The CHIPS Act already puts America first. Scrapping it would poison the well for US investment.

In his recent speech to Congress, President Donald Trump claimed that companies receiving CHIPS and Science Act subsidies “take our money, and they don’t spend it,” urging lawmakers to “get rid of” it. (CHIPS stands for Creating Helpful Incentives to Produce Semiconductors.) Trump touted tariffs as a better tool to induce investment in the US, citing the Taiwan Semiconductor Manufacturing Corporation’s (TSMC) recent commitment to boost its US investment from $65 billion to $165 billion after he threatened to impose duties on chip imports.

However, revoking the 2022 CHIPS Act would threaten not only the current boom in US semiconductor investment but also any future programs the government devises to incentivize investment that competes with China. Instead, the Trump administration would do better by making the legislation’s funding more targeted and efficient, including by reforming its over-budget tax credit. The CHIPS Act has brought enormous investment into the US from top semiconductor firms around the world, bringing US manufacturing capabilities closer to the cutting edge and improving US national security. Even after stripping out post-2019 inflation, the CHIPS Act spurred an estimated over $110 billion in 2019 dollars of investment in the US (see figure below), more than the entire amount of real investment in facilities to manufacture electronics and computers from 2007 until 2020, when the precursor to the CHIPS Act was approved. The CHIPS Act passed Congress with bipartisan support, and it grew out of the National Defense Authorization Act for Fiscal Year 2021 passed in the last days of the first Trump administration.

Unlike a tariff-based approach, the CHIPS Act has spurred investment without raising costs that would reduce the competitiveness of the far more numerous semiconductor–using industries like AI companies locked in a high stakes race with China.

If the US government pulls the rug out from under the promised incentives, chipmakers will cancel or scale down their planned investments in the US. Firms in any other industries the US wants to support in the future will then question whether they should invest based on US policies that could be canceled on a whim.

CHIPS Act subsidies and tax credits successfully stimulated US production

While the US dominates much of the chip supply chain, by 2020 it held a small and shrinking share of semiconductor fabrication, with no production of advanced chips on American soil. US policymakers believed that attracting production to US territory would reduce the risk that virtually all the world’s advanced semiconductor manufacturing could be shut off in an invasion or blockade of Taiwan and would benefit US national security and hopefully lead to positive spillovers for the US economy and technological innovation. The CHIPS Act induced many of the world’s leading semiconductor companies to produce chips in the US by offering a generous advanced manufacturing tax credit of 25 percent to facilities that make semiconductors or equipment that makes chips (see figure below). Essentially any firm investing in making either is eligible for the tax credit, which has no cap on the amount. The tax credit will end up being by far the largest subsidy in the CHIPS Act, although most attention has been focused on a targeted program of $32 billion in grants and $5.5 billion in loans. Other crucial programs include aid for workforce training and research and development (R&D) support Workers with needed skills are scarce around the world, and without them the US cannot make chips here. These programs also include powerful guardrails to ensure recipients cannot also expand chip production in China.

Though it is more expensive to build and run semiconductor manufacturing in the US than in East Asia, figure 1 shows that the CHIPS Act incentives overcame the cost differential to spark a surge in investment far beyond expectations.

Holding prices constant at the level of the fourth quarter of 2019 to avoid counting post-pandemic inflation as higher investment, calculations show investment in US facilities to produce electronics, computers, and electricals rocketed from averaging just under $7 billion per year in the decade through 2020 to over $90 billion in 2024. Such real investment is now consistently running at around $7.5 billion per month. Importantly, these data are for realized investment in construction, not future promises that could be pared back. Without correcting for inflation, I estimate that around $158 billion in investment was catalyzed by the CHIPS Act already, averaging around $5.7 billion in investment per month for 2023 and 2024.[1] This estimate is extremely conservative because about half of the cost of building a fabrication plant to make semiconductors is in specialized equipment worth hundreds of millions of dollars and are not included in these figures.[2]

Announcements of investments including future spending are even greater, at more than $500 billion, according to the Semiconductor Industry Association. Nothing other than the CHIPS Act could plausibly have caused such a huge increase in investment, which has not ﷟ happened in other chip producing countries. The higher-than-expected demand for grant funds gave the US government’s Chips Program Office more leverage to condition grants on making larger investments in manufacturing more advanced chips, giving the US an even better deal than lawmakers hoped.

Chips production shall not live by tariffs alone

TSMC’s promised $165 billion investment seems to improve on what the Biden administration negotiated, but it builds on the CHIPS Act’s success and was not driven solely by tariffs. Commerce Secretary Howard Lutnick is right to say that the investment is larger and came without additional grants. Yet, TSMC presumably still expects to receive the 25 percent tax credit, which would be a subsidy of $25 billion on $100 billion of investment, a subsidy rate of 25 percent. That would be a larger total subsidy amount than the roughly $23 billion in grants and likely tax credits on TSMC's previous $65 billion of investment, a subsidy rate of 35 percent, that it received during the Biden administration. TSMC’s commitment will bring its US fabrication plants, or fabs, closer to the cutting edge, locate advanced packaging in the US so at least some of the chips made in Arizona will no longer need to be shipped to Asia before being incorporated into devices, and for the first time fund major research and development in the US.

Yet, it is surely thanks in part to the CHIPS Act that TSMC is willing to invest more with a lower subsidy rate. The risk of investing only a few years ago was far greater, because it was unclear whether it would be a success. There was no advanced chip production in the US, and TSMC had no experience building advanced fabs outside of Taiwan. TSMC has more confidence now that it has overcome issues with unions, zoning, and cultural differences to produce advanced 4 nanometer chips like those used today in iPhones. The share of defective chips made in the US, a key driver of whether production is economically viable, is on par with, if not better than, production in Taiwan. It would not have had this experience without the CHIPS Act. The AI boom also helped, because many of TSMC’s biggest spending customers like Apple and Nvidia are based in the US.

While tariffs could incentivize investment in the US by raising the costs of imported chips, there may be other reasons for TSMC to make the investment, such as shoring up US support of Taiwan’s security or avoiding being pushed into a potential joint venture with Intel.

Tariffs also face two challenges. The first is that it would be challenging to apply tariffs on Taiwanese semiconductors because most are imported to the US after being tested and packaged or incorporated into devices elsewhere, in either case entering the US from different countries in different customs categories. Taiwan only exported $8.2 billion in semiconductors to the US in 2023, so even substantial tariff rates would not justify such a large TSMC investment, and the US would need a (nonexistent) sizable apparatus to track down and impose tariffs on chips.

Second, building semiconductor fabs takes years (TSMC’s Arizona fab took four), so near term tariffs cannot bring near term increases in US supply. Instead, tariffs on semiconductors would lead to higher prices for everything containing chips in the short term for a potential benefit of inducing more production here years in the future. Firms may hesitate to make such investments considering that the additional chip supply would largely come after Trump completes his second term in office, when the tariffs that justified that investment may no longer be in place.

The US government should be cautious about inferring from TSMC’s big move that other CHIPS Act recipients would also expand their US investments without subsidies. TSMC is in a uniquely strong financial position because of its dominance in advanced chip manufacturing. Samsung’s foundry business, for example, has been losing money and thus would have a hard time justifying major investments in a high-cost jurisdiction like the US without subsidies. A PIIE analysis of the CHIPS Act by Gary Clyde Hufbauer and Megan Hogan found that over 80 percent of the investment projects by value that received grants probably would not have happened without them, and those investments support over 40,000 permanent jobs. The analysis concluded that some smaller projects might have been undertaken without a CHIPS Act grant or loan, but “for the larger projects, CHIPS Act grants might have been essential.”

Improving the CHIPS Act

The CHIPS Act’s grants and loans, awarded only to projects US government officials judge to be most likely to contribute to US security and manufacturing prowess, have a hard budget constraint. If investments continue at current levels until 2026, the cost of the tax credit alone could reach over $73 billion, far above the Congressional Budget Office’s initial $24.25 billion estimate. This tax credit, by contrast, is available to any semiconductor manufacturing project in the US without any cap.

To control costs but avoid disrupting investments that have already been made, policymakers could consider more exacting eligibility requirements for the tax credit for projects that either have not yet broken ground or expand their scope. The legislation could be revised to include such rules or empower the CHIPS Program Office to make eligibility subject to a cost-benefit analysis similar to that already used for grant proposals. Such reviews would ensure projects claiming the tax credit are worth the expense and help catch any firms trying to illegally claim it. To implement this, the US government needs to retain the kind of expertise the CHIPS Program Office has built inside government to ensure the government can verify whether companies have lived up to their end of the bargain.

The CHIPS Act could also be tweaked to focus it more narrowly on its objective, removing provisions like childcare requirements that have been criticized for trying to achieve progressive priorities unrelated to semiconductors.

Conclusion: Keep CHIPS

The CHIPS Act has exceeded expectations and is worth keeping both for its own sake and to signal that the US stands behind its commitments when investors build here. TSMC’s recent commitment to increased investment is an impressive achievement, but it is not only due to the Trump administration's tariff threats. Tariffs would take many years to boost domestic production and would not likely have as strong an effect. Meanwhile, they would make everything from phones, laptops, and AI infrastructure more expensive for US consumers, businesses, and government agencies that use chips. Careful adjustments to make it more cost effective could make sense, but removing support from existing projects that counted on the CHIPS Act would damage US credibility with investors who put billions of dollars to work based on US government promises. It is not worth the risk to future investment in industries crucial to national security and economic growth.

Notes

1. To calculate this figure, I subtract the Clean Investment Monitor’s figures for investment in production of green technologies including batteries, electrolyzers, fueling equipment, solar, and wind from the US Census Bureau figures, which from the author’s correspondence with Census staff should include these categories. I also subtract a baseline of investment at average rates from 2011 to 2020 to avoid counting investments that would have happened without the CHIPS Act.

2. Author conversation with Census staff strongly suggests semiconductor manufacturing equipment is not part of these figures. Per Census staff, “Cost and installation of production machinery and equipment items are excluded unless the equipment is fixed, largely site-fabricated equipment not housed in a building,” consistent with the data’s definitions.