States has continued to cooperate with Taiwan on dual-use electronics, to the benefit of both parties.

For all Trump’s bravado, he will be well aware that Taiwanese chips are indispensable to U.S. defense. To take just one example, TSMC makes the advanced AI chips used in U.S. F-35 fighter jets, yet there is no suggestion that these will be produced at the Arizona fab any time soon.

That cooperation between Washington and Taipei turned to Taiwan’s advantage in the 1980s was down to the foresight of planners and policymakers in Taipei and the initially largely U.S.-trained professionals that they recruited. Foremost among these is Morris Chang, who in 1986 was lured from his role as COO at General Instrument in Pennsylvania, by Taiwan’s “godfather of technology,” Li Kwoh-ting, to head the Industrial Technology Research Institute (ITRI).

Contrary to popular belief – or at least widespread reporting in international media – Chang had never lived in Taiwan prior to his recruitment. He had left China for Hong Kong in 1948, then the U.S. the following year, where he majored in mechanical engineering at MIT. On becoming president of ITRI in 1973, Chang was given virtual carte blanche to proceed in the direction he saw fit. Arguments remain over who first came up with the foundry model, but Chang turned it into a reality, founding TSMC with official backing in 1987. 

And here we come to an important point, which Tinn emphasizes in “Island Tinkerers”: The idea that this mass manufacturing of wafers was stolen and transplanted from the U.S. is not only a fiction but ignores an integral piece of the historical puzzle – Washington and the U.S. chips giants of the era rejected the foundry model as “unfeasible.” On multiple trips to the United States to procure investment, Chang was routinely ignored by most of the companies he pitched his plan to and declined by the only two that did receive him – Texas Instruments, where he had worked for 25 years, rising to vice president, and Intel. It is notable that, some 40 years later, Trump specifically named the latter company as a victim of Taiwan’s supposedly underhanded tactics.

Dutch multinational Philips was the only foreign investor, with an initial 27.5 percent stake in TSMC, but American firms soon got on board, By the mid-1990s, the lion’s share of TSMC’s chips were supplied to U.S.-based fabless design firms. And this arrangement has continued to this day: Taiwan concentrates on manufacturing, while the U.S. specializes in design. 

While Trump’s xenophobia may not discriminate, his misplaced accusations about Taiwan are the continuation of long-held but perennially evolving stereotypes of “Chinese” or Asians in the United States.

“Asian Americans and Asian communities in the U.S. have always been seen as a foreign threat, which is tied to their technological advancements,” says Wen Liu, an associate research professor with the Institute of Ethnology at Academia Sinica. “We call this type of xenophobia ‘Techno-Orientalism,’ meaning that it’s fear and othering of Asian supremacy and trying to find ways to control it,” continues Liu, author of the 2024 work “Feeling Asian American: Racial Flexibility between Assimilation and Oppression.”

Outlining the ways in which perceptions of Asianness in the U.S. have evolved from the Chinese Exclusion Act of 1882 through the internment of Japanese Americans during WWII, and the surge in Asian hate crimes during the COVID-19 pandemic, when Trump fanned the fire, infamously dubbing the virus the “kung flu,” Liu sees Trump’s stance on chips as the latest instantiation in a century-plus trajectory of racist views and policies.

“The current geopolitical struggle has the same type of dynamics, where the Trump administration sees the advancement of chip production in Taiwan as a challenge to American supremacy over tech,” says Liu. 

This assessment chimes with Tinn’s observations on media depictions of Asians as “scoundrels,” and “counterfeiters” during the 1980s. Many of the articles and even public hearings into copyright infringement during the era were premised on such depictions and, Tinn reveals, were rife with misrepresentations, inaccuracies, and outright falsehoods. Plus ça change.

There are other historical factors in the “migration” of the chip industry to Taiwan, which have little to do with the transplantation of the technology and much more to do with the emergence of increasingly attractive and lucrative prospects back in Taiwan.

Even by the standards of Taiwan’s tech supremos, Yang Guang-lei’s resume is exemplary, earning him a place in a select group known as the “six knights of R&D.”

Yang, who is better known as Konrad Young outside Taiwan, studied, lived, and worked in the U.S. for most of his life before returning to his homeland in 1995. A Berkeley graduate, he served as TSMC’s director of R&D for 20 years, as well as an adviser for Intel between 2021 to 2023, which led him to conclude that the Taiwanese company has a huge advantage in terms of its R&D system and that Intel’s prospects of replicating a successful foundry model are minimal.

“TSMC is way ahead,” he says.

Having rebuffed several offers from Taiwanese companies in the late 1980s and early 1990s, Young was finally persuaded by a remarkable change in circumstances that suddenly made a return to Taiwan extremely attractive.

To mitigate the relatively paltry salaries on offer, Taiwanese tech firms began offering company stock with a face value of NT$10 per share. These “paper stocks” were not yet publicly traded and of uncertain value. “This was before IPO, and we couldn’t calculate what they were worth,” says Young. “Even at ten dollars, it was less than my salary, so the only reason for me to come back would be patriotism or a better working environment,” he adds.

As the latter was far from encouraging, something had to give. This happened in 1994, with the IPOs of tech companies, including TSMC and Taiwan’s other chip giant United Microelectronics Corporation (UMC) –  also established under ITRI auspices. Stock continued to be issued at NT$10 but now had a market value of up to eight times as much. This built-in bonus more than compensated for the lower salaries and saw a wave of returns to Taiwan.

“We had 30 years of talent reserve –  that’s a lot of people,” says Young. “Suddenly, almost overnight, you had this amazing injection of talent into Taiwan in that period. I was just one of them.”

The stock practice was phased out between 2006 to 2008, likely at least partly under U.S. pressure, but it helped further cement Taiwan’s chip dominance. While state interventions such as this and the targeted tapping of overseas Taiwanese talent demonstrate the importance of top-down strategies, Young – a devout Christian –  suggests that only a divine planner could have envisaged Taiwan’s meteoric rise. 

“Who made this happen?” he asks. “No government could have planned for that. Maybe it doesn’t have to be God – but call it heaven or just an act of fortune.”

As for the possibility of reshoring the semiconductor industry to the U.S., Young challenges the notion that this is impossible. “I think it will happen,” he says. “You have to understand the U.S. was the initiator of the semiconductor industry. They still do the funding and have very strong upstream equipment and material. This is their world.”

Others remain ambivalent. “Technically, TSMC cannot give up its advantages,” says Chang. “And it will struggle to build advanced fabs outside Taiwan,” he adds, noting the regulatory looseness it enjoys on home soil.

He also points out that TSMC’s success is built not only on the company itself but on a tightly woven ecosystem: “flexible regulations, efficient supply chains of materials, a cluster of cooperative firms located in the same region, and partnership among local governments, labor communities, and residents.”

Chang concludes, “This type of environment takes a long time to develop – how long is hard to say, but I would guess at least a decade.”