Engineers at American innovation hubs such as Bell Labs, IBM, Texas Instruments, and Fairchild Semiconductor led the development of the transistor and the integrated circuit in the middle of the last century. Robert Noyce, credited with first deploying such a circuit on a silicon chip, went on to co-found Intel with Gordon Moore, whose famous “Moore’s Law” predicted the doubling of transistors on a chip every two years. The United States dominated the market for semiconductors, particularly at the high end, with Intel controlling 90 percent of the market for PC microprocessors in 1997. “Intel sets release dates for new chips,” wrote Time magazine that year, “dictating the pace of the computer industry with the confident aplomb of fashion designers raising or lowering hemlines.”

While semiconductors have become ever more important to modern civilization, providing the brains for not only computers but also phones, cars, and missiles, Intel no longer sets the standard. That distinction belongs to the Taiwan Semiconductor Manufacturing Company (TSMC), which manufactures chips with transistors just five nanometers (nm) wide. Intel remains mired two generations behind, at 10nm, and won’t achieve 7nm until 2023 at the earliest — by which time TSMC will likely be at 3nm. Globally, the U.S. share of semiconductor manufacturing has fallen by two-thirds since 1990, to 13 percent, and is still falling.

The decline of America’s semiconductor industry is in many ways the story of the nation’s major economic missteps in recent decades, as economists misunderstood the dynamics of globalization, policy-makers ignored the importance of technological leadership, and business leaders chose shareholder payouts over investment. Michael Boskin, chairman of George H. W. Bush’s Council of Economic Advisers, famously quipped, “Potato chips, computer chips, what’s the difference?” Quite a lot, it turns out.

Fortunately, the industry also illustrates the potential for, and pathway toward, redemption. It faced and overcame a similar challenge in the 1980s, thanks to assertive Reagan-administration trade policy, robust private-sector investment, and an innovative public–private partnership. Last year, a good bit of Trump-administration arm-twisting led TSMC to announce a $12 billion investment in a new Arizona foundry. The National Defense Authorization Act for 2020 established a public-policy framework to provide the industry with support.

This year, Intel has rejected pressure from an activist hedge fund to consider the abandonment of manufacturing altogether and instead committed $20 billion to strengthening its position. Amid renewed Chinese rumblings in the Taiwan Strait and a global chip shortage, the bipartisan National Security Commission on Artificial Intelligence called in March for a major public investment in “rebuilding domestic chip manufacturing” and “a strategy to stay at least two generations ahead of China.” The topic is ripe for bipartisan cooperation, and success could lay the groundwork for other efforts to ensure that America regains its role as global leader in industry and innovation.

The semiconductor industry includes “fabless” companies, which design chips, creating blueprints for the vast arrays of transistors whose embedded logic makes the digital world work; “foundries,” which operate the massive fabrication plants etching those blueprints into silicon; and integrated device manufacturers, which do both. Under the logic of comparative advantage, often cited by economists to explain how all countries can benefit from international trade by specializing where they are relatively better suited, one might expect America to be a global leader in both highly sophisticated parts of the supply chain. Yes, it might make sense for lower-cost Chinese labor to make T-shirts for Americans, but presumably it would be Americans using state-of-the-art technology to put microscopic transistors on chips for the Chinese. Thus the “trade” in “international trade.”

This is not how things have worked out. Beginning in the 1990s, South Korea, Taiwan, and China embarked on aggressive programs to create their own comparative advantage and win global leadership in the manufacturing process, offering generous subsidies for foundries to set up shop. Today, an advanced foundry costs 30 to 60 percent more to build in the United States than in Taiwan, Korea, or China, but not because of cheaper labor or better technology. In each case, the vast majority of that difference is explained by government support. China, pursuing its stated policy of becoming a global leader in all segments of the industry, provides land grants, cash grants, tax credits, below-market loans, and direct equity investments — with larger subsidies available to firms that agree to share their technology with local partners.

American firms have clamored to accept the offer. Domestic U.S. manufacturing capacity grew only half as fast as global capacity from 1990 to 2010, causing market share to fall from 37 percent to 13 percent. What began as losses in less sophisticated segments of the market ultimately became failures at the leading edge, as Asian firms gained experience and expertise. Intel competitor Advanced Micro Devices (AMD) abandoned its foundries in 2008 and now relies on TSMC for all its advanced chips. Though Apple used Intel chips in its laptops, it turned to TSMC and Samsung for the iPhone. As Intel struggled to move from 14nm to 10nm chips in the late 2010s, TSMC passed it and then lapped it again, marking the first time in five decades that Intel was beaten to market. In 2017, Samsung passed Intel to become the world’s largest chipmaker. TSMC’s stock price has increased fivefold in the past five years, and its market capitalization now doubles Intel’s.

Although Intel has excelled in shareholder payouts, TSMC passed it for the first time in capital expenditures in 2015 and has outspent it since. Its technological prowess slipping, Intel continued to step up its share buybacks and dividends each year, returning more than $70 billion to shareholders over the past five years — twice TSMC’s level. In 2018, following the abrupt resignation of CEO Brian Krzanich, Intel turned to its CFO to fill the top spot and expanded its buybacks further.

Thanks to their leadership in design, American firms still account for nearly half of the chip industry’s sales. But this is an industry in which manufacturing matters. Whereas the sneaker industry has only 4 percent of its value-add tied to manufacturing, in semiconductors the figure is 45 percent. With little manufacturing, the American share of the industry’s total value-add has fallen below 25 percent, while China’s share now exceeds 30 percent.

Beyond missed opportunity, the loss of manufacturing leadership is likely over time to erase design leadership, too. The engineering challenges in chip production are closely intertwined with the design process, and where the former goes, the latter will follow. Andy Grove, the longtime Intel CEO who succeeded Noyce and Moore, warned of this progression in 2010:

Each company, ruggedly individualistic, does its best to expand efficiently and improve its own profitability. However, our pursuit of our individual businesses, which often involves transferring manufacturing and a great deal of engineering out of the country, has hindered our ability to bring innovations to scale at home. Without scaling, we don’t just lose jobs – we lose our hold on new technologies. Losing the ability to scale will ultimately damage our capacity to innovate.

Using the battery industry as an illustration, Grove observed: “The U.S. lost its lead in batteries 30 years ago when it stopped making consumer electronics devices. Whoever made batteries then gained the exposure and relationships needed to learn to supply batteries for the more demanding laptop PC market, and after that, for the even more demanding automobile market. . . . I doubt [U.S. firms] will ever catch up.” He was correct. China has seven of the ten largest makers of electric-vehicle batteries, which account for 80 percent of the market; the U.S. has none.

A similar trend appears to be under way in semiconductors, whose creators tend to develop in clusters. Silicon Valley got its name for a reason. Fairchild Semiconductor alumni in California created both Intel and AMD, whose alumni later created Nvidia. As semiconductor manufacturing has shifted to East Asia, the center of gravity for innovation has gone with it. From 1998 to 2000, American firms received 86 percent of the industry’s venture-capital investment. From 2018 to 2020, that figure was 32 percent, compared with 37 percent for China.

The domestic industry’s failures are not only economically costly but also downright dangerous in national-security terms. “Semiconductors are the new oil,” as a Fortune headline put it in March, and America has once again allowed itself to become dependent on a vulnerable foreign source. U.S. military capabilities in areas such as cybersecurity, AI, autonomous systems, advanced imaging, and secure communications rely on the most advanced semiconductor technologies. The Pentagon works closely with civilian firms to design chips that are then manufactured abroad because of inadequate domestic capacity. For instance, the F-35 uses chips from American designer Xilinx that are made by TSMC in Taiwan.

In 2018, Congress created the National Security Commission on Artificial Intelligence “to consider the methods and means necessary to advance the development of artificial intelligence, machine learning, and associated technologies to comprehensively address the national security and defense needs of the United States.” The commission’s final report, issued in March, put the matter succinctly: “After decades leading the microelectronics industry, the United States is now almost entirely reliant on foreign sources for production of the cutting-edge semiconductors that power all the AI algorithms critical for defense systems and everything else. Put simply: the U.S. supply chain for advanced chips is at risk without concerted government action.”

Opponents of government action argue that the semiconductor industry’s prospects are rosier. Jessica Anderson, the executive director of Heritage Action, told the Wall Street Journal in March that Congress should not give federal support to a “profitable” industry. In a recent report published by the Cato Institute, Scott Lincicome cited statistics on the industry’s total size (large), the number of people it employs (many), the amount it spends on research and development (a lot), and the share of U.S. companies’ production that still occurs domestically (less than half). He acknowledged that U.S. capabilities could no longer be considered leading-edge but minimized the extent of Intel’s failures by assuring readers that it remains “financially healthy” and “is looking at free cash flow of $17.5 billion” this year.

In emphasizing large numbers and plentiful cash, Lincicome avoids grappling with the downward trends and loss of leadership that should be the focus for policy-makers. For instance, he provided a reassuring quote from the industry’s annual report: “The semiconductor manufacturing base in the United States remains on solid footing.” But that’s extracted from a more troubling sentence, which notes that, “while the semiconductor manufacturing base in the United States remains on solid footing, capacity growth around the world has outpaced that in the United States, which in turn has steadily eroded the share of global manufacturing capacity at home.” The report continues, “In 2019, all six of the new semiconductor fabs that opened globally were outside the United States, with four being built in China, which is putting significant government dollars behind these new facilities.”

American firms have made clear that they will gladly abandon their own manufacturing capabilities for the heavily subsidized and now technologically superior capacity available abroad. If regaining leadership in production is a national priority, then public policy will have to play a role in advancing it. This might take two forms: first, intervention to level a playing field distorted by foreign efforts to claim leadership, and second, collaboration to jump-start innovation and overcome technical challenges.

The situation in many ways mirrors that of the late 1980s, when a U.S. semiconductor industry reeling from intense Japanese competition saw its share of global sales plunge from more than 55 percent to less than 40 percent. The Reagan administration, having pursued an agreement with Japan to halt dumping into the American market, then imposed stiff, punitive tariffs on Japanese firms when the practice continued. The federal government also partnered with top U.S. firms on the creation of Sematech, a consortium that pursued research-and-development projects whose output could be shared industry-wide. The private sector and the government each footed half of the annual $200 million budget, and firms provided top talent for six- to 30-month stints. Robert Noyce, long since retired from his role as Intel’s first CEO, was drafted to head the effort.

“With a few vocal exceptions,” according to the National Academy of Sciences, “Sematech is widely credited within the U.S. industry with some role in stimulating a resurgence among U.S. semiconductor producers in the 1990s.” New generations of chip miniaturization were achieved more quickly and at lower cost. Perhaps the ultimate endorsement came in the mid 1990s, when federal funding ended but firms saw sufficient value in the effort to step up their own contributions and continue work. American market share stopped falling and then climbed back to near its early-1980s high.

The challenge this time around is greater. America has fallen further behind, and foreign governments are supporting their producers more aggressively. “An important factor in whether Samsung moves forward” with a $17 billion U.S. foundry, as reported in the Wall Street Journal, “will be the availability of U.S. federal government incentives to offset those offered by foreign countries.” Taiwan and South Korea hold the edge in sophistication, but on the horizon sits China, already No. 1 in both total industry value-add and venture-capital investment, and champing at the bit to claim technological leadership.

While many consider a world of free markets and small government ideal, globalization has left American policy-makers to face the rather different attitudes of other countries, and thus to face a choice: use public policy to support domestic manufacturing, or lose it for good. They look poised to choose the former. Last year, the Senate voted 96–4 to incorporate an aggressive investment framework into the National Defense Authorization Act, including production incentives, research-and-development funding, and creation of the National Semiconductor Technology Center, which could play the role of a next-generation Sematech. The Biden administration’s $2 trillion infrastructure package includes $50 billion of funding for such initiatives, a proposal that hopefully will garner bipartisan support as well.

Perhaps the greatest cause for optimism is something that didn’t happen. In December, activist hedge-fund manager Dan Loeb disclosed that he had taken a $1 billion stake in Intel and asked the company to consider divesting its manufacturing business — something CFO-turned-CEO Bob Swann had indicated he was considering. The move would have freed Intel’s design business to use Asian foundries, perhaps boosting its short-term prospects and generating a tidy profit for Loeb. But it would have devastated America’s last chance for semiconductor-manufacturing leadership. Instead, Intel replaced Swann in February with a celebrated engineer, Pat Gelsinger, who announced plans in March for a $20 billion investment to build Intel’s foundry business into a direct TSMC competitor that manufactures for fabless companies as well as executing Intel’s own designs. The next week, TSMC announced plans for $100 billion in new investment of its own. Game on.

–Mr. Cass is the executive director of American Compass, where Mr. Oyeniran is an analyst.