Independence Day: Washington is paying factories to come home

Through the CHIPS and Science Act and the Inflation Reduction Act, Washington is handing out tens of billions of dollars in grants, loans and tax credits to bring factories home. Deciding where to build a plant used to be a private calculation. It now looks more like a subsidised land grab, and semiconductor and clean-energy companies are enjoying the government's generosity.

Two laws, one summer

Both laws were signed in August last year, within days of each other. The CHIPS Act provides $52.7bn for American semiconductor manufacturing, research and workforce training, the bulk of it earmarked for grants to build or expand fabrication plants, alongside a 25 per cent investment tax credit for chip factories. The Inflation Reduction Act was scored at the time at about $369bn in climate and energy spending, most of it delivered through tax credits that carry no fixed cap. If companies build more than expected, the bill simply grows.

The corporate response was quick. Intel had already announced a $20bn complex in Ohio, TSMC raised its planned investment in Arizona to $40bn, and Micron said it could spend up to $100bn in upstate New York over two decades. Battery makers from South Korea and Japan have lined up joint ventures with Detroit's carmakers across the Midwest and the South. Much of that money was committed on the expectation of federal support rather than after receiving it.

Uncle Sam lowers the bar

By paying for up-front capital spending and guaranteeing credits for domestic output, the government is pushing down the return a project needs to clear. Investments that would have failed an ordinary discounted cash flow test on cost alone now add to earnings from the start. Corporate treasurers know the money comes with regulatory strings attached, including prevailing wage rules. Building at home also keeps getting dearer.

The arithmetic is simple. A 25 per cent credit on a chip plant means the company's own capital covers only three-quarters of the sticker price, before any grant. Every dollar of subsidy received in the first years is worth far more to a discounted cash flow model than a dollar of profit a decade out, so front-loaded support has an outsized effect on the internal rate of return. The IRA's advanced manufacturing credit works on the other end, paying fixed amounts per unit produced, such as $35 per kilowatt-hour of battery cells. That turns a thin-margin commodity business into one with a guaranteed top-up on every unit.

The IRA also lets many companies sell their credits to other taxpayers or, in some cases, receive them as direct payments. That matters for young manufacturers with no taxable profits to offset. It turns a tax break into something close to cash, and it lets banks and large companies with big tax bills act as financiers of the transition.

The strings attached

The support is not free money. Clean-energy projects earn the full credit only if they pay prevailing wages and use registered apprentices; without them, the credit is cut sharply. The Commerce Department's rules for chip grants require applicants seeking more than $150m to share profits above their projections with the government, to set out childcare plans for their workers, and to limit expansion of advanced capacity in China for a decade. Each condition is defensible on its own terms, and each adds cost or constrains flexibility.

Construction costs push the same way. Building a fab in America takes longer and costs more than in Taiwan or South Korea, partly because of permitting and partly because the country has lost much of its specialist construction workforce. Chipmakers competing for electricians and pipefitters in Arizona are bidding against battery plants in Georgia and Tennessee that are drawing on the same subsidies.

Precedents are mixed

Industrial policy in America has a patchy record. The Synthetic Fuels Corporation, created in 1980 to wean the country off imported oil, was wound down within a few years once oil prices fell. Solyndra, the solar firm that received a $535m federal loan guarantee, went bankrupt in 2011 and became a byword for picking losers. On the other side of the ledger, federal money for defence research helped create the semiconductor industry in the first place, and Sematech, the government-backed chip consortium of the late 1980s, is widely credited with helping American manufacturers recover from Japanese competition.

The difference between those outcomes tended to be whether the subsidised product could eventually compete on price. Where the subsidy bridged a technology to scale, it paid off. Where it propped up an activity that was structurally more expensive at home, the plants closed when the money stopped.

Winners and a counterargument

The clearest winners are equipment makers, construction contractors, skilled trades and the states that won the plants, along with incumbents large enough to handle the application process. The losers are taxpayers, whose bill for uncapped credits could run well above the original estimates, and allies such as the European Union and South Korea, which have complained that the rules discriminate against their producers. Europe has answered with its own Chips Act, and a global subsidy race is under way.

The best defence of the policy is that price is the wrong test. Most of the world's advanced chips are made in Taiwan, a concentration that looks reckless if relations with China deteriorate further. On that view the subsidies buy insurance and a domestic supply chain, and a plant that earns a modest return is still worth having. Learning effects might also narrow the cost gap as American plants mature.

That argument justifies the spending as a matter of national security. It does not make the investments good business. The chip plant credit applies only to construction that begins before 2027, and the IRA's per-unit manufacturing credits phase down from 2030. The factories are coming home. Plenty of them were justified by federal patronage rather than private economics, and their margins will look thin once the subsidies expire, most visibly for battery and component makers early in the next decade.