Last week, assembly lines at Ford and General Motors ground to a halt—all for lack of a few pieces of silicon smaller than your pupil. The pandemic and the move to remote work has put unprecedented pressure on semiconductor supplies. Like the two carmakers, companies everywhere are scrambling to purchase them or simply shutting down.
Advocates of “industrial policy,” or government support for specific industries, have seized on the shortage to argue for more federal money. The CHIPS for America Act and the American Foundries Act would spend tens of billions of dollars to encourage semiconductor manufacturing in the United States. Both bills have found bipartisan support in Congress.
Yet America has already tried semiconductor industrial policy in the past, with poor results, and the case for it today is even weaker for semiconductors than it is for other industries. The last thing we should do is turn a short-term supply shortage into an endless government dependency.
Back in 1979, the United States was concerned that it was losing its semiconductor edge to Japan, which had an industrial policy favoring the industry. In response, the federal government set up and funded the Very High-Speed Integrated Circuits Program (VHSIC). Over a decade, it produced nothing. In 1987, the Department of Defense tried again, helping to form SEMATECH, an Austin, Texas-based nonprofit with $500 million in both public and private funds. Though occasionally celebrated as a model, SEMATECH mainly funded research into memory chips which did not represent the future of the industry. A recent dissertation demonstrated that the U.S. military received “no tangible benefits” from the program, and that claims that SEMATECH rebirthed the industry “remain dubious.” The most likely result of the government subsidies was to allow the industry to spend less on research.
In the ensuing years, our onetime Japanese competitors’ share of the global semiconductor market dropped from about half to less than 15 percent. Those who had touted Japan’s silicon policy as a model lost their strongest case. South Korean and Taiwanese companies, led mainly by Samsung and TSMC, captured much of Japan’s global market share. Despite sporadic subsidies, the South Korean government actually disfavored semiconductors relative to old-fashioned industries like steel and chemicals. Taiwan’s success came not from shutting the world out but from inviting more foreign investment into the sector, especially from the United States. Meantime, U.S. companies’ share of the market remained stable at around 50 percent.
Industrial advocates argue that, even though U.S. semiconductor firms continue to dominate, they produce most of their semiconductors abroad—especially in China. Yet China is home to only 15 percent of global semiconductor production and almost no leading companies in the industry. China, in fact, is the world’s largest importer of semiconductors, largely from South Korea and Taiwan; indeed, it recently spent more on importing semiconductors than it did on crude oil. By contrast, semiconductors are America’s fifth-largest export, and the U.S. still maintains a trade surplus in them.
Industrial-policy advocates claim that more domestic manufacturing is necessary to keep our semiconductors at the cutting edge. Yet many of our most innovative semiconductor companies, such as Nvidia and AMD, are “fabless”—that is, they don’t own any “fabs,” or chip-manufacturing plants, instead contracting the manufacturing work to other companies and countries. This has not been a burden for them. Nvidia and AMD stock prices have both gone up more than 1,500 percent in the past five years. And contrary to the subsidy advocates, these companies have not thrived by skimping on research. The U.S. semiconductor industry devotes a greater proportion of its revenue to research than any country.
The federal government also just dropped a lawsuit against the fabless semiconductor company Qualcomm, after years of accusations that it had monopolized the mobile-phone chip market. It would be exceedingly odd for the federal government to start subsidizing and directing any of these wildly successful companies, especially ones that Washington has accused recently of being too powerful.
Finally, many industrial advocates claim that we need to reshore our manufacturing for strategic and military reasons—above all, to offset China’s growing power. But the U.S. also needs its long-time Pacific Rim partners to counter China. Any attempt to boost our semiconductors at the expense of the biggest players—South Korea, Taiwan, and Japan—would only serve to drive them into Beijing’s arms. A similar pattern played out when America left the Trans-Pacific Partnership, a free-trade pact meant to counter China, only to have most of the Pacific Rim sign an agreement with China itself in 2020. Such isolation makes neither economic nor strategic sense.
Many economists hoped that the decline of Japanese dominance in semiconductors and in other industries would convince politicians of the folly of state-directed economies. Governments understand neither what they should invest in nor how they should invest in it. If American companies are concerned about their supply chains, by contrast, they have the incentive and funds to move them. Why our federal government wants to take one of America’s most successful industries and turn it into a ward of the state is a mystery.
Photo: MACRO PHOTO/iStock