Will the CHIPS Act Make American Semiconductor Manufacturing Great Again?

Seven reasons you shouldn’t count on it

Editor’s Note: The following is a guest author post from Kevin Klyman, a technology policy researcher at Harvard’s Kennedy School. He has worked on international technology policy at the UN and Human Rights Watch and his writing in Foreign Policy and The American Prospect has broken news about export controls and digital trade deals. You can follow him on Twitter @kevin_klyman

When President Biden signed the CHIPS and Science Act into law, he said, “the future of the chip industry is going to be made in America,” by “American companies.”

The legislation made a serious bet that American semiconductor giants would help the US regain technology leadership, despite their history of union busting, outsourcing production to East Asia, prioritizing stock prices above national security, and failing to stay at the technological frontier.

The core premise of the CHIPS Act is that increased government spending can make the US a semiconductor manufacturing powerhouse by boosting domestic production of advanced chips and increasing America’s share of global semiconductor production. In the words of National Security Advisor Jake Sullivan:

America now manufactures only around 10 percent of the world’s semiconductors, and production—in general and especially when it comes to the most advanced chips—is geographically concentrated elsewhere. This creates a critical economic risk and a national security vulnerability.

While the CHIPS Act was an unprecedented investment in American high-tech manufacturing that could ease barriers to future government spending, it’s far from clear that it will make American semiconductor manufacturing great again. This cornerstone of Bidenomics is unlikely to achieve its aims for no less than seven reasons.

1) Multinationals Won’t Do the Right Thing

Multibillion dollar semiconductor companies do not have the interest of the nation or their workers at heart.

They will prioritize their shareholders above all else unless and until the government forces them not to. In the case of the CHIPS Act, Washington did not do nearly enough to ensure that semiconductor companies use subsidies to bolster economic development rather than their bottom lines.

Progressives notched a significant win in the negotiations over the CHIPS Act by pressuring the Commerce Department to ban companies from using government subsidies to buy back their own stock. But these restrictions are imperfect, ignoring the fact that an increase in taxpayer dollars coming into companies’ coffers frees up other, unrelated money that can then be used to reward shareholders.

For instance, the day after the CHIPS Act passed, Intel slashed capital expenditure by $4 billion in order to increase its dividend to shareholders by $4.5 billion. Intel is not alone in increasing the portion of its profits it pays to shareholders in the wake of the CHIPS Act—Micron hiked its dividend by 15% three days after the bill passed, and Broadcom has increased its dividend by 12% in 2023. This is part of a broader trend: from 2011-2020, Broadcom, IBM, Intel, Qualcomm, and Texas Instruments spent $249 billion on stock buybacks, equivalent to 70% of their profits. The world’s most valuable semiconductor company, US chip designer Nvidia, announced in August that it will use record profits from sales of AI chips to give handouts to shareholders to the tune of $25 billion.

The Commerce Department claims to address these concerns by prioritizing funding for companies that commit to limiting stock buybacks, but this is not a coherent strategy. There are only a handful of companies that are in contention for major subsidies, and stock buybacks will not stop Commerce from giving preference to top US companies over foreign firms or smaller players. Lobbying is the elephant in the room here—US semiconductor companies broke industry lobbying records for their lavish spending to persuade lawmakers to pass the CHIPS Act, with companies like Intel and Micron doubling their spending.

A modest approach to curtailing corporate greed just won’t cut it. As long as the federal government allows semiconductor companies to invest in shareholders instead of factories, industrial policy will not make the US a global leader in semiconductor manufacturing.

2) Ends-Means Imbalance

$75 billion in spending over 10 years is insufficient to do more than nudge America’s overall market share in semiconductor manufacturing, to say nothing of recapturing global leadership.

Morris Chang, founder of Taiwan Semiconductor Manufacturing Co., the world’s largest chipmaker, said of the CHIPS Act:

right now you’re talking about spending only tens of billions of dollars of money of subsidy … it will be a very expensive exercise in futility. The US will increase onshore manufacturing of semiconductors somewhat. But all of that will be very high-cost increase, high unit cost. It will be noncompetitive in the world markets where you compete with factories like TSMC.

TSMC alone spends $32 billion on capital expenditure annually—four times as much as the CHIPS Act, even without accounting for generous subsidies from the Taiwanese government.

While the CHIPS Act may create sufficient domestic production capacity to meet demand from the Pentagon and achieve other incremental goals, it is unlikely to transform America’s position in the global semiconductor market. The canonical 2020 study on the issue from BCG and the Semiconductor Industry Association projected that a $50 billion investment in semiconductor manufacturing would increase US global market share in chip production from 10% to just 13-14%. In October, Goldman Sachs estimated the CHIPS Act would increase the US share of global semiconductor production by less than 1%, as the capital intensity of semiconductor manufacturing is increasing faster than expected due to higher costs associated with next generation chips.

If only there were somewhere we could find more money to spend on economic priorities…

Stephen Semler

3) Chips Don’t Pay

Manufacturing chips in the US is extremely inefficient.

Compared with semiconductor factories in East Asia, US factories take 50% longer to build, they produce half as many chips, and construction is 400% more expensive. The US produces very few advanced chips today, but even with the full force of Washington behind them US factories will be able to meet only one-third of domestic demand for advanced chips.

When President Biden spoke at the ribbon-cutting of TSMC’s Arizona factory in December, he said the factory would produce “the most advanced semiconductor chips on the planet.”

But this is no longer true: TSMC has delayed production at its Arizona plant by a year due to cost overruns, and by the time that plant produces chips in 2025 TSMC’s factories in Taiwan will be shipping even more sophisticated chips.

4) Foreign Firms Dominate Semiconductor Manufacturing

US firms are likely too far behind industry leaders to catch up.

TSMC receives over half of all global orders to produce chips, which has enabled it to perfect its manufacturing processes and mass produce chips at lower prices. TSMC produces 90% of the world’s leading-edge logic chips, while Samsung produces the rest; Intel remains two generations behind.

Samsung, SK Hynix, and Tokyo-based Kioxia control 70% of the market for memory chips, with Micron controlling just 30% of the market for dynamic random access memory chips and 10% of flash memory. Even if the CHIPS Act increases US semiconductor production, foreign firms will have an edge because they produce so many more chips that they have a near-insurmountable lead in process engineering.

While Biden has called Intel’s new production site in Ohio a “field of dreams,” the reality is that Intel has lost its manufacturing leadership as it has consistently failed to meet its technology targets. Financial analysts noted this month that major production issues will lead Intel to pay foreign companies nearly $20 billion annually to manufacture its chips, outsourcing $10 billion worth of production to TSMC in 2025.

5) Insufficient Demand

The CHIPS Act was poorly timed as the semiconductor industry is in recession.

For the first time in four years, overall capital expenditure in the industry will decrease in 2023 due to a weak electronics market and overproduction in the wake of pandemic-induced shortages. Prices for memory chips have declined 50% year-over-year (YoY) even as the AI bubble provides some relief.

The recession has dealt a blow to US chipmakers: Intel’s revenue fell by 20% YoY in 2022, and in the second quarter of 2023 Texas Instruments’ revenue fell 13% YoY and Micron’s fell by 50%. Profits have cratered, leading Intel and Micron to lay off more than 10% of their workers. The US is pumping money into semiconductor manufacturing precisely when companies are pulling back due to a lack of demand.

6) The Toll of China’s Retaliation

China has retaliated against US restrictions on its semiconductor industry, seriously damaging America’s national champions.

For example, China’s antitrust regulator, the State Administration for Market Regulation, recently blocked Intel’s acquisition of Tower Semiconductor, upending Intel’s business strategy. Intel’s plan to compete with TSMC and Samsung is predicated upon mimicking their business model by becoming a contract manufacturer that produces chips for third parties. In order to do this, Intel needs to acquire a contract manufacturer to jumpstart the process, but Tower was the only viable option. China was concerned that US restrictions on its tech sector would make Tower less likely to partner with Chinese firms if it were purchased by Intel, leading Beijing to sink the deal.

In May, China limited procurement of Micron’s chips by operators of “critical information infrastructure” after an investigation by the Cyberspace Administration of China alleged (without evidence) that Micron’s products pose a cybersecurity threat. Though Micron initially downplayed the significance of these rules, it later admitted that it could lose up to 12% of its revenue as a consequence. China’s restrictions are a direct response to US rules prohibiting companies that receive CHIPS Act subsidies from expanding factories in China.

7) US Strategy Relies on Restricting Its Allies

Finally, the US is at odds with its allies regarding its semiconductor strategy.

The White House was triumphant after Biden’s meeting at Camp David with the leaders of Japan and South Korea, but the summit papered over significant disagreements on chips. While the US hopes to undermine China’s semiconductor industry and “slow down China’s rate of innovation” per Commerce Secretary Gina Raimondo, firms in Japan and South Korea rely on China as their most important market. Semiconductors are South Korea’s largest export by far, and more than half of them go to China.

The US has placed significant restrictions on firms in Japan in its effort to beat China in the tech war. The Commerce Department’s October 2022 semiconductor export controls put a target on the back of Tokyo Electron and Nikon, threatening them with sanctions if they continued to sell advanced semiconductor manufacturing equipment to Chinese firms. By contrast, the US issued waivers to chipmakers like Samsung, SK Hynix, and TSMC, allowing them to continue importing certain equipment for their factories in China.

Tokyo Electron and Nikon immediately complied with US export controls, leading to a reduction in sales for such equipment. Japan subsequently adopted its own export controls to bring it into compliance with US law, with an anonymous Japanese official remarking that Tokyo was bemused that its sovereignty could be disregarded so easily.

In 2019, an executive at Tokyo Electron explained that the company must comply with these types of US restrictions on China’s tech sector because “it’s crucial for us that the US government and industry see us as a fair company,” while an executive at another Japanese equipment supplier said:

we could be in deep trouble if we take advantage of the US export ban to expand businesses with China.

The risks of expanding capacity in China became clear for Korean companies in the wake of Beijing’s restrictions on procurement of Micron’s chips. Washington has lobbied the government of South Korea to prevent Samsung and SK Hynix from backfilling Micron’s supply of memory chips to China, causing Seoul to change its stance and refrain from encouraging such backfilling.

Samsung and SK Hynix are vulnerable to US restrictions on China because they produce 40% of their chips in Chinese factories. The US has ample leverage over South Korean chipmakers. If Samsung or SK Hynix step out of line, Washington could theoretically let export control waivers expire. Both firms are also seeking waivers on CHIPS Act rules barring companies that receive US subsidies from expanding capacity in China; the Commerce Department could reject these requests if Samsung and SK Hynix opt to increase sales to Micron’s former clients. 

But coercion breeds ill will and only goes so far. In the near term, Beijing is likely to retaliate against South Korea and Japan if their semiconductor firms refuse to do business with China, putting both nations in an uncomfortable double-bind. The more restrictions the US imposes on firms in South Korea and Japan, the more likely they are to engineer US components out of their products over time and seek sources of funding with fewer geopolitical strings attached.

“Made in America” Isn’t All It’s Cracked Up to Be

Each of these issues with the CHIPS Act raises the question of why the US should strain to produce more semiconductors domestically.

One line of reasoning often cited by policymakers is that high-tech manufacturing creates good, union jobs and spurs US industrial capacity. While this may be true on aggregate, workers’ rights were not advanced by the CHIPS Act—in the final negotiations, rules requiring firms to pay prevailing wages and remain neutral in union elections were nixed.

A more pressing justification for increasing domestic semiconductor manufacturing is the idea that the US could lose in a war with China over Taiwan if it cannot access Taiwanese chips or produce substitutes at home. But a sufficient stockpile of chips would solve this issue, obviating the need for the US to lead in semiconductor manufacturing.

A prior question is whether the US should formulate its economic policy on the basis of an imminent war with China—which may in fact increase the likelihood of war—or if it should do everything in its power to prevent such a war.

Though taking every conceivable step to prepare for conflict is often framed as sensible policymaking, the safest option would be to make avoiding World War III the first priority.