When President Biden visits a microchip factory under construction in Arizona on Tuesday, it may look like a political victory lap: The factory will bring $12 billion and thousands of jobs to an important swing state that just elected a Democratic governor and senator. But the chips the factory will manufacture carry far more significance than being a partisan maneuver. They are essential to American security.
Silicon chips, or semiconductors, the tiny integrated circuits that power electronic devices, are the reason we can send texts or turn on the television. They are the means by which pilots can fly aircraft safely and militaries can monitor missiles on radar. The potency of every chip depends on the number of transistors squeezed onto its surface, and because that number is growing exponentially, so too is what the chips can accomplish. The inventions that emerge, especially in artificial intelligence and supercomputing, will determine not only who will lead the global economy but also who will win wars. The future, in short, depends on chips.
The problem is that, right now, the United States can’t depend on a steady supply of chips, even though the design and software are mostly developed here. After half a century of global outsourcing, the manufacturing process has been dispersed, with each step along the way becoming highly concentrated in a few countries. The silicon wafers that carry the transistors are mainly made in Japan. The lithography tools that pattern chips are mainly made in Japan and the Netherlands. Processor chips themselves are fabricated largely in Taiwan, notably by the company TSMC (which is building the factory Biden will visit in Arizona). The chips are then tested and packaged into devices, primarily in China.
The dangers of such a system are obvious: Taiwan plays an important role in the production of all types of chips, but it produces a whopping 90 percent of the most advanced semiconductor chips — the ones essential to innovation. In the industry, these are called “leading edge.” This puts the United States in a tenuous position. If China were to invade Taiwan, which it claims is an “inalienable part” of its territory, the United States could lose access to the bulk of the tiny components that make our country run. We got a taste of that when the covid-19 pandemic disrupted supply chains and, suddenly, it took half a year for a new refrigerator to arrive. An invasion of Taiwan would mean a repeat — but on a catastrophic scale and with control in the hands of an adversary. The Pentagon, let’s not forget, deems China the top threat to U.S. security.
The Chips and Science Act passed by Congress this summer is supposed to help address the problem by funneling more than $50 billion in subsidies to onshore manufacturing as well as bolstering research and development. Whether it will work depends on implementation. The key will be directing the subsidies not just toward the fabrication of chips but to testing and packaging, too; otherwise the United States would remain reliant on China. Another challenge will be ensuring that research and development money produces results beyond shiny new buildings at universities. Innovative technologies must move — as the jargon puts it — from the lab to the fab(rication). Oversight of how the money is spent is crucial.
In a parallel move, in October, the Commerce Department banned the export of leading-edge chips used in military applications as well as advanced chipmaking tools. The export controls also forbid “U.S. persons” from servicing advanced chipmaking facilities in China. Taken together, these steps could suffocate China all along the supply chain. Even foreign companies are barred from selling their leading-edge chips to China without a U.S. government license if they want to continue using American technology — which almost every semiconductor firm does.
The aim is not only to make it harder for China to buy chips but also to make it harder for China to build chips — thereby stymying both its work in AI today and its hopes of harboring an entirely domestic microchip industry tomorrow. Xi Jinping’s regime has already invested heavily in this effort.
If China did achieve supremacy in advanced chip manufacturing, military and economic supremacy could easily follow. That kind of power, wielded by a regime defined by its dystopian surveillance systems and the violent repression of a cultural minority, is a frightening specter. But we’re far from that reality. China spends as much money importing chips as it does oil. China may be able to outcompete the United States when it comes to building the most ships or the most drones — but this country can probably beat it in building the best systems to control them. That’s the edge we must maintain.
Some argue that the White House’s aggressive actions risk what could have been a permanent U.S. economic advantage by forcing China to develop its own chipmaking capabilities that may eventually surpass our own. But China was already building its domestic industry. The export controls are intended to put China so far behind that catching up will prove difficult. However, that depends on the measures actually being effective.
One challenge is enforcement. The Commerce Department’s plate is overflowing. Congress must not underfund it. Another challenge is bringing allies on board. At the moment, the United States is so integral to the semiconductor supply chain that pretty much no country or company can produce leading-edge chips without its involvement. But the controls could create pressure for companies in Japan and Europe to manufacture the kind of tools that are restricted and provide them to China themselves, rather than accept the loss of a valuable customer. Diplomacy matters here.
The good news is this: In the public mind, software has long been the big story — why has the Chinese company TikTok so consumed members of Generation Z? How should Facebook and Twitter treat posts from former president Donald Trump? There’s been far too little focus on the hardware without which none of these platforms could exist at all. Now, as the beginnings of comprehensive semiconductor strategy come into view, it looks like that’s finally changing.