U.S. Plan to Save Semiconductors Misses the Mark, Defense Firms Say
Companies that make microelectronics for the Pentagon argue that the current bill could maintain U.S. defense dependency on China rather than fix it.
In the 2006 film The Departed, Massachusetts State Police take down a mob boss who is smuggling U.S.-made microprocessors to China. Those microprocessors, one of the cops says, “are the kind of processors they put into computers to put a cruise missile up the ass of a camel from a couple hundred miles away.” These “little pieces of plastic,” he adds emphatically, “are about a hundred grand apiece.”
Fourteen years later, the competition between U.S. and Chinese microelectronics manufacturing is hotter than ever as the rivals pursue 5G technology, hypersonic weapons, and quantum computing. For U.S. lawmakers, rebuilding a domestic supply chain for critical materials—whether semiconductors or rare-earth elements—has become a central part of the economic and technological rivalry with China.
To lure back some of the semiconductor and microelectronics manufacturing that in the last three decades has fled to Asia, especially China, lawmakers are turning to incentive grants. What began as several separate congressional bills ended as a section of the 2021 National Defense Authorization Act (NDAA) that would offer companies up to $3 billion to build new, U.S.-based semiconductor fabrication plants. Congress hopes to agree on a final version of the defense funding bill ahead of the Nov. 3 presidential election and vote on it soon after.
But there’s one big problem, say companies that build the specialized semiconductors used by the U.S. Defense Department in, for example, radiation-hardened chips on space satellites, anti-jam GPS receivers, and the F-35 fighter. The bill’s current language focuses too much on state-of-the-art commercial chips and not enough on the application-specific, small-order chips that the Pentagon needs to build directed energy weapons, they argue.
More than a dozen companies—concerned less with what could power your laptop and more on what might steer that cruise missile—are lobbying Congress to change the defense bill’s language to include “mission-critical and integration-level semiconductors” before the compromise language is finalized. This addition would put a spotlight on Pentagon-qualified semiconductor manufacturing and integration and help secure an increasingly fragile global supply chain. Without it, their collective employees’ livelihoods—about 50,000 jobs across the country—are at stake, those companies said.
But the incentive program, even though it’s tucked into a defense bill, is meant to be a broader fix for the decades-long disappearance of U.S. chip manufacturing. The current NDAA language is meant to improve the entire U.S. supply chain for semiconductor manufacturing, said a Hill staffer for a senator involved in the process.
“Our goal as legislators is to create a broad program that can help solve this problem, and we want to make sure that it’s not overly specific so we don’t limit our ability to innovate,” the staffer said. The section, multiple sources noted, was included in the NDAA not only because it has national security implications but because it’s pretty much the only bill Congress can pass with a broad majority anymore.
While members of Congress are taking the defense industries’ concerns into account as they finalize the NDAA, the Hill staffer said the inclusion of “mission-critical and integration-level semiconductors” risked limiting the flexibility of the proposed grant incentives.
“There’s nothing in the current language that would make them ineligible from the program or participating in the program. They just have to compete,” the staffer continued.
The grants, to be distributed via the Commerce Department, will have a finite amount of funding. Multiple senators had previously advocated to authorize $15 billion, per documents viewed by Foreign Policy, but the total amount is expected to be included in the compromise NDAA.
The language of the NDAA section is meant to rejuvenate the whole gamut of high-end U.S. semiconductor design and manufacture, noted James Lewis, the director of the Strategic Technologies Program at the Center for Strategic and International Studies, giving the United States the ability to make a play for the high ground of the new global economy. Adding more sector-specific language to the semiconductor section to assuage defense firms risks watering down that intent, he said, as more firms will be taking a slice out of a single pie.
“We want to spend money on keeping America as the technological leader in semiconductors,” Lewis said. “The Chinese cannot duplicate what they get from Qualcomm and Intel. They cannot make 5G without America.” Meanwhile, the United States could easily replicate what it gets from China, “but we got out of that business … because it’s not where you make the most money.”
For decades, lawmakers, industry, and experts have all watched with concern as the U.S. share of global semiconductor manufacturing has plummeted, while China’s share has risen. In 1990, the United States made 37 percent of the world’s semiconductors, according to a September report by the Semiconductor Industry Association and the Boston Consulting Group. It’s now at 12 percent and is projected to drop to 10 percent by 2030.
By 2030, 77 percent of computing chips will likely be produced in Asia, primarily in China and Taiwan, with 24 percent and 21 percent of the share, respectively, the report said. Industry experts noted that should tensions between China and Taiwan worsen, the U.S. semiconductor makers that rely on parts made by the Taiwan Semiconductor Manufacturing Co. (TSMC) and other foundries in the country would suffer greatly.
“Because 75% of the global capacity is already concentrated in East Asia, maintaining domestic manufacturing capabilities is essential to ensure the US semiconductor industry has a highly resilient, geographically diversified supply chain,” the report’s authors said, noting that this issue is particularly critical for computer chips used in U.S. defense systems.
Ellen Lord, the Pentagon’s chief acquisition czar, said at a think tank event in July that the country has become entangled in a “potentially compromised position, where we have U.S. intellectual property, in terms of designs, going offshore for fabrication and packaging.” This leaves the United States with vulnerabilities that are “unacceptable moving forward,” she added.
The commercial semiconductor industry has begun to respond. TSMC, which owns the world’s largest foundry, announced this year that it would invest $12 billion to build a new fabrication facility, or fab, in Arizona. Currently, most of its facilities are in Taiwan, with two in mainland China and one in Washington state. Intel has also expressed interest in working with the Defense Department to build a new U.S.-based foundry.
Ironically, U.S. defense needs are one driver of the exodus of chipmakers. Fewer companies are able to meet the stringent Defense Department requirements for design and manufacturing, said John Costello, the vice president of government affairs for Microchip Technology, which develops circuits for defense and aerospace systems. It makes better financial sense for some to exit the market and then adopt a “fab-less model and outsource a lot of stuff so [they] can go after that high-volume commercial application,” he said.
Costello, whose company is part of the group lobbying Congress to change the language, underscored the threat of offshore supply chains, even if they’re in a friendly place like Taiwan. If there were a disruption in Taiwan, Microchip Technology would need to find another supplier of silicon wafers for its microchips, he said, and would have to requalify its whole line of products.
Other chipmakers acknowledge that reshoring will mean higher costs—but also greater resilience. China has the benefit of “very, very cheap labor,” and manufacturing and integration should be reshored to the United States, said Terry Spitzer, the CEO of Global Technical Systems. His company provides cybersecurity and security products to the U.S. military and is also advocating for the language change.
“Will it cost a little more? It will cost a little more,” he said. “But in the long run, we’ll have those jobs in this country, we’ll have the technology … and we’ll have the secure supply lines.”
For Lewis, the whole fight over which firms and which technologies should benefit from multibillion-dollar government grants has a familiar ring.
“This is the usual unseemly battle over who gets to feed at the federal trough,” he said.
Those involved in the joint industry effort insist that’s not the case. “We take this issue very seriously, as our country’s competitive edge from global rivals, our national security, and more than 50,000 American jobs in at least 16 U.S. cities in nine different states are at stake,” the group said.