Who’s Going to Pay for American-Made Semiconductors?

Soon, we’ll see rising costs from domestic semiconductor fabs in the price tags on consumer goods.

Written by Nate Farshchi
Published on Mar. 23, 2022
Who’s Going to Pay for American-Made Semiconductors?
Brand Studio Logo

With billions committed to the construction of new U.S. fabrication plants to manufacture semiconductors, it may appear on the surface that the chip shortage problem will soon be solved. But the reality is not so straightforward. 

There will continue to be ripple effects on chip customers and end users thanks to high demand for technology, labor costs and pressures around environmental, social and governance issues associated with the foundry expansions in the United States.

Beginning in 2020, the chip shortage put the largely behind-the-scenes industry of semiconductor fabrication on center stage. These small microchips suddenly became the largest production choke point for industries and sectors throughout the economy. Then came 2021, when notably immense capital expenditure announcements signaled the return of semiconductor fabrication to the United States. Incentives at the state and federal level — the largest of which has been the $52 billion in federal funding and incentives set aside through the CHIPS for America Act — have cleared a path forward for addressing semiconductor supply issues.

Rising Domestic Semiconductor Costs Still Face a Big Challenge

Federal incentives alone won’t be able to completely offset the incremental cost of bringing production back to the United States — nor can they make up for opportunities that companies may have already lost because of chip challenges. 

 

A Surge in Semiconductor Investments

A number of companies have put forth investments for semiconductor manufacturing facilities set to come online in the next few years. In January, Intel announced plans to spend $20 billion to construct two state-of-the-art foundries in central Ohio, a commitment that came as a mirror to the $20 billion announced in early 2020 for the construction of two foundries in Arizona. 

But building a foundry is a far cry from the apartment complex you watched rise over a summer, like a time lapse, during your pre-pandemic commute home from the office. “The factories that build microprocessors cost between $10-$20 billion and can take three to five years to build,” as this ExtremeTech article explains. 

Meanwhile, the Semiconductor Industry Association notes that “the U.S. share of global semiconductor manufacturing capacity has eroded from 37 percent in 1990 to 12 percent” in late 2021. To the delight of businesses and politicians alike, both Taiwan Semiconductor Manufacturing Co. and Samsung Electronics announced three-year global capital expenditure plans in 2021 of approximately $100 billion each, both of which include commitments to expand fabrication in the United States. Samsung Electronics also plans to build a $17 billion semiconductor factory just outside of Austin, Texas. In an optimistic estimation, the capital expenditures announced in 2021 would likely begin producing chips in late 2023 or early 2024. 

But such investments do not exist in a vacuum. Plans for these fabs come at a time when companies and consumers alike are increasingly prioritizing ESG as the nation grapples with unique labor market challenges — all of which will factor into the process of developing these manufacturing facilities and getting them online. 

Read More RSM US Analysis on SemiconductorsWhat the U.S. Is Doing About Semiconductor Shortages — and How It May Affect Your Operations

 

Environmental and Talent Challenges

Investors, customers, employees and the government have all made it clear that ESG issues are paramount for businesses across the economy. From an environmental perspective, foundries require massive amounts of water and power to operate. Taiwan Superconductor Manufacturing Company (TSMC) foundries in Phoenix need around 4.7 million gallons of water per day to support production, TSMC technical director Tony Chen told CNBC. (Chen has overseen 17 fabrication construction projects in his 23 years with TSMC.) 

According to Taiwan News, “Factories used 6 percent of [Arizona’s] water in 2019, but with TSMC and Intel both planning new facilities, the fear is that number will grow.” The good news for environmentalists and the people of Arizona is that TSMC plans to build an on-site water treatment plant capable of recycling up to 90 percent of the water used in the fabrication process. This aims to address many of the sustainability questions and viability concerns associated with operating in Arizona. 

Beyond accounting for responsible construction of the foundries from an ESG perspective, the facilities will need to be filled with complex machinery and a highly educated, diverse workforce. TSMC’s Phoenix foundry will be the first five-nanometer facility in the United States. Its size and advanced technologies mean talent will need to be cultivated or imported due to the facility fabricating the most advanced chips produced in the United States. Most of the foundry announcements have been accompanied by $100 million partnerships with local universities to develop a talent pipeline, but like construction, growing a talent pool will also take time.

The cost of labor varies by country, and this is also true for semiconductor jobs. The average annual salary for a semiconductor engineer in the United States is $118,300, while the average annual salary for a similar job in Taiwan is NT $980,000 — equivalent to $35,200, according to Payscale. This represents an increase in the cost of direct labor of approximately 350 percent, and higher costs of labor in the United States extend beyond direct semiconductor positions to construction as well. 

Get More Hardware IntelAre You Prepared for the Quantum Revolution?

 

Who Is Going to Pay for This? 

The short answer to this question is that we, the end users, will be footing the bill for this coming expansion of U.S. semiconductor manufacturing. The return of domestic chip fabrication will raise global output sometime between 2023 and 2025, and these chips will be more expensive to produce based on both direct and indirect costs. Because our society relies so heavily on chips, fabs will pass the costs on to their customers — who will in turn pass them onto us. 

“Technology has become like traditional infrastructure in that it is indispensable in the modern economy,” as a recent PitchBook report put it. “According to this theory, technology companies will succeed in passing inflationary costs along to their consumers.” 

The likening of technology in a broad sense to infrastructure is apt, and perhaps even more so when considering chips. Semiconductors are a fundamental requirement for any offering in the technology, media or telecommunications space, and they play a part in many other spaces. Take the auto industry, for instance: “A modern car can easily have more than 3,000 chips. But cars account for a tiny share of chip demand,” the New York Times reported last year. “TSMC is one of the few makers of a variety of chips vital to auto manufacturing, but in 2020 carmakers generated only 3 percent of the company’s sales.”

 

The Takeaway 

We can expect to see the infrastructure costs of these facilities show up in the coming years for a variety of goods, whether it’s for an iPhone 16 or a 2025 Ford F-150. The future price increases to technology, media and telecom goods and services can be expected to include the incremental costs associated with manufacturing chips in the United States. 

The hope is that in exchange for higher prices, semiconductor fabricators will be able to move production closer to their customers, diversify their geographical presence and ultimately streamline supply chains.

Hiring Now
JPMorganChase
Financial Services
SHARE