When Congress passed the CHIPS Act, a bill seeking to boost U.S. semiconductor manufacturing, the Washington Post trumpeted it as “a new era of industrial policy.” One might also call it a version of “capitalism with Chinese characteristics”—a foray into state-directed market policy. But set aside competition with China for a moment and remember that building more factories leads to more energy demands. It takes about 1,000 times more energy to fabricate a pound of silicon than it does to make a pound of steel. And the world already consumes silicon devices by the tens of kilotons every year. Meantime, as President Biden hails the green energy subsidies in the Inflation Reduction Act as “the biggest step forward on climate ever,” that policy sends the U.S. down the same energy path as Europe: high prices, low reliability, nearly intractable dependencies, and the prospect of deindustrialization as energy-intensive factories close down. That’s hardly a business climate conducive to expanding any energy-hungry industry.
In any case, the CHIPS Act is not the federal government’s first foray into industrial policy. In 1983, the Reagan administration, in “its desperation to do something that will help this country compete with Japan,” proposed (but failed to create) a new federal department in the model of Japan’s Ministry of International Trade and Industry. At the time, Japan had surpassed America in automobile and steel manufacturing and was, observers believed, about to pass the U.S. in computers. But the government never passed anything nearly as ambitious or expensive as the CHIPS Act, and its restraint succeeded. America boomed; Japan didn’t.
Of course, certain circumstances are different today. Semiconductors and their manufacturing plants (known as “fabs”) matter now more than ever for both military technology and the overall economy. Annual global purchases of semiconductors have passed $650 billion, now exceeding polymers ($600 billion) and approaching the $900 billion global steel industry. These basic inputs are the building blocks used to create everything else in civilization.
Public awareness of the importance of semiconductor supply chains came courtesy of the Covid lockdowns. When Reagan was president, semiconductors and electronics constituted about 10 percent of the cost of the bill of materials to produce a car. That share is now 40 percent and rising. And electric cars—which stand to benefit from those IRA subsidies—use at least twice as much silicon as conventional ones.
The importance of semiconductors extends beyond cars. Chips are crucial to the emerging industry of mobile robots, machines once relegated to fantasy but now increasingly common in the automated warehouses at the center of e-commerce. Half a million mobile robots were built last year, with growth expected to jump tenfold over a decade. Meantime, semiconductors are the building blocks of the cloud, where global capital spending exceeds that of the world’s electric utility industries. And they figure into rising demands to digitalize medicine, education, and training. As I’ve documented in my book The Cloud Revolution, all these markets for chips are still in the early innings.
The U.S. still possesses 50 percent of the market share of today’s global semiconductor industry, but its share of most advanced fabs has shrunk. In 1990, at the beginning of the tech bubble, the U.S. and Europe combined had a roughly 80 percent share of global fab capacity (roughly equally divided between the continents). Today, that share is about 20 percent (again about equally divided), with the other 80 percent in Asia. Some 90 percent of leading-edge fabs are in Taiwan and South Korea. Congress’s motivation for the CHIPS Act is no mystery.
The U.S. is not alone in subsidizing domestic chip fabs. Including Europe and Japan, about $100 billion has been pledged in chip subsidies (with China’s true number unknown). But the reshoring of American fabs was underway before the new law passed, as firms rediscovered the value of resilience, security, and location amid geopolitical turmoil. And major chipmakers have collectively announced plans for more than $650 billion in spending. While some semiconductor firms lobbied hard for the CHIPS Act, it requires a remarkable faith in the power of subsidies to believe that what amounts to a 15 percent subsidy is the primary catalyst for such ambitious capital commitments.
Driving this expansion is instead a conviction about the magnitude of future silicon demands. Consider the private partnership announced between Brookfield Asset Management and Intel. The Canadian fund recently signed a $30 billion deal to help finance more Intel U.S. chip fabs and split the revenue. Brookfield’s CEO said that the strategy follows the same deal-making common in other capital-intensive industries. Companies will happily take any supplemental money, but the market opportunity, not the subsidies, underpins such big investment gambles. The key, to use the famous truism from Walter Wriston (the former, storied CEO of Citibank), is that “capital will always go where it’s welcome and stay where it’s well treated. Capital is not just money. It’s also talent and ideas.”
Moreover, the CHIPS Act ignores the rest of the labyrinthine semiconductor supply chain. Silicon chips, once fabricated, must be put into packages, an industry the U.S. no longer has onshore. Similarly, an array of chemicals and components are critical inputs for chip fabs, and most are not found in America. For example, with over 60 percent of global supply, China dominates the mining of fluorspar, which is used to produce critical chemicals for chip fabs. Two Ukrainian companies produce over half of the worlds semiconductor-grade neon, another critical input for fabs. U.S. production of such materials has declined in recent decades because America’s policies have been hostile to chemical refining and mining industries. The CHIPS Act does nothing to ameliorate that.
The regulatory state, not any lack of subsidies, is the principal reason that it’s so hard to build big, industrial, supply-chain-critical projects of any kind in the U.S. In the 1970s, filing an environmental-impact statement took only a few years. Now it takes over eight years, on average. It took five years to get a permit merely to expand the Bayonne Bridge in New York. Completing the environmental review process to dredge the supply-chain critical Port of Savannah took 14 years.
Even government subsidies don’t trump regulations. (President Obama discovered as much with his administration’s own “shovel-ready” industrial stimulus.) Though Senator Joe Manchin insists that Democrats have promised to reform regulations and permitting, the party’s major legislative achievements—the CHIPS Act, the IRA, and the Infrastructure Investment and Jobs Act—won’t change anything in that regard.
If the next Congress wants to deliver on what’s promised in the CHIPS Act, it doesn’t need to dole out any more money. It needs to rein in the regulatory state.
Photo by Justin Sullivan/Getty Images