After the formal passage of the CHIPS and Science Act in August 2022, uncertainty lingered as to when and how much exactly of the $39 billion promised manufacturing incentives will be delivered. In early 2023, the first Notice of Funding Opportunity (NOFO) for major Commercial Fabrication Facilities with costs exceeding $300 million was issued, [1] followed by a second NOFO for materials and manufacturing equipment suppliers with investment below $300 million in September 2023. [2] CHIPS funding comes in the form of direct funding/grants or loans at a cheaper rate compared to that obtainable from the market [3], supplemented by investment tax credits granted by the Inland Revenue Service. For major fabs, about 5-15% of projected capital expenditures cost will be supported by direct funding in addition to the tax credits, while for materials and equipment suppliers, 10% of projected CAPEX will be the default. For foreign firms like TSMC, which have committed substantial investment in anticipation of the CHIPS Act subsidies, the exact amount of the subsidies and the terms attached are of much importance. Not just because they determine the projects’ short-term commercial viability but also because they signal the current administration’s strategic intentions that shape long-run relationships. Ascertaining the core tenets of the American semiconductor strategy is critical for Taiwanese firms and the Taiwanese government, as it enables understanding of what attractive offers they could make in future negotiations with US policymakers.
This piece examines the funding selection criteria and conditions spelled out in the two rounds of the Notice of Funding Opportunity (NOFO) issued by CHIPS for America, which sheds light on the multi-dimensional objectives the Biden administration is attempting to achieve through the CHIPS program and its broader overall semiconductor strategy. The terms of the CHIPS program are crafted to reflect multiple objectives that the Biden administration is pursuing: (1) restrain China’s progress in advanced nodes and stifle the development of Chinese computational and AI capabilities; (2) ensure the stability of the supply chain to mature node semiconductors essential to the vehicle, electronics, and military equipment production; (3) the relocation of “good manufacturing jobs” coupled with sufficient training programs and community investment to localities distressed by US deindustrialization, providing economic opportunities to alleviate perceived structural inequities; (4) ensure long-run commercial profitability and competitiveness of semiconductor manufacturing in the US through improving economic-industrial fundamentals and rebuilding a viable supply chain ecosystem.
The CHIPS Act is designed to kill two birds with one stone: through subsidy incentives, key global semiconductor manufacturing players are induced to invest in America and refrain from engagement with China, which in principle serves both domestic industrial policy and foreign security policy goals. In exchange for the subsidies that will be distributed only conditional on reaching agreed-upon milestones, the recipients have to implement plans and measures that accord with the funding terms. As long as the subsidies remain lucrative enough for participation to make commercial sense under the current geopolitical environment, the terms would constrain the firms to relocate substantial investment capacity to the United States in the medium to long term while closing immediate possibilities to expand capacity or conduct R&D exchanges with China. The next few paragraphs discuss the specific terms of the NOFO funding conditions that correspond to each of the four objectives and their implications to major recipients of CHIPS funding, especially to foreign firms like TSMC and Samsung.
Guardrails against China
For competition with China in advanced, current-generation, and mature node semiconductors, the most salient conditions to CHIPS funding are the “guardrails” clauses relating to “foreign entities of concern” which targets mainly China. [4] Obviously projects directly or indirectly controlled by foreign entities of concern will not be funded, but also there are “clawback clauses” pertaining to any possible future joint R&D or technology licensing agreement with Chinese entities. Furthermore, for 10 years beginning on the date of award, the recipient could no longer expand manufacturing capacity in China, except only if the expansion in question relates to legacy semiconductors that predominantly serve foreign countries of concern. [5] CHIPS Act documents describe legacy logic semiconductors as technologies relating to the 28-nanometer generation or older [6], and mature node semiconductors, in general, are defined as (a) logic and analog chips that are not based on FinFET, post-FinFET or any other sub-28 nm transistor architectures; (b) discrete semiconductor devices such as diodes and transistors; (c) optoelectronics and optical semiconductors; and (d) sensors. [7] Thus, restraints on semiconductor manufacturing in China pertain mostly to leading-edge (defined as logic chips produced by EUV, 3D NAND with > 200 layers, and DRAM with half-pitch below 13 nm) and current generation semiconductors (not leading edge up to 28nm), where Chinese manufacturers did not have yet the competitive technology, reliable and cost-efficient manufacturing equipment, and capacity to serve domestic needs. Yet for mature node capacity, expansion in foreign countries of concern is still eventually restricted to be at most 10 percent of original capacity, and mature chips relating to natural security concerns, such as those with defense or sensing applications, cannot be manufactured in China. [8]
Failure to comply with conditions in the clawback clauses would allow the Secretary of Commerce to recover the full amount of assistance provided. Thus, the CHIPS subsidies now become a bargaining chip that dynamically increases the opportunity cost of expanding capacity in China. This, in combination with existing export control measures and other latent policy tools, means that it is unlikely that Intel, Samsung, TSMC, Micron, and other announced recipients will make any significant moves to build or expand their presence in China in the foreseeable future. Although current geopolitical tensions have already made such moves unlikely, formally enshrining this in the CHIPS agreement provides a commitment that is strategically valuable. Any lingering uncertainty as to whether firms like TSMC and Samsung could continue to serve and expand operations in technologies more advanced than 28nm in China is resolved and communicated to all parties, including the Chinese government, which may be interested in implementing similar policies to subsidize foreign logic and memory manufacturers. Furthermore, as the condition is a 10-year commitment, this implies that facilities with processes more advanced than 28nm that may become mature in the upcoming years will still be presumably prohibited from being built in China.
（Source：iStock）
Mature Node and Leading Edge
There is a specific emphasis in the documents on supporting current generation and mature node semiconductors that are essential inputs to a set of “critical manufacturing sectors”, including automobiles, aerospace, and defence. This is in part motivated by the 2020 chip shortage that saliently affected automobile production and other basic electronics. Nevertheless, a less discussed but critical supply chain vulnerability for the US is access to semiconductor inputs for military purposes, which the CHIPS Act could remedy by supporting domestic manufacturing capacity designated and verified to be safe for military production, which commands a higher cost compared to technologically equivalent products.
An additional 2 billion dollars is specifically allocated to mature node manufacturing, which as of now seems will mostly be allocated to GlobalFoundries, the key supplier of chips for defense purposes by being the only commercially high-volume “Trusted Foundry” of the Department of Defence. [9] GlobalFoundries have two accredited trusted foundries in Vermont and Malta, New York that have gone through extensive security checks of both the staff and the physical & electronic production processes to prevent sabotage, intellectual property theft, or violation of export control restrictions. Now, the CHIPS funding will be directed to support new facilities at the two trusted fabs that will deliver 22-nanometer 22-FDX chips to the Department of Defense, accordingly, the most advanced technology available for satellites and other defense uses. [10]
For leading-edge technologies, an interesting requirement for CHIPS Act recipients is to participate in various CHIPS Act R&D programs, including the National Semiconductor Technology Center (NSTC) and the National Advanced Packaging Manufacturing Program (NAPMP), dedicated to the development of new semiconductor products and test, assembly, and packaging technologies respectively. Participants have to commit to reserve manufacturing capacity at below market rates to projects supported by the NSTC and NAPMP and also to rotate technical staff to or help train technical staff for these two entities. The participants may also need to disclose process data, provide NSTC/NAPMP staff access to R&D facilities and equipment/design tools, or even disclose to the public or industry process design kits (PDKs) of mature processes through open-source licenses. [11] A more cynical view of these requirements is that they are not quid pro quo scientific exchanges but venues that induce technology transfers of critical know-how and trade secrets of foreign firms to the US. Applicants with advanced technologies must, therefore, weigh the benefits of the subsidies and continued cordial relationship with the US government against the potential costs of accommodating federal R&D efforts to future business profits and competitiveness.
Workplace Development and Community Investment
Another strong focus of CHIPS Act funding conditions is on workplace development and community investment. A key prior condition to acquire federal CHIPS funding is to obtain subsidy packages from state or local governments, who are motivated to have the recipient commit to long-term investment and community building at the local level.[12] By tying firms to local governments, CHIPS funding could piggyback on local governments’ knowledge and relationships to ensure the funds are allocated to productive ends suitable for long-term community development. Thus CHIPS funding is designed to be place-based rather than firm-based: i.e., awards are targeted to individual manufacturing sites and surrounding community development projects at a locality rather than to a specific firm. Firm-specific local subsidies such as direct tax abatements that do not tie incentives to specific project development are therefore discouraged and will be discounted in the federal government’s evaluation. Also, the use of funds for stock buybacks or other schemes that enrich corporate shareholders is prohibited. As in general, money is fungible, and firms can always indirectly use the expanded budget constraint from the subsidies to spend on shareholders, future stock buybacks are discouraged and may hurt the chances of getting awarded funding.
Applicants to the funds must detail plans to educate and train local manufacturing and construction workers and support them with sufficient pay, benefits, and amenities that include childcare support. Training relatively low-skilled semiconductor technicians in collaboration with community college systems is a big focus of the program. The often-cited US shortage in talent for semiconductor manufacturing centers on the lack of engineers and R&D talent that manufacturers find hard to retain due to superior pay and work conditions in the chip design and software sectors. On the other hand, theoretically, a few weeks to one or two years of training could prepare anyone without prior experience or education in semiconductor-related fields for a job in the fab. [13] Thus, the CHIPS NOFO stipulates that firms must offer a comprehensive plan to help local governments train the workforce to meet future needs, drawing from those that have been earning much less than a semiconductor technician. Such jobs should still provide less skilled and educated workers with better pay and work conditions than what they could have earned elsewhere. The policy thus targets the segment of the US population that has been underemployed due to deindustrialization. Had manufacturing remained competitive in America, low-skilled workers could have been employed in higher value-added manufacturing jobs rather than in the services. In contrast, hardware engineering jobs that have to compete with software engineers and chip designers are not as emphasized in the CHIPS subsidies workforce plan requirement. This reflects both practical considerations on the relative advantages and appeal of US industries in the labor market and the political-redistributional motive to bring “good manufacturing jobs” back to demographics that have been staunch supporters of the Democratic Party until recently.
Commitments to promote diversity, equity, and inclusion goals are also expected, rewarding serious plans to train and employ women and minority groups and projects that contract with minority and women-owned businesses. CHIPS funding may also be contingent on collaboration with local education and community groups as well as labor unions. To compete for CHIPS funding, Micron made concessions in a project labor agreement with unions to prioritize hiring union labor [14] and has made commitments to labor peace talks (to allow workers to organize) as a part of its 6.1 billion CHIPS deal. [15] Community investments that address local or the workforce’s needs to issues like transportation, housing, or education are encouraged, as it is expected that such investments that address local deficiencies would also help with the progress of the CHIPS project in question. Thus, the CHIPS Act funding can also be considered an indirect redistribution program that pushes firms to target jobs, education, and other development opportunities to minority groups, females, veterans, and small businesses that the government wishes to promote at a fraction of a cost to direct distribution of funding. Such public-private partnership comes with a few advantages theoretically: first, the costs of such welfare distribution are shared between private firms and the government; second, private profit-seeking organizations should be better than bureaucrats in creating workforce development programs that actually train useful and productive workers.
However, these requirements impose extra costs on the recipients, especially to foreign firms unfamiliar with navigating American unions and minority communities. It is also unclear how well TSMC and Samsung could replace local bureaucratic and social welfare institutions in supporting the community. Concerns have surfaced recently on whether they could train technicians adequately and cost-efficiently without fostering workplace conflict between workers and managers of different cultural backgrounds. Delays in the progress of fabs have also been creating supply-demand mismatches: graduates of short semiconductor technician programs that have swiftly started after the announcement of CHIPS investments found that they have no jobs to apply to. Any skills learned from the programs would be lost before the fabs actually go into operation and start employing workers, rendering the programs ineffective. Once the large-scale projects got up and running, there might still be friction in integrating workers into actual production processes and managerial practices. Semiconductor business cycles may also affect the continued employability of these lesser-skilled technicians. With CHIPS funding and associated semiconductor industrial policy being largely tied to labor requirements, firms must leverage the US government’s interest in supporting work in disadvantaged communities to make sure public funding shares some costs of training and employing Americans.
（Source：iStock）
Commercial Viability and Taxpayer Protection
Finally, applicants must submit evidence on the commercial viability and financial strength of the projects, including the competitiveness of future products, sources of finance, likely customers, potential downside risks and mitigation measures, and purchase commitments and collaborations across the supply chains. The documents thus recognize that most manufacturing projects without subsidies are less likely to make economic sense, and applicants have to demonstrate both the necessity and sufficiency of how could the CHIPS funding be utilized for profitable opportunities. Adequate screening of applicants is warranted for the limited funding to be directed to ends most likely to be economically successful. In the end, there is a shared interest between the US government and the recipients in the commercial successes of the manufacturing projects to rebuild US semiconductor manufacturing capacity. If the CHIPS Act were to be a serious industrial policy program, it should not give constant unconditional support to some national champions without consideration of their profitability. The belief that U.S. semiconductor manufacturing is a potential “infant industry” that just needs proper nurturing to grow is the core tenet of the program. Once initial roadblocks are overcome, increasing returns through production and supply chain coordination would be realized, making semiconductor manufacturing viable in the US eventually without government support.
To ensure that CHIPS funding is distributed efficiently for their purposes, “taxpayer protection” clauses are designed to (a) recover profits from projects that are too successful and (b) prevent future funding to failing projects or those that are noncompliant with the requirements. To ensure that the incentives are not merely subsidizing production that will occur with or without the CHIPS Act, applicants have to justify that the CHIPS incentives are necessary for the investment to take place. Major recipients receiving more than $150 million are also required to share with the US government a portion of cash flows that exceed the recipients’ projections above a pre-negotiated threshold. This first induced firms to not strategically underreport their projected returns so as to satisfy construction and operational milestones, facilitating the monitoring of firm outcomes. Secondly, it makes sure that the government has some equity in the less likely but possible event that the CHIPS Act becomes extraordinarily successful. Thus, for major funding recipients, the design of CHIPS subsidies also provides some form of insurance that sacrifices some profit in good times to exchange for more funding support in the initial difficult phases of the investments. However, a downside of the profit-sharing clauses is that they might prevent firms from voluntarily investing or allocating extra capacity to the awarded projects to prevent profits from exceeding the threshold that warrants profit-sharing. Periodic report requirements can partially ameliorate this issue through regular checkups to see whether firms adhere to their promises. Much of CHIPS funding will be disbursed in tranches contingent on the reaching of negotiated milestones relating to capital expenditures, workplace development, and operational status. This ensures that not too much funding will be allocated to projects that eventually fail.
Understanding that a vibrant semiconductor ecosystem involves not just manufacturing giants but also smaller firms across the supply chain, the second NOFO targets smaller materials and equipment suppliers whose investment will complement the big manufacturers’ operations. Total funding for this scheme amounts to $500 million, much less than the billions awarded to major manufacturers. Each small-scale facility project is expected to be subsidized by around ten percent of the projected capital expenditures cost. [16] Applicants are evaluated by similar requirements as that of commercial fabrication facilities, such as workforce development, economic and national security, community investment, and commercial viability. To promote the development of regional clusters, applicants are encouraged to apply as a part of a consortium of multiple suppliers, local government, and an anchor institution[17], like a semiconductor fab. Small-scale projects that are not part of the cluster have a higher burden of proof to prove that they can strengthen supply chain resilience or advance U.S. technology leadership. State and local governments are strongly encouraged to support such a cluster by implementing corresponding policies in workforce education, infrastructure investment, and the streamlining of regulations. Such an approach again points to the place-based nature of the CHIPS subsidies that do not reward any single set of champion firms. Instead, both the big fish and other smaller organisms in a regional semiconductor ecosystem are rewarded. Taiwanese suppliers interested in joining TSMC’s venture in Arizona must, therefore, anticipate making long-term commitments to Phoenix’s local community and economy to receive support from CHIPS funding.
US Multidimensional Strategy: Taiwan’s Response
For recipients of the CHIPS program, the announcement of the subsidies is just the beginning of a dynamic game with federal and state governments as well as local communities. Repeated renegotiations with respect to both new investments and existing agreements are expected, as there will be economic cycles and geopolitical shocks that change fundamentals that warrant renegotiations and adjustments of past agreements. The Federal government surely cannot be satisfied with the extent of investments that TSMC, Samsung, Intel, and Micron have committed, and will demand more fabs with more advanced technologies to be built. For Taiwanese firms such as TSMC and its suppliers committed to America for the long term, constant communication and coordination with federal agencies are key to making sure that subsidy allocation and the terms attached are in their best interest. The Taiwanese government should also participate in the coordination process by leveraging their access to Taiwanese firms that the US policymakers are eager to attract. The announcement pertaining to the guardrails provisions mentions the importance of maintaining coordination and close contacts with U.S. partners and allies, including Korea, Japan, India, and the UK. Taiwan was conspicuously left out of the discussion despite TSMC’s investment being perhaps the most high-profile and celebrated project funded by the CHIPS Act. [18]However, the Taiwanese government has an arsenal of policy tools that could determine the relative attractiveness of Taiwanese locations to that of the US for Taiwanese semiconductor firms. The government should leverage this fact to remind U.S. policymakers that communication and coordination with the Taiwanese government prevent excessive subsidy competition for TSMC and other star Taiwanese firms.
To navigate the complex US policy process motivated by different objectives, maintaining cordial relationships and trust with the US government is essential for Taiwanese firms. Taiwanese firms committed to investing in the US have to be prepared to cut ties with China. They will first have to refrain from future expansion in China and should consider phasing out Chinese production to mature technologies less susceptible to regulatory scrutiny. Taiwanese firms should also recognize US policymakers’ interest in regional community building and workforce development. To secure equal footing in future funding decisions, they should be prepared to make commitments that go against their desire to minimize costs. It is unlikely that the US government is interested in continuing to support Taiwanese fabs in America that employ mostly Taiwanese immigrant workers rather than locals, so figuring out how to efficiently train and employ American workers without sacrificing productivity is essential. Finally, to not repeat the Foxconn debacle in Wisconsin, Taiwanese firms should not back out of their commitments easily. The US government is interested in long-term investment in a local community resulting from joint efforts from multiple firms to form a viable industrial cluster. Mistakes made by the Trump Administration regarding Wisconsin will not be easily repeated, and Taiwanese firms must be wary of the financial and reputational consequences of backing out of their deal. Thus the stipulated project milestones must be negotiated carefully beforehand, supported by informed research of on-the-ground conditions. In future negotiations on new projects or renegotiations on existing deals, firms must invoke prospects that appeal to US policymakers, such as aligning the sustained profitability of the project to local economic wellbeing. The Taiwanese government can play an indirect or direct role in the process by controlling the costs of foreign investments with corresponding domestic policies such as tax abatements on suppliers undertaking significant investments abroad. These policy tools potentially could be great bargaining chips for Taiwanese firms. They could either increase the value of Taiwanese firms’ outside options by making investment in Taiwan more attractive, which warrants higher.
US subsidies to compensate or reduce the costs of Taiwanese foreign investment efforts.
Conclusion
Previously interconnected semiconductor supply chains are being split into two blocs. One bloc is that of China, with a burgeoning domestic supply chain serving mainly Chinese customers. The other is composed of US allies that remain interconnected, though each allied country is instituting significant efforts to promote domestic firms and building domestic supply chain capacity to avoid overreliance on other allies. In an age where semiconductor R&D and production decisions are driven increasingly by political logic rather than economic rationality, economic investment, and production costs should be weighed against the costs stemming from geopolitical uncertainty. Rising economic costs from cutting Chinese ties and investing in America should then be considered a price for security and supply chain risk diversification. The price could be tolerable provided that the US and allied governments provide sufficient incentives, but obtaining good deals with the US government requires a proper understanding of US objectives.
The simultaneous emphasis on workforce development, national security, supply chain stability/economic security, and commercial viability in the CHIPS NOFOs, encapsulated in the terms and conditions for funding, reflect the current administration’s view that all four objectives are mutually complementary rather than conflicting with each other. Under the belief that disadvantaged populations and communities could be adequately trained to serve in semiconductor manufacturing and that reasonably cost-efficient supply chain clusters could be set up given proper incentives, CHIPS funding is viewed as a necessary remedy to address various market failures that have hitherto constrained semiconductor manufacturing competitiveness. The substantial investment commitments of the CHIPS participants are anticipated to kick start a virtuous cycle of continued investment and commercial production that lowers production costs over time, contrary to suggestions that the CHIPS Act will be another failed industrial policy that does not respect differences in comparative advantage between the US and East Asia. [19]
Whether the grand economic strategy could succeed in the long run remains to be seen, and significant costs and construction/production delays will almost surely persist in the short run. Yet, now that most of the heavily anticipated first-round funding to major recipients like TSMC, Intel, and Samsung has been announced, some preliminary strategic objectives of the CHIPS Act have been achieved. The semiconductor giants’ acceptance of the CHIPS funding and the conditions attached entail acknowledgment of or even agreement to the Biden administration’s strategic priorities. This, along with the significant resources committed to projects in America, will steer the subsidy recipients towards an investment and business roadmap that will dynamically reduce the costs of continued investment in America while increasing the costs of re-engagement with China, solidifying the economic incentives supporting adherence to the American semiconductor strategy. Now that the Taiwanese government and firms have taken the path to embrace such a strategy, they must learn how to accommodate these US policy goals without sacrificing too much self-interest. Continued dialogue and careful identification of common grounds of mutual interest are then essential to maintain profitable relationships into the distant future.
[4] 15 U.S.C. § 4652(6) (C) (i): “ the covered entity shall enter into an agreement with the Secretary specifying that, during the 10-year period beginning on the date of the award, subject to clause (ii), the covered entity may not engage in any significant transaction, as defined in the agreement, involving the material expansion of semiconductor manufacturing capacity in the People’s Republic of China or any other foreign country of concern.”
[5] 15 U.S.C. § 4652(6) (C) (ii):The prohibition in the agreement required under clause (i) shall not apply to—
(I)existing facilities or equipment of a covered entity for manufacturing legacy semiconductors; or
(II)significant transactions involving the material expansion of semiconductor manufacturing capacity that—
(aa)produces legacy semiconductors; and
(bb)predominately serves the market of a foreign country of concern.
2024 / 5 / 22
Assessing Interconnected Security and Socioeconomic objectives in the CHIPS Act Funding Conditions: How should Taiwan Respond?
作者：Ming-Yen Ho
After the formal passage of the CHIPS and Science Act in August 2022, uncertainty lingered as to when and how much exactly of the $39 billion promised manufacturing incentives will be delivered. In early 2023, the first Notice of Funding Opportunity (NOFO) for major Commercial Fabrication Facilities with costs exceeding $300 million was issued, [1] followed by a second NOFO for materials and manufacturing equipment suppliers with investment below $300 million in September 2023. [2] CHIPS funding comes in the form of direct funding/grants or loans at a cheaper rate compared to that obtainable from the market [3], supplemented by investment tax credits granted by the Inland Revenue Service. For major fabs, about 5-15% of projected capital expenditures cost will be supported by direct funding in addition to the tax credits, while for materials and equipment suppliers, 10% of projected CAPEX will be the default. For foreign firms like TSMC, which have committed substantial investment in anticipation of the CHIPS Act subsidies, the exact amount of the subsidies and the terms attached are of much importance. Not just because they determine the projects’ short-term commercial viability but also because they signal the current administration’s strategic intentions that shape long-run relationships. Ascertaining the core tenets of the American semiconductor strategy is critical for Taiwanese firms and the Taiwanese government, as it enables understanding of what attractive offers they could make in future negotiations with US policymakers.
This piece examines the funding selection criteria and conditions spelled out in the two rounds of the Notice of Funding Opportunity (NOFO) issued by CHIPS for America, which sheds light on the multi-dimensional objectives the Biden administration is attempting to achieve through the CHIPS program and its broader overall semiconductor strategy. The terms of the CHIPS program are crafted to reflect multiple objectives that the Biden administration is pursuing: (1) restrain China’s progress in advanced nodes and stifle the development of Chinese computational and AI capabilities; (2) ensure the stability of the supply chain to mature node semiconductors essential to the vehicle, electronics, and military equipment production; (3) the relocation of “good manufacturing jobs” coupled with sufficient training programs and community investment to localities distressed by US deindustrialization, providing economic opportunities to alleviate perceived structural inequities; (4) ensure long-run commercial profitability and competitiveness of semiconductor manufacturing in the US through improving economic-industrial fundamentals and rebuilding a viable supply chain ecosystem.
The CHIPS Act is designed to kill two birds with one stone: through subsidy incentives, key global semiconductor manufacturing players are induced to invest in America and refrain from engagement with China, which in principle serves both domestic industrial policy and foreign security policy goals. In exchange for the subsidies that will be distributed only conditional on reaching agreed-upon milestones, the recipients have to implement plans and measures that accord with the funding terms. As long as the subsidies remain lucrative enough for participation to make commercial sense under the current geopolitical environment, the terms would constrain the firms to relocate substantial investment capacity to the United States in the medium to long term while closing immediate possibilities to expand capacity or conduct R&D exchanges with China. The next few paragraphs discuss the specific terms of the NOFO funding conditions that correspond to each of the four objectives and their implications to major recipients of CHIPS funding, especially to foreign firms like TSMC and Samsung.
Guardrails against China
For competition with China in advanced, current-generation, and mature node semiconductors, the most salient conditions to CHIPS funding are the “guardrails” clauses relating to “foreign entities of concern” which targets mainly China. [4] Obviously projects directly or indirectly controlled by foreign entities of concern will not be funded, but also there are “clawback clauses” pertaining to any possible future joint R&D or technology licensing agreement with Chinese entities. Furthermore, for 10 years beginning on the date of award, the recipient could no longer expand manufacturing capacity in China, except only if the expansion in question relates to legacy semiconductors that predominantly serve foreign countries of concern. [5] CHIPS Act documents describe legacy logic semiconductors as technologies relating to the 28-nanometer generation or older [6], and mature node semiconductors, in general, are defined as (a) logic and analog chips that are not based on FinFET, post-FinFET or any other sub-28 nm transistor architectures; (b) discrete semiconductor devices such as diodes and transistors; (c) optoelectronics and optical semiconductors; and (d) sensors. [7] Thus, restraints on semiconductor manufacturing in China pertain mostly to leading-edge (defined as logic chips produced by EUV, 3D NAND with > 200 layers, and DRAM with half-pitch below 13 nm) and current generation semiconductors (not leading edge up to 28nm), where Chinese manufacturers did not have yet the competitive technology, reliable and cost-efficient manufacturing equipment, and capacity to serve domestic needs. Yet for mature node capacity, expansion in foreign countries of concern is still eventually restricted to be at most 10 percent of original capacity, and mature chips relating to natural security concerns, such as those with defense or sensing applications, cannot be manufactured in China. [8]
Failure to comply with conditions in the clawback clauses would allow the Secretary of Commerce to recover the full amount of assistance provided. Thus, the CHIPS subsidies now become a bargaining chip that dynamically increases the opportunity cost of expanding capacity in China. This, in combination with existing export control measures and other latent policy tools, means that it is unlikely that Intel, Samsung, TSMC, Micron, and other announced recipients will make any significant moves to build or expand their presence in China in the foreseeable future. Although current geopolitical tensions have already made such moves unlikely, formally enshrining this in the CHIPS agreement provides a commitment that is strategically valuable. Any lingering uncertainty as to whether firms like TSMC and Samsung could continue to serve and expand operations in technologies more advanced than 28nm in China is resolved and communicated to all parties, including the Chinese government, which may be interested in implementing similar policies to subsidize foreign logic and memory manufacturers. Furthermore, as the condition is a 10-year commitment, this implies that facilities with processes more advanced than 28nm that may become mature in the upcoming years will still be presumably prohibited from being built in China.
Mature Node and Leading Edge
There is a specific emphasis in the documents on supporting current generation and mature node semiconductors that are essential inputs to a set of “critical manufacturing sectors”, including automobiles, aerospace, and defence. This is in part motivated by the 2020 chip shortage that saliently affected automobile production and other basic electronics. Nevertheless, a less discussed but critical supply chain vulnerability for the US is access to semiconductor inputs for military purposes, which the CHIPS Act could remedy by supporting domestic manufacturing capacity designated and verified to be safe for military production, which commands a higher cost compared to technologically equivalent products.
An additional 2 billion dollars is specifically allocated to mature node manufacturing, which as of now seems will mostly be allocated to GlobalFoundries, the key supplier of chips for defense purposes by being the only commercially high-volume “Trusted Foundry” of the Department of Defence. [9] GlobalFoundries have two accredited trusted foundries in Vermont and Malta, New York that have gone through extensive security checks of both the staff and the physical & electronic production processes to prevent sabotage, intellectual property theft, or violation of export control restrictions. Now, the CHIPS funding will be directed to support new facilities at the two trusted fabs that will deliver 22-nanometer 22-FDX chips to the Department of Defense, accordingly, the most advanced technology available for satellites and other defense uses. [10]
For leading-edge technologies, an interesting requirement for CHIPS Act recipients is to participate in various CHIPS Act R&D programs, including the National Semiconductor Technology Center (NSTC) and the National Advanced Packaging Manufacturing Program (NAPMP), dedicated to the development of new semiconductor products and test, assembly, and packaging technologies respectively. Participants have to commit to reserve manufacturing capacity at below market rates to projects supported by the NSTC and NAPMP and also to rotate technical staff to or help train technical staff for these two entities. The participants may also need to disclose process data, provide NSTC/NAPMP staff access to R&D facilities and equipment/design tools, or even disclose to the public or industry process design kits (PDKs) of mature processes through open-source licenses. [11] A more cynical view of these requirements is that they are not quid pro quo scientific exchanges but venues that induce technology transfers of critical know-how and trade secrets of foreign firms to the US. Applicants with advanced technologies must, therefore, weigh the benefits of the subsidies and continued cordial relationship with the US government against the potential costs of accommodating federal R&D efforts to future business profits and competitiveness.
Workplace Development and Community Investment
Another strong focus of CHIPS Act funding conditions is on workplace development and community investment. A key prior condition to acquire federal CHIPS funding is to obtain subsidy packages from state or local governments, who are motivated to have the recipient commit to long-term investment and community building at the local level.[12] By tying firms to local governments, CHIPS funding could piggyback on local governments’ knowledge and relationships to ensure the funds are allocated to productive ends suitable for long-term community development. Thus CHIPS funding is designed to be place-based rather than firm-based: i.e., awards are targeted to individual manufacturing sites and surrounding community development projects at a locality rather than to a specific firm. Firm-specific local subsidies such as direct tax abatements that do not tie incentives to specific project development are therefore discouraged and will be discounted in the federal government’s evaluation. Also, the use of funds for stock buybacks or other schemes that enrich corporate shareholders is prohibited. As in general, money is fungible, and firms can always indirectly use the expanded budget constraint from the subsidies to spend on shareholders, future stock buybacks are discouraged and may hurt the chances of getting awarded funding.
Applicants to the funds must detail plans to educate and train local manufacturing and construction workers and support them with sufficient pay, benefits, and amenities that include childcare support. Training relatively low-skilled semiconductor technicians in collaboration with community college systems is a big focus of the program. The often-cited US shortage in talent for semiconductor manufacturing centers on the lack of engineers and R&D talent that manufacturers find hard to retain due to superior pay and work conditions in the chip design and software sectors. On the other hand, theoretically, a few weeks to one or two years of training could prepare anyone without prior experience or education in semiconductor-related fields for a job in the fab. [13] Thus, the CHIPS NOFO stipulates that firms must offer a comprehensive plan to help local governments train the workforce to meet future needs, drawing from those that have been earning much less than a semiconductor technician. Such jobs should still provide less skilled and educated workers with better pay and work conditions than what they could have earned elsewhere. The policy thus targets the segment of the US population that has been underemployed due to deindustrialization. Had manufacturing remained competitive in America, low-skilled workers could have been employed in higher value-added manufacturing jobs rather than in the services. In contrast, hardware engineering jobs that have to compete with software engineers and chip designers are not as emphasized in the CHIPS subsidies workforce plan requirement. This reflects both practical considerations on the relative advantages and appeal of US industries in the labor market and the political-redistributional motive to bring “good manufacturing jobs” back to demographics that have been staunch supporters of the Democratic Party until recently.
Commitments to promote diversity, equity, and inclusion goals are also expected, rewarding serious plans to train and employ women and minority groups and projects that contract with minority and women-owned businesses. CHIPS funding may also be contingent on collaboration with local education and community groups as well as labor unions. To compete for CHIPS funding, Micron made concessions in a project labor agreement with unions to prioritize hiring union labor [14] and has made commitments to labor peace talks (to allow workers to organize) as a part of its 6.1 billion CHIPS deal. [15] Community investments that address local or the workforce’s needs to issues like transportation, housing, or education are encouraged, as it is expected that such investments that address local deficiencies would also help with the progress of the CHIPS project in question. Thus, the CHIPS Act funding can also be considered an indirect redistribution program that pushes firms to target jobs, education, and other development opportunities to minority groups, females, veterans, and small businesses that the government wishes to promote at a fraction of a cost to direct distribution of funding. Such public-private partnership comes with a few advantages theoretically: first, the costs of such welfare distribution are shared between private firms and the government; second, private profit-seeking organizations should be better than bureaucrats in creating workforce development programs that actually train useful and productive workers.
However, these requirements impose extra costs on the recipients, especially to foreign firms unfamiliar with navigating American unions and minority communities. It is also unclear how well TSMC and Samsung could replace local bureaucratic and social welfare institutions in supporting the community. Concerns have surfaced recently on whether they could train technicians adequately and cost-efficiently without fostering workplace conflict between workers and managers of different cultural backgrounds. Delays in the progress of fabs have also been creating supply-demand mismatches: graduates of short semiconductor technician programs that have swiftly started after the announcement of CHIPS investments found that they have no jobs to apply to. Any skills learned from the programs would be lost before the fabs actually go into operation and start employing workers, rendering the programs ineffective. Once the large-scale projects got up and running, there might still be friction in integrating workers into actual production processes and managerial practices. Semiconductor business cycles may also affect the continued employability of these lesser-skilled technicians. With CHIPS funding and associated semiconductor industrial policy being largely tied to labor requirements, firms must leverage the US government’s interest in supporting work in disadvantaged communities to make sure public funding shares some costs of training and employing Americans.
Commercial Viability and Taxpayer Protection
Finally, applicants must submit evidence on the commercial viability and financial strength of the projects, including the competitiveness of future products, sources of finance, likely customers, potential downside risks and mitigation measures, and purchase commitments and collaborations across the supply chains. The documents thus recognize that most manufacturing projects without subsidies are less likely to make economic sense, and applicants have to demonstrate both the necessity and sufficiency of how could the CHIPS funding be utilized for profitable opportunities. Adequate screening of applicants is warranted for the limited funding to be directed to ends most likely to be economically successful. In the end, there is a shared interest between the US government and the recipients in the commercial successes of the manufacturing projects to rebuild US semiconductor manufacturing capacity. If the CHIPS Act were to be a serious industrial policy program, it should not give constant unconditional support to some national champions without consideration of their profitability. The belief that U.S. semiconductor manufacturing is a potential “infant industry” that just needs proper nurturing to grow is the core tenet of the program. Once initial roadblocks are overcome, increasing returns through production and supply chain coordination would be realized, making semiconductor manufacturing viable in the US eventually without government support.
To ensure that CHIPS funding is distributed efficiently for their purposes, “taxpayer protection” clauses are designed to (a) recover profits from projects that are too successful and (b) prevent future funding to failing projects or those that are noncompliant with the requirements. To ensure that the incentives are not merely subsidizing production that will occur with or without the CHIPS Act, applicants have to justify that the CHIPS incentives are necessary for the investment to take place. Major recipients receiving more than $150 million are also required to share with the US government a portion of cash flows that exceed the recipients’ projections above a pre-negotiated threshold. This first induced firms to not strategically underreport their projected returns so as to satisfy construction and operational milestones, facilitating the monitoring of firm outcomes. Secondly, it makes sure that the government has some equity in the less likely but possible event that the CHIPS Act becomes extraordinarily successful. Thus, for major funding recipients, the design of CHIPS subsidies also provides some form of insurance that sacrifices some profit in good times to exchange for more funding support in the initial difficult phases of the investments. However, a downside of the profit-sharing clauses is that they might prevent firms from voluntarily investing or allocating extra capacity to the awarded projects to prevent profits from exceeding the threshold that warrants profit-sharing. Periodic report requirements can partially ameliorate this issue through regular checkups to see whether firms adhere to their promises. Much of CHIPS funding will be disbursed in tranches contingent on the reaching of negotiated milestones relating to capital expenditures, workplace development, and operational status. This ensures that not too much funding will be allocated to projects that eventually fail.
Understanding that a vibrant semiconductor ecosystem involves not just manufacturing giants but also smaller firms across the supply chain, the second NOFO targets smaller materials and equipment suppliers whose investment will complement the big manufacturers’ operations. Total funding for this scheme amounts to $500 million, much less than the billions awarded to major manufacturers. Each small-scale facility project is expected to be subsidized by around ten percent of the projected capital expenditures cost. [16] Applicants are evaluated by similar requirements as that of commercial fabrication facilities, such as workforce development, economic and national security, community investment, and commercial viability. To promote the development of regional clusters, applicants are encouraged to apply as a part of a consortium of multiple suppliers, local government, and an anchor institution[17], like a semiconductor fab. Small-scale projects that are not part of the cluster have a higher burden of proof to prove that they can strengthen supply chain resilience or advance U.S. technology leadership. State and local governments are strongly encouraged to support such a cluster by implementing corresponding policies in workforce education, infrastructure investment, and the streamlining of regulations. Such an approach again points to the place-based nature of the CHIPS subsidies that do not reward any single set of champion firms. Instead, both the big fish and other smaller organisms in a regional semiconductor ecosystem are rewarded. Taiwanese suppliers interested in joining TSMC’s venture in Arizona must, therefore, anticipate making long-term commitments to Phoenix’s local community and economy to receive support from CHIPS funding.
US Multidimensional Strategy: Taiwan’s Response
For recipients of the CHIPS program, the announcement of the subsidies is just the beginning of a dynamic game with federal and state governments as well as local communities. Repeated renegotiations with respect to both new investments and existing agreements are expected, as there will be economic cycles and geopolitical shocks that change fundamentals that warrant renegotiations and adjustments of past agreements. The Federal government surely cannot be satisfied with the extent of investments that TSMC, Samsung, Intel, and Micron have committed, and will demand more fabs with more advanced technologies to be built. For Taiwanese firms such as TSMC and its suppliers committed to America for the long term, constant communication and coordination with federal agencies are key to making sure that subsidy allocation and the terms attached are in their best interest. The Taiwanese government should also participate in the coordination process by leveraging their access to Taiwanese firms that the US policymakers are eager to attract. The announcement pertaining to the guardrails provisions mentions the importance of maintaining coordination and close contacts with U.S. partners and allies, including Korea, Japan, India, and the UK. Taiwan was conspicuously left out of the discussion despite TSMC’s investment being perhaps the most high-profile and celebrated project funded by the CHIPS Act. [18]However, the Taiwanese government has an arsenal of policy tools that could determine the relative attractiveness of Taiwanese locations to that of the US for Taiwanese semiconductor firms. The government should leverage this fact to remind U.S. policymakers that communication and coordination with the Taiwanese government prevent excessive subsidy competition for TSMC and other star Taiwanese firms.
To navigate the complex US policy process motivated by different objectives, maintaining cordial relationships and trust with the US government is essential for Taiwanese firms. Taiwanese firms committed to investing in the US have to be prepared to cut ties with China. They will first have to refrain from future expansion in China and should consider phasing out Chinese production to mature technologies less susceptible to regulatory scrutiny. Taiwanese firms should also recognize US policymakers’ interest in regional community building and workforce development. To secure equal footing in future funding decisions, they should be prepared to make commitments that go against their desire to minimize costs. It is unlikely that the US government is interested in continuing to support Taiwanese fabs in America that employ mostly Taiwanese immigrant workers rather than locals, so figuring out how to efficiently train and employ American workers without sacrificing productivity is essential. Finally, to not repeat the Foxconn debacle in Wisconsin, Taiwanese firms should not back out of their commitments easily. The US government is interested in long-term investment in a local community resulting from joint efforts from multiple firms to form a viable industrial cluster. Mistakes made by the Trump Administration regarding Wisconsin will not be easily repeated, and Taiwanese firms must be wary of the financial and reputational consequences of backing out of their deal. Thus the stipulated project milestones must be negotiated carefully beforehand, supported by informed research of on-the-ground conditions. In future negotiations on new projects or renegotiations on existing deals, firms must invoke prospects that appeal to US policymakers, such as aligning the sustained profitability of the project to local economic wellbeing. The Taiwanese government can play an indirect or direct role in the process by controlling the costs of foreign investments with corresponding domestic policies such as tax abatements on suppliers undertaking significant investments abroad. These policy tools potentially could be great bargaining chips for Taiwanese firms. They could either increase the value of Taiwanese firms’ outside options by making investment in Taiwan more attractive, which warrants higher.
US subsidies to compensate or reduce the costs of Taiwanese foreign investment efforts.
Conclusion
Previously interconnected semiconductor supply chains are being split into two blocs. One bloc is that of China, with a burgeoning domestic supply chain serving mainly Chinese customers. The other is composed of US allies that remain interconnected, though each allied country is instituting significant efforts to promote domestic firms and building domestic supply chain capacity to avoid overreliance on other allies. In an age where semiconductor R&D and production decisions are driven increasingly by political logic rather than economic rationality, economic investment, and production costs should be weighed against the costs stemming from geopolitical uncertainty. Rising economic costs from cutting Chinese ties and investing in America should then be considered a price for security and supply chain risk diversification. The price could be tolerable provided that the US and allied governments provide sufficient incentives, but obtaining good deals with the US government requires a proper understanding of US objectives.
The simultaneous emphasis on workforce development, national security, supply chain stability/economic security, and commercial viability in the CHIPS NOFOs, encapsulated in the terms and conditions for funding, reflect the current administration’s view that all four objectives are mutually complementary rather than conflicting with each other. Under the belief that disadvantaged populations and communities could be adequately trained to serve in semiconductor manufacturing and that reasonably cost-efficient supply chain clusters could be set up given proper incentives, CHIPS funding is viewed as a necessary remedy to address various market failures that have hitherto constrained semiconductor manufacturing competitiveness. The substantial investment commitments of the CHIPS participants are anticipated to kick start a virtuous cycle of continued investment and commercial production that lowers production costs over time, contrary to suggestions that the CHIPS Act will be another failed industrial policy that does not respect differences in comparative advantage between the US and East Asia. [19]
Whether the grand economic strategy could succeed in the long run remains to be seen, and significant costs and construction/production delays will almost surely persist in the short run. Yet, now that most of the heavily anticipated first-round funding to major recipients like TSMC, Intel, and Samsung has been announced, some preliminary strategic objectives of the CHIPS Act have been achieved. The semiconductor giants’ acceptance of the CHIPS funding and the conditions attached entail acknowledgment of or even agreement to the Biden administration’s strategic priorities. This, along with the significant resources committed to projects in America, will steer the subsidy recipients towards an investment and business roadmap that will dynamically reduce the costs of continued investment in America while increasing the costs of re-engagement with China, solidifying the economic incentives supporting adherence to the American semiconductor strategy. Now that the Taiwanese government and firms have taken the path to embrace such a strategy, they must learn how to accommodate these US policy goals without sacrificing too much self-interest. Continued dialogue and careful identification of common grounds of mutual interest are then essential to maintain profitable relationships into the distant future.
[1] https://www.nist.gov/system/files/documents/2024/04/19/Amended%20CHIPS-Commercial%20Fabrication%20Facilities%20NOFO%20Amendment.pdf
[2] https://www.nist.gov/system/files/documents/2023/09/29/CHIPS%20-%20Facilities%20for%20Semiconductor%20Materials%20and%20Manufacturing%20Equipment%20NOFO.pdf
[3] “The interest rate on a CHIPS Loan will generally be based on the cost of funds to the Department of the Treasury for obligations of comparable maturity plus a portion of the spread to the market rate for commercial loans or debt of similar risk, tenor, and terms.” https://www.nist.gov/system/files/documents/2024/04/19/Amended%20CHIPS-Commercial%20Fabrication%20Facilities%20NOFO%20Amendment.pdf page 10
[4] 15 U.S.C. § 4652(6) (C) (i): “ the covered entity shall enter into an agreement with the Secretary specifying that, during the 10-year period beginning on the date of the award, subject to clause (ii), the covered entity may not engage in any significant transaction, as defined in the agreement, involving the material expansion of semiconductor manufacturing capacity in the People’s Republic of China or any other foreign country of concern.”
[5] 15 U.S.C. § 4652(6) (C) (ii):The prohibition in the agreement required under clause (i) shall not apply to—
(I)existing facilities or equipment of a covered entity for manufacturing legacy semiconductors; or
(II)significant transactions involving the material expansion of semiconductor manufacturing capacity that—
(aa)produces legacy semiconductors; and
(bb)predominately serves the market of a foreign country of concern.
[6] https://www.law.cornell.edu/definitions/uscode.php?width=840&height=800&iframe=true&def_id=15-USC-102889058-1966900778&term_occur=999&term_src=title:15:chapter:72A:section:4652
[7] https://www.nist.gov/system/files/documents/2024/04/19/Amended%20CHIPS-Commercial%20Fabrication%20Facilities%20NOFO%20Amendment.pdf Page 5 of 76
[8] https://www.nist.gov/news-events/news/2023/09/biden-harris-administration-announces-final-national-security-guardrails
[9] https://www.forbes.com/sites/tiriasresearch/2024/03/18/globalfoundries-trusted-foundry-status-helped-secure-chips-act-investment/?sh=30028abc4ce4
[10] https://www.commerce.gov/news/speeches/2024/02/remarks-commerce-secretary-gina-raimondo-globalfoundries-announcement https://gf.com/gf-press-release/globalfoundries-and-biden-harris-administration-announce-chips-and-science-act-funding-for-essential-chip-manufacturing/
[11] https://www.nist.gov/system/files/documents/2024/04/19/Amended%20CHIPS-Commercial%20Fabrication%20Facilities%20NOFO%20Amendment.pdf page 23,24
[12] 15 U.S.C. § 4652(a)(2)(B)(ii)(I)
[13] https://www.npr.org/2023/12/19/1219830288/biden-semiconductor-chips-workforce-jobs
[14] https://www.bloomberg.com/news/articles/2023-12-08/micron-mu-strikes-union-deal-for-15-billion-idaho-chips-plant
[15] https://news.bloomberglaw.com/daily-labor-report/micron-to-broker-labor-peace-agreement-under-6-1-billion-deal
[16] https://www.nist.gov/system/files/documents/2023/09/29/CHIPS%20-%20Facilities%20for%20Semiconductor%20Materials%20and%20Manufacturing%20Equipment%20NOFO.pdf
[17] https://www.nist.gov/news-events/news/2023/09/biden-harris-administration-announces-final-national-security-guardrails
[18] https://www.nist.gov/news-events/news/2023/09/biden-harris-administration-announces-final-national-security-guardrails
[19] https://www.project-syndicate.org/commentary/us-chips-act-pushing-tsmc-into-wasteful-unpromising-projects-by-chang-tai-hsieh-et-al-2024-02
Related Articles
Disinformation and Civil Defence: How Did Taiwan’s Civil Society Counter Foreign Information Manipulation?
作者：You-Hao Lai
2024 / 6 / 16
Potential Human Rights Violations in Risk Assessment of Artificial Intelligence Systems: Two Cases of Recidivism Prediction Tools
Author：Janet Zhenrong Gan 、 Wei-Lin Chen
2024 / 2 / 29