Three years ago, the Biden administration, backed by big bipartisan majorities in Congress, launched the CHIPS and Science Act to revitalize domestic semiconductor manufacturing. After decades of outsourcing that reduced the U.S. production share of advanced chips from 37 percent to 10 percent, this “industrial policy” investment was justified as necessary for national security and as a measure to reinvigorate manufacturing.
To date, Congress has appropriated over $37 billion in CHIPS grants and loans to 32 companies and 48 projects across the country. Despite the billions in spending, this legislation is only the beginning of an answer to our semiconductor challenges.
A secure semiconductor supply chain requires three components: manufacturing infrastructure, the technology to power it, and the workforce to operate it. Early on, both the government and the electronics industry recognized that building factories and installing machines was just part of the solution — a skilled workforce would also be essential. With persistent labor shortages, especially in manufacturing, this represented a major task. So, Congress included hundreds of millions for workforce development in the Department of Commerce fab grants and in separate initiatives at the National Science Foundation, Department of Defense and Department of Energy.
Congress prioritized workforce development because staffing fabs is as hard (or harder) than building them. The electronics industry lacks the talent pipelines to recruit and train workers. In a tight labor market, companies often end up recruiting from a pool of workers with limited skills or significant socioeconomic employment barriers. Training these workers falls mostly on employers, who risk seeing them hired away by competitors. This churn deters private investment in training needed to meet current and future demand.
Within the manufacturing sector, wages are another confounding factor. The median wage for electronics production workers is $24 per hour — considerably lower than machinists, welders and auto workers. Raising wages significantly could price U.S. products out of the global market or require public subsidies to make them competitive. For defense-related production, subsidies can be justified on national security grounds. In the consumer market, these types of subsidies would be prohibitive and, in the long run, politically untenable.
The lack of a viable workforce and a domestic cost structure that is uncompetitive has already complicated investment in domestic chip capacity.
In 2021, Taiwan Semiconductor Manufacturing Company announced a $12 billion fab in Phoenix. Though the company completed construction in July 2022, high-volume production was delayed until late 2024, mainly due to a lack of trained workers. Meanwhile, in Japan, a similar facility broke ground 18 months after the Arizona fab, yet went into full production at the same time. Access to trained labor made the difference. So long as better-paying jobs are plentiful, new semiconductor plants are likely to face the same problem finding skilled workers.
Even if workforce and wage challenges are overcome, another issue looms: U.S. manufacturing capacity has withered over several decades and would have to be rebuilt. As we’ve seen with President Trump’s tariffs on the auto industry and consumer electronics manufacturing chains — products passing back and forth across national boundaries during assembly, sometimes multiple times — are complex, fragile and hard to redesign.
Chips don’t work alone; they require substrates like printed circuit boards and other components that are also not produced domestically. A semiconductor without a printed circuit board is like a car engine without a frame — sophisticated and potentially powerful but effectively useless. Without a plan to use our new semiconductor resources, we risk filling warehouses with chips that have nowhere to go.
To make use of the tens of millions of newly available, domestically produced semiconductors, the U.S. faces a clear choice: either invest in the long, expensive and uncertain effort of rebuilding electronics manufacturing capacity, or rely on a form of distributed, global assembly — often called “friendshoring” — that combines American design know-how with less expensive labor.
In other words, end-to-end electronics production that could make use of domestically produced chips means rebuilding a version of the international manufacturing system that is currently undergoing massive disruption. If we end up with a “friendshored” assembly system, we will, to paraphrase T.S. Eliot, arrive where we started and understand globalization for the first time.
Brent Orrell is a senior fellow at the American Enterprise Institute, specializing in job training and workforce development. David Hernandez is the Vice President of Education at the global trade association IPC.
