Shubham Dwivedi Gregory D. Wischer
Semiconductors are critical for U.S. national security and a key input in everything from computers and smartphones to fighter jets and military communications. The most important of these are leading-node logic chips, which are necessary for all kinds of advanced technologies like artificial intelligence, quantum computing, and advanced wireless networks. While this makes leading-node logic chips a national security priority, the supply chains for these chips are vulnerable to market and geopolitical risks. For example, demand for emerging technology like self-driving vehicles will strain existing capacity for these advanced chips, and only Asian firms that depend heavily on Chinese customers—Taiwan-based TSMC and South Korea-based Samsung—fabricate them.
Given the national security implications of leading-node logic chips, the U.S. government should build resilient supply chains for these chips. Importantly, Washington must address existing vulnerabilities across the entire supply chain, which includes design, materials, equipment, fabrication and assembly, testing, and packaging. To that end, there is much that both Congress and the Biden administration can do to aid this endeavor.
What Congress Can Do
Grants. Congress should authorize and appropriate grants to address supply chain vulnerabilities for leading-node logic chips. These grants should fortify the entire supply chain, including the production of essential materials such as gases and chemicals.
With the capital-intensive nature of semiconductor supply chains—especially for leading-node logic chips—substantial government grants are necessary. For instance, one advanced logic fab costs $30-40 billion, and Boston Consulting Group estimates indicate that building five to six new U.S. leading-node fabs that produce 35,000 wafers per month by 2030 will only cover 25 percent of U.S. consumption.
The Semiconductor Industry Association and Boston Consulting Group have proposed a federal government incentive program totaling $20-$50 billion, which would make the cost-structure of constructing two to three new state-of-the-art fabs in the United States competitive with Asia. They estimate that building these fabs by 2030 will cover U.S. consumption for advanced logic chips in critical infrastructure applications. Over a period of ten years, these new U.S. fabs would require $40-$50 billion in private sector investment and $15–$20 billion in government incentives. Eventually, over this time period, government incentives may comprise 30-40 percent of total fabrication costs.
Given supply chain synergies in their current regions of operation—Asia—semiconductor fabrication firms without U.S. operations will not be able to significantly expand their U.S. fabrication capacity. The cold truth is that, absent a domestic manufacturing footprint, the U.S. semiconductor industry may also lose its technological design edge, which is critical for advances in chip technology. However, government incentives can concomitantly bolster U.S. semiconductor manufacturing and design. Moreover, these government incentives will actually be effective due to America’s pre-existing manufacturing infrastructure and chip design industry.
In short, the United States has a robust infrastructure and skilled manufacturing workforce, but it simply lacks sufficient incentives for substantial capital expenditures in leading-node logic supply chains.
Tax Incentives. Congress should offer tax incentives for investment in supply links for leading-node logic chips. Primarily, this means lower corporate taxes in the semiconductor industry, targeting not only design and fabrication but also material and equipment.
Without sufficient high-quality materials and equipment, fabrication of leading-node logic chips is impossible. Targeting supply links upstream from fabrication will also likely strengthen downstream links by encouraging geographic clustering of related supply links, making the entire supply chain more resilient.
To that end, Congress should establish a federal investment tax credit (ITC). An ITC would “offer ongoing, predictable incentives” for supply chains of leading-node logic chips. A report from SEMI, a leading semiconductor association, notes that a well-planned ITC “would be the single most impactful federal policy to help build new US semiconductor manufacturing capacity.” This measure should target the entire supply chain, from design to assembly, testing, and packaging. This includes, and should especially emphasize, targeting materials. An ITC would also likely attract investments from foreign semiconductor firms like TSMC and Samsung.
Additional tax incentives should also target research. As it stands, U.S. tax incentives for research are low compared to other countries. For example, the United States has a 9.5 percent subsidy and 14-20 percent tax credit for research, while China has a 29.1 percent subsidy and 175 percent super-deduction rate for research. The United States also lags China in research growth: from 2003 to 2017, U.S. government research increased 2 percent from $121 billion to $124 billion, while Chinese government research increased 330 percent from $23 billion to $98 billion.
What the Biden Administration Can Do
Allocation Criteria. The Biden administration should issue criteria that prioritizes funding for vulnerabilities in supply chains of leading-node logic chips.
Congressional legislation generally only provides high-level guidance on funding programs, often failing to go into detail. For example, the CHIPS Act has an overriding concern for fabrication while neglecting the critical materials necessary for fabrication in the first place. In its own supply chain report, while the administration does recommend funding for “domestic manufacturing of leading edge semiconductors,” it should also target domestic materials production. Likewise, the CHIPS Act aims to invest in semiconductors in a broad manner rather than focusing on leading-node logic chips, which are of greatest importance. Will Hunt, a research analyst at Georgetown University’s Center for Security and Emerging Technology, recommends allocating at least $23 billion of the $37 billion manufacturing incentives in the CHIPS Act for leading-node logic chips.
The administration should also issue a report or more on supply chain vulnerabilities for leading-node logic chips, which will assist Congress in crafting semiconductor legislation. Given that Congressional staff and committees are limited by both time and scope constraints, the White House should provide in-depth research analyzing supply chain vulnerabilities and estimating requisite funds for mitigating these vulnerabilities. While the administration’s 100-day supply chain review provides decent background information on supply chain issues, future reports should detail specific supply chain vulnerabilities, projects that will alleviate such vulnerabilities, and related program costs for addressing each vulnerability. These types of reports would help congressional staff and lawmakers better understand supply chain vulnerabilities and costs.
Streamlined Regulations. The Biden administration should streamline health, safety, and environmental regulations impacting supply chains for leading-node logic chips.
Onerous domestic regulations affecting wages, work conditions, and environmental protections increase costs, de-incentivizing domestic investment and encouraging overseas investment. The Semiconductor Industry Association notes that “[d]elays in regulatory decisions and restrictions on new chemicals can make these new technologies unavailable in US and drive R&D and use of these new technologies offshore.” Additionally, many countries have more streamlined regulations in place compared to the United States, and foreign governments are often more resilient to public pressure concerning burdensome regulations. A stable regulatory environment is a strong investment incentive for large-scale capital expenditures, including the fabrication of leading-node logic chips.
To that end, the administration should permit national security exemptions and regulatory expeditions for certain health, safety, and environmental regulations. For example, the U.S. Environmental Protection Agency (EPA) issued a rule banning Phenol, Isopropylated Phosphate (PIP) from U.S.-sold products, which disrupts semiconductor supply chains since the industry uses PIP in equipment wiring. The EPA has also restricted the use of new chemical compositions that buttress semiconductor manufacturing innovation and production. By invoking national security, the White House can streamline regulations and expedite permitting processes for key semiconductor projects. National security justifications may also strengthen the government’s defense of streamlined regulations against legal challenges, especially from overly activist environmental groups.
Additionally, the administration should coordinate with state authorities to expedite permitting for leading-node logic supply chain projects. State and local governments have different regulations from the federal government, and companies face significant costs ensuring compliance with all pertinent regulations. Semiconductor companies may be more inclined to invest in long-term projects if the White House were to develop harmonized plans with state governments—particularly those states with attractive regulatory regimes—for obeying regulations and receiving approval for various projects, from fabs to air purification facilities. Streamlined regulations also reduce the costs and time spent ensuring regulatory compliance.
Potential Counterarguments
There will be individuals who will argue that the aforementioned policy recommendations will be ineffective. More specifically, they will argue that industrial policy—government efforts to support particular industries—is an ineffective tool.
Gary Clyde Hufbauer and Euijin Jung at the Peterson Institute for International Economics, for instance, contend that “the track record of such efforts has achieved mixed success at best.” Yet this line of argumentation tends to focus on the American context over a short and recent timeframe—coinciding with the popularization of free-market economics—while ignoring foreign developments. Consider that industrial policy has proved critical in building strong semiconductor industries in Taiwan and South Korea. These governments successfully incentivized semiconductor fabrication through grants and chip purchases, and continue to provide financial support to their respective national champions: TSMC and Samsung. Other foreign governments, including China, are also offering incentives to grow domestic semiconductor capacity and gain global market share.
Another argument will be that the above policy recommendations are too expensive.
Semiconductor industry analysts estimate that the upfront cost of self-sufficient U.S. semiconductor supply chains is in the range of $350–$420 billion. These analysts posit that “complete autarchy or full semiconductor self-sufficiency appears to be more a theoretical concept rather than an attainable policy goal.” However, our recommendations do not advocate for “complete autarchy or full semiconductor self-sufficiency”—we seek self-sufficiency in a subsegment of semiconductor supply chains: leading-node logic chips. Furthermore, China is seeking to establish self-sufficient semiconductor supply chains, investing over $150 billion in manufacturing subsidies so far. In order to adequately compete against this, the United States, which has a larger economy than China, should increase its government incentives for semiconductors. Failure to do amounts to surrendering this essential sector to foreign competitors.
Finally, some may argue that the above policy recommendations will be wasteful—that rapidly growing supplies of leading-node logic chips via government incentives as demand increases gradually may inadvertently harm semiconductor firms.
For example, when demand for GPUs was high in 2018, Nvidia increased supplies. However, demand dropped, harming Nvidia financially. Scott Lincicome, director of economics and trade at the Cato Institute, notes that industrial policy has cost overruns, including deadweight losses and unintended consequences. But any government program can have and often has these costs—did cost overruns prevent the creation of the F-35 stealth fighter? Furthermore, the United States has the appropriate domestic semiconductor environment—established firms, robust infrastructure, and skilled workers—to ensure semiconductor incentives are well-used.
A Twenty-First Century Industrial Strategy
Supply chain vulnerabilities for leading-node logic chips endanger U.S. national security. The plain truth is that these chips are critical in emerging technology affecting key sectors. Simply put, America needs a secure supply of these chips, lest we be relegated to second-tier technological power status. Congress should pass laws that offer grants and tax incentives for supply links of leading-node logic chips, and the White House should prioritize funding for supply links and streamline federal regulations that affect semiconductor supply chains. Though some may argue that these recommendations are ineffective, expensive, and wasteful, foreign governments like Taiwan and South Korea have proven that industrial policy can effectively bolster the quality and capacity of supply chains for leading-node logic chips.
In May 2022, U.S. secretary of state Antony Blinken highlighted America’s need for “a modern industrial strategy to sustain and expand our economic and technological influence, make our economy and supply chains more resilient, sharpen our competitive edge.” He urged Congress to “make historic investments in research and innovation, including in fields like artificial intelligence, biotechnology, quantum computing. These are areas where Beijing is determined to lead.” But to make America the global leader in emerging technologies, the U.S. government—through bills like the CHIPS Act—needs to prioritize funding supply chains for leading-node logic chips specifically, not semiconductors broadly.
No comments:
Post a Comment