U.S. President Donald Trump has vowed to impose tariffs on Taiwan’s semiconductor industry and has previously accused Taiwan of stealing the U.S. chip industry. These accusations should be a concern for Taiwan, but—like much of Trump’s talk—is likely intended as a negotiating tactic.
The primary strategic goal for the administration is to revitalize advanced semiconductor manufacturing in the United States. In 2022, the U.S. accounted for less than 1 percent of global fabrication capacity in advanced logic chip manufacturing—although the figure is projected to grow to 28 percent by 2032 as foreign chipmakers, such as Taiwan Semiconductor Manufacturing Company (TSMC) in Arizona and Samsung in Texas, gradually ramp up advanced chip production. Both were incentivized by the Biden administration’s CHIPS and Science Act. Now, Trump is also seeking to rescue struggling U.S. chipmaker Intel—particularly its foundry division, Intel Foundry Services (IFS).
For TSMC, Taiwan’s chip manufacturing giant, maintaining a strong relationship with the Trump administration is critical. The North American market accounts for 70 percent of the company’s revenue. Strained ties with the White House could expose TSMC to heightened regulatory scrutiny—or accusations from Trump or figures close to him.
Similarly, TSMC carefully manages its relationship with the Taiwanese government. While the company enjoys strong support from the administration of President Lai Ching-te due to its critical place in Taiwan’s foreign and security policy and its historically positive ties to the government, its overseas expansion also faces scrutiny from policymakers.
Although the details remain unclear, Trump has pledged to renegotiate key provisions of the CHIPS Act and review the previously approved funding. Any changes to the U.S. commitment to TSMC outline in the act could jeopardize the company’s $65 billion in planned investments in Arizona. So far, TSMC has only received $1.5 billion out of the $6.6 billion in CHIPS Act funding that it was offered during the Biden administration.
The likely impact of tariffs on TSMC could be limited, as the company largely does not directly export finished products to the United States. Instead, TSMC supplies semiconductors to downstream partners for assembly and integration before they reach end customers. Thus, the effect of tariffs on the chip industry could be indirect and manageable unless the U.S. imposes component tariffs—the import duties on specific parts or materials used in manufacturing. TSMC would also be able to raise prices to pass the tariffs to its supply chain partners or end users through proper negotiation. As the negotiations between TSMC and the White House unfold, several options are emerging.
The most discussed option is a deal between TSMC, Intel, the U.S. government, and U.S. chip designers such as Broadcom and Qualcomm. Multiple report indicate that the White House has proposed that would have TSMC acquire a stake in Intel Foundry Services (IFS) and take a leading role in its operations after IFS separated from Intel. Other reports suggest a potential joint venture involving TSMC, Intel, the U.S. government, and industry partners, with technology transfer and technical support from TSMC.
If TSMC took a leading operational role in IFS through stake acquisition or a joint venture, it would face the complex challenge of integrating two fundamentally different foundry ecosystems. This includes coordinating distinct manufacturing technologies, equipment, supply chains, and workforces, all of which comprises an extraordinarily costly and time-consuming process.For instance, TSMC would have to decide whether to continue Intel’s 20A process node as well as the planned 18A and 14A process nodes at IFS, or alternatively, to replace them with TSMC’s N3, upcoming N2, or even future A16 nodes. A “process node” refers to a chip manufacturing technology where smaller nodes—such as TSMC’s 3 nanometer (N3) and 2 nm (N2)—are typically more enhanced than larger nodes in performance, efficiency, and density, though definitions vary by company.
Continuing the use of Intel’s process node risks direct competition with TSMC’s own business while offering an uncertain guarantee of success, given Intel’s past struggles. Conversely, adopting TSMC’s methods would be costly and time-consuming for TSMC and other stakeholders of IFS, as TSMC would have to retrain workers, replace equipment, restructure its supply chain, and more. On top of that, there would likely be serious problems in addressing high-level differences such as working culture and management styles.
TSMC has only established joint ventures for less advanced chips, ranging from 7 nm to 28 nm, to meet specific regional demand. This is evident in its European Semiconductor Manufacturing Company (ESMC) joint venture in Germany with Infineon, NXP, and Bosch, as well as Japan Advanced Semiconductor Manufacturing (JASM) in Japan with Sony, Denso, and Toyota.
A joint venture poses a significant risk of exposing TSMC’s most advanced semiconductor technologies, which was not the case for ESMC and JASM given that those two joint ventures are focusing on less advanced chips. This is not the case for a joint venture with Intel, which is likely focusing on cutting-edge chips. Partnering with a rival foundry to produce cutting-edge chips poses a significant risk of leaking its industry-leading technology This makes the agreement for a joint venture more challenging—especially considering that TSMC’s 3 nm, 5 nm, and 7 nm advanced chips account for nearly 70 percent of its revenue, according to its fourth-quarter earnings in 2024.
Zooming out, both stake acquisition and joint venture would also put TSMC in an awkward position within the chipmaking business landscape. Given the company’s market dominance, it would face a conflict in encouraging clients to use IFS chips instead of its own, unless IFS adopts TSMC’s process node. Many clients are unlikely to wait for this long and uncertain process.
Moreover, it would blur TSMC’s long-standing neutral position as a pure chipmaker for its clients because collaboration with Intel could conflict with clients’ interests. This could erode one of the company’s key advantages over integrated device manufacturers such as Samsung and Intel, which not only manufacture chips but also provide design and packaging services for both internal products and external partners.
There are also regulatory risks to consider. Antitrust reviews could be launched globally against TSMC and Intel over collaboration, given a deal’s potential to create a new form of monopoly in the chipmaking business.
Furthermore, it remains unclear how Intel’s $7.9 billion CHIPS Act funding and $3.5 billion contract with the Department of Defense would be allocated under this partnership. It’s also doubtful whether Trump likes the idea of having a foreign chipmaker running a national chipmaking champion either in the format of joint venture or stake acquisition. While theoretically viable, both TSMC and the White House may hesitate to pursue such a time-consuming and complex deal with foreseeable downside risks.
Although there have been rumors about technology transfers, that possibility remains the least likely outcome of the TSMC-White House negotiations. The company is unlikely to agree to such a deal with the Trump administration, whether through technology transfers, R&D presence in the United States, or intellectual property licensing, as its competitive edge lies in its technological leadership and R&D centered within Taiwan’s semiconductor ecosystem.
Historically, TSMC has been extremely cautious about sharing its technology, especially if said sharing involves its key competitors. Such a move would also contradict its global expansion strategy, which has focused on adding production capacity and fulfilling specific regional demands rather than technology sharing, as we have seen in its past projects in Germany, China, Japan, and the United States.
The Taiwanese government’s position could further complicate the equation. The Lai administration is likely to express objection against any technology transfer or the establishment of TSMC research and development efforts abroad, as such moves could erode Taiwan’s strategic importance in the global economy and supply chain, particularly in the event of a Chinese invasion, thus undermining Taiwan’s national security.
There are more practical options. Although the plan is still in the early stages, TSMC is likely considering building a new advanced packaging plant in Arizona as part of its expanded investment in the United States. This packaging plant is expected to focus on CoWoS (Chip-on-Wafer-on-Substrate) packaging—an essential advanced packaging technology for most AI chips.
This would streamline TSMC’s end-to-end process from fabrication to packaging to delivery, benefiting its U.S. clients. More importantly, this aligns with the Trump administration’s push for a more complete U.S. semiconductor supply chain, as TSMC controls roughly 90 percent of the annual global capacity for CoWoS.
However, the expansion does pose a risk of diluted margins and operational challenges for TSMC due to higher costs and limited revenue generation in Arizona.
The idea of setting up a new advanced packaging plant is a viable one. TSMC has already secured land in Arizona in collaboration with the U.S. government for six factory sites, with three of these already occupied by the company’s three fabrication plants (fabs)—factories where semiconductors are manufactured. To that end, TSMC can build up a new packaging plant in Arizona with fewer regulatory and physical limitations. TSMC’s expanded packaging presence in Arizona could also benefit a local packaging player called Amkor, as TSMC could subcontract a portion of its CoWoS orders under existing partnership with the company. It could also attract potential additional investment from a Taiwanese top packaging firm called ASE Technology, considering the two firms’ existing close cooperation in CoWoS packaging in Taiwan.
Besides, TSMC is likely to accelerate the production timeline for its second and third fabs These two fabs will manufacture cutting-edge 3 nm, 2 nm and A16 chips, aiming to rapidly boost the U.S. chipmaking capability. The company is already rumored to be accelerating the timeline for its third fab in Arizona, with construction set to begin as early as June 2025. This could bring the production timeline of the third fab about a year ahead. Similarly, TSMC aims to begin volume production at its second U.S. fab in the second half of 2027, one year ahead of schedule.
TSMC appears open to expanding its investment in the United States and accelerating its advanced semiconductor production timeline. Rather than pushing for a technology transfer—an unspoken red line for both TSMC and the Taiwanese government—the Trump administration should focus on securing further investment commitments from TSMC and facilitating its collaboration with local players.
Given the risks and complexities of a potential deal involving Intel, TSMC, and other stakeholders, the administration would be better served by directly supporting Intel financially through government funding, regulatory incentives, and contracts. This is important at a time when Intel shows confidence in the capability of its upcoming 18A process node chipmaking to compete with TSMC as well as its strategic role in the U.S. semiconductor industry and national security.
Ultimately, the administration should adopt a dual-track strategy, securing additional investment commitments from TSMC while offering Intel the time and support necessary for a potential turnaround. This approach will strengthen the country’s advanced semiconductor manufacturing and technological leadership without taking unnecessary risks.