The automotive industry is perhaps most widely known to have been affected by the chip shortage. But it isn’t just automakers and other end customers who care about chip shortages, the entire supply chain cares too. Most supply chains are designed to be consolidated and cost-effective, but they can be brittle as a result. Limited visibility and lack of real-time communication between supplier tiers can lead to a “bullwhip effect” where small shifts in demand are amplified, resulting in high cumulative demand volatility.9
Chipmakers are scrambling to catch up. The world’s three largest semiconductor manufacturers announced cumulative annual capital expenditures of more than US$60 billion for 2021 and will likely spend even more in 2022.10 Some of that is increasing capacity at existing fabs, but some is construction of new facilities, such as Intel’s two new fabs in Arizona for US$20 billion-plus.11 In addition, aggregate venture capital investment in startup chip companies will have more than tripled in 2021 and 2022 compared to the annual average of the previous 15 years. Even though they are mostly focused on designing chips rather than manufacturing them, these companies will all want to make chips to use up still-tight capacity.12
To guard against future shortages, governments are pushing to increase local supply. As of 2020, 81% of semiconductor contract manufacturing was based in Taiwan or South Korea.13 The United States,14 the European Union,15 and China16 have all committed to growing their country or region’s semi fabricating capacity, a process called localization. Localization is not just about avoiding shortages, but also about enhancing national security: The proposed US$52 billion CHIPS for America Act was a part of the National Defense Authorization.17
These localization initiatives are an effort to reduce the risk created by the chipmaking industry’s historic concentration of manufacturing in a very few geographic areas: Silicon Valley in the past, and Taiwan and South Korea more recently. This clustering improved efficiency, turnaround times, and profitability in good times—but as we have seen, it also amplifies risk. If, as seems likely, multiple countries decide to mitigate that risk by building their own manufacturing capacity, the overall industry capacity utilization rate may trend lower compared to the last decades, though it will likely remain volatile. In the long run, this would likely mean fewer shortages at the cost of some efficiency.
Localization efforts, however, will take time. Increasing chip manufacturing capacity is a slow process, and rightly so: Cutting-edge chips have been called the most complex devices ever made, and it takes the most expert chipmakers in the world billions of dollars, years, and all of their expertise to get a new plant up and running.18
Complicating things further are shortages that casual observers may not know are key parts of chipmaking. One is a shortage of packaging substrates—the miniature interface layers in packaged chips. This shortage has constrained chip manufacturing for some time now, with lead times of one year or more.19 Additionally, to make chips, manufacturers need not just buildings and wafers, but equipment such as photolithography tools and wirebonders, which respectively print nanometer-scale patterns on semi wafers and add thin wire interconnects to chip packages. Equipment of both kinds, new and even used, is in short supply. Photolithography lead times are more than 10 months, and lead times for wirebonders, which are normally abundant, stand at over six months.20