At one time, the chipset cemented America’s position as a preeminent technological power. From the 1950s to 1970s, the US government and semiconductor companies like Fairchild Semiconductor and Intel worked together to turn America into the world’s biggest manufacturer of semiconductors. Semiconductors were the technological foundation of the Apollo missions, the first heat-seeking missiles and advanced rocketry, the first PCs, mobile phones and televisions.
The industrial policy that built American technological power was thought to be a relic of the Cold War. But now, industrial policy is back. On Friday, the House of Representatives passed the America Competes Act, a bill that promises $52 billion in government funding to bring back American leadership in semiconductor design and manufacturing. Wide-ranging but thinly spread, the text calls for the creation of a consortium called the National Science and Technology Center and other government programs, to compete with Chinese technology.
The White House backed the bill, which looks likely to pass in the Senate. “The House took a critical vote today for stronger supply chains and lower prices, for more manufacturing – and good manufacturing jobs – right here in America,” President Biden said in a statement. “I look forward to the House and Senate quickly coming together to find a path forward and putting a bill on my desk as soon as possible for my signature.”
It sounds like the right idea. But the bill doesn’t have the concentrated financial firepower that’s needed for a successful industrial policy. If it’s executed without focus, it could fail to restore America’s semiconductor presence against China. The US government might only have one shot before it’s fallen too far behind.
Successful industrial policies have almost always earmarked government funding for a handful of high-performing companies and industries. For the past fifty years, the typical semiconductor consortium in America, South Korea, Japan and Taiwan consisted of no more than five major companies and a dozen at the most.
Semiconductors are a capital-intensive, unpredictable industry. They require multibillion dollar investments five to ten years in advance, with no certainty as to whether the market will turn boom or a bust.
These market cycles can be ruthless for private enterprises that don’t have state backing. One year, the industry might be flooded with too many of a single semiconductor, causing prices to plummet. After competitors close or exit, there will often be a dearth of semiconductors, raising profits but leading to damaging shortages that hurt national security and technological innovation.
We are in the latter cycle right now, with America and its allies South Korea and Taiwan unable to produce enough advanced chipsets. The result is supply chain instability, shortages, rising prices, and stalled technological innovation.
This is why most governments, except for the US, subsidize their semiconductor industries. State consortiums in Japan, South Korea and Taiwan have succeeded when companies have pooled together some of their intellectual property, in exchange for government money and lucrative defense contracts.
But the America Competes Act reads like a diverse platter of unrelated political one-ups, rather than a target investment in America’s strategic industries. It contains provisions to fight deforestation, help coral reefs and engage in “climate diplomacy provisions” that fall outside of the bill’s focus and are unrealistic with constraints in time and funding.
These are all causes on their own, but they detract from the core issue of technological competition. The bill is spread too thin. The America Competes Act earmarks $52 billion for semiconductor manufacturing, but this is not enough considering the scope of the bill. The money is to be divided between new fab plants, research and development initiatives, and new government offices. But these endeavors are costly.
Consider that Taiwan’s TSMC plans to spend $9.4 billion building a new fab plant in Taiwan. It also believes that the cost of future, cutting-edge plants could reach $20 billion. The top ten companies, including TSMC, spend $43.5 billion a year on R&D. When we add up those numbers, $53 billion falls short of a decade-long effort to build a stable, influentialsemiconductor industry.
As the Congress works to conference the COMPETES Act with the Senate-passed U.S Innovation and Competition Act, lawmakers will need to focus the legislation to focus on core areas of agreement. Members of Congress should consider the long-term needs of the ensuring American competitive advantages in the chip sector and ensure that enough resources are available.
China has the benefit of planning for the long view. With a strong central government and a long history of industrial planning, it has a leg up against America. American policymakers must not allow their decentralized, federalist democracy to become a weakness in the race for the next semiconductor. As our Cold War history has shown us, democracy is our biggest asset to staying ahead of the curve. We just need to provide the resources, the funding and the direction to our strategic industries to keep them ahead.
Troublemakers: Silicon Valley’s Coming of Age by Leslie Berlin
How Asia Works: Success and Failure in the World’s Most Dynamics Region by Joe StudwellThe Intel Trinity: How Robert Noyce, Gordon Moore, and Andy Grove Built the World's Most Important Company by Michael S. Malone