Fixing the U.S. Semiconductor Supply Chain

The persistent global shortage of semiconductors has prompted governments and chip manufacturers around the world to take actions to expand production and prevent such shortages from occurring again. But in formulating their plans, they should include measures to ensure that future supply chains for computer chips are resilient. A study that we conducted with DENSO, a global automotive component manufacturer whose headquarters are in Japan, highlighted how vulnerable the semiconductor supply chain is to disruptions. We found that a short disruption of a semiconductor fabrication facility, or “fab,” in Taiwan for 10 days, could cause a flurry of additional disruptions across the entire supply chain that would last almost a year.

The aim of our study was to understand the resiliency of the semiconductor supply chain and to identify strategies that companies and the federal government could employ to improve it. To that end, we built a model, a mathematical description of the supply chain, which allowed us to simulate a supply chain disruption and its ripple effects throughout the value chain.

The vast majority of semiconductor fabs are located in Taiwan, mainland China, and South Korea. The production lead time for a batch of chips — the time between when the fab begins production and completion — varies from 20 to 60 days, depending on the plant and the complexity of the integrated circuit. These chips are subsequently shipped to assembly and testing facilities, almost all of which are in Asia, to produce various components; these processes can take 30 to 40 days. Finally, components are used in a variety of parts, modules, and systems (e.g., in vehicles’ body control modules, electronic lighting modules, keyless entry systems, inverters, brakes, power steering, and infotainment systems), which are then assembled into finished products. This implies that a shortage beginning at a wafer fab will become transparent to the vehicle manufacturer only after a long delay, when it is notified by suppliers that they cannot ship parts.

Our model incorporates a variety of performance measures, some of which were developed in earlier research described in this HBR magazine feature and this HBR follow-up digital article. A central element is time to survive (TTS): the maximum duration that the supply chain can match supply with demand after a facility disruption. Another is time to recovery (TTR): the time it would take for a particular node (such as a supplier facility, a distribution center, or a transportation hub) to be restored to full functionality after a disruption. Our simulation revealed that TTS associated with a disruption at a fab is as short as a few days. We found that the time it would take the supply chain to recover, or return to normal operations, from a 10-day disruption of production at one fab for would be nearly 12 months and would create significant financial losses across all supply chain partners.

In our current research, we are complementing TTR and TTS with a new measure: TTR Inventory, or the time it takes the inventory levels of the entire supply chain to return to normal after the disruption is over. Our analysis suggests that after a 10-day disruption in one fab, it will take at least 300 days before chip inventory is normal.

To mitigate U.S. dependency on foreign chip production, Congress passed the CHIPS and Science Act, which President Joe Biden signed into law in August, and the Commerce Department unveiled its strategy for “rebuilding American leadership in semiconductor industry” in early September, which includes establishing and expanding domestic production of leading-edge chips, building a sufficient and stable supply of older-technology semiconductors, and investing in R&D to ensure next generation semiconductor technology is developed and produced in the United States. The act is expected to spur the construction of a half-dozen semiconductor manufacturing facilities in the United States. Such measures will increase and secure supplies of semiconductors, which are important not only for the economy but also for national security.

Given the lessons of the last three years and our research, policymakers and industry leaders should take these actions in creating and implementing their plans:

Build for Resiliency.

When designing supply chains for the newly constructed semiconductor facilities, focus on resiliency. Making a supply chain resilient entails a time-consuming process.

Reshoring alone does not necessarily create resiliency — a reality highlighted by the 2020 meat shortage that occurred in the United States during the Covid-19 pandemic. This industry’s supply chain is entirely domestic. To reduce costs, many companies had focused on consolidating manufacturing activities. The result: A relatively small number of slaughter plants process much of the beef and pork consumed in the United States. The pandemic revealed that shutting down one plant, even for a few weeks, has a major impact throughout the country; it crushes the prices paid to farmers and leads to months of meat shortages.

Keeping this in mind, U.S. policymakers should incentivize industry to stress test the new U.S. semiconductor supply chains and, if needed, to invest in creating dual sources of products.

Don’t just focus on advanced chips.

It is important that policymakers consider more than just cutting-edge process nodes at fabs. (A process node is an operation that manufactures a specific generation, or architecture, of semiconductor circuits). Much of the investment being discussed by policymakers is focused on advanced process technology nodes such as those that make chips with transistor elements that are 7 nanometers (nm) and below. But ensuring the supply of older-generation chips (45 to 65 nm) to industries such as automotive and aerospace as well as the military is critical. These technology nodes are responsible for the majority of the chips used by the automotive industry.

Significant risks to supply chains will continue if semiconductor chips produced in North America need to travel back to Asia for assembly and testing. With this in mind, the federal government should provide the industry with incentives to develop these capabilities in the United States.

Fashion a short-term response.

The construction of new semiconductor facilities will take a long time; the optimistic estimate is at least two years. Until that time, the federal government should ensure the appropriate supply of chips for key national security components. There are multiple ways to achieve this objective. They include:

• Building a chip reserve capacity, much like the strategic petroleum reserve
• Encouraging original equipment manufacturers (OEMs) and components suppliers to re-use chips from products that are obsolete
• Developing a national IT tool that captures and makes visible to users information on the amount and locations of inventories of available chips and obsolete products that include chips so supplies can be allocated efficiently to satisfy national security and commercial demands

Establish component standards.

Unlike the PC industry, there are limited common standards across many industries — in particular, the automotive industry — for electronic components, including communication, power, and memory chips. Increased standardization could improve the ability of both producers and consumers of chips to respond to a supply disruption.

Our recommendations could also be applied to supply chains for other products, such as the one for batteries and magnets used in electric vehicles (EVs). To protect supplies critical to the U.S. economy and security, adding domestic production and boosting R&D will not alone suffice. The new supply chains must also be designed and built to be resilient.