Semiconductors sit at the heart of many of today’s most consequential technologies, powering advances in artificial intelligence, quantum computing, and next-generation wireless networks. Leadership in semiconductor design and manufacturing has become a measure of national and international strength and economic security—prompting governments and companies around the world to invest heavily in research, development, and production capacity.
Willy Shih is the Cizik professor of management practice in business administration at Harvard Business School, where he focuses on manufacturing strategy, innovation, and supply chains. Before arriving at Harvard, Shih spent 28 years working across the technology sector at firms such as IBM, Silicon Graphics, Eastman Kodak, and Thomson SA. In this conversation, he explains why semiconductors have become so central to global innovation—and what their rise means for the future of the U.S. economy.
Why have semiconductors moved from a behind-the-scenes technology to a central concern for economic and national security?
Top of mind is the rivalry between the U.S. and China and the competition for leadership in AI. That race depends on computing hardware, which uses chips from companies like Nvidia that are built with the most advanced semiconductor processes in the world. The Taiwan Semiconductor Manufacturing Company (TSMC) controls the vast majority of that capacity. China claims Taiwan is a rebellious province and has been carrying on numerous naval exercises in the waters surrounding it. TSMC is building new factories (called “fabs”) in Arizona, but they currently lag the ones in Taiwan in having the most advanced processes.
Add to this the anxiety that happened during the pandemic when manufacturers of vehicles and appliances suddenly realized that they were overdependent on an undiversified supply of chips coming out of Asia. This was brought back into focus last fall when Nexperia, a major supplier of automotive chips, got caught in a dispute between the Dutch and Chinese governments. That’s why there is so much concern across governments and companies who use semiconductors: concentrated supply in geopolitically fraught regions.
What can policy realistically achieve to bring significant semiconductor manufacturing to the U.S., and what can’t it?
While policy initiatives can try to make investment conditions more favorable in the U.S., market forces are still at work. Costs in the U.S.—for fab construction, for operations, for many raw material inputs—are higher than in Asian countries, so domestically produced chips can be a lot more expensive. I have advocated for demand-side incentives that would cause companies to preferentially purchase American-manufactured chips while those new fabs improve their costs.
Having a capable workforce is another area the government can help with. STEM education is one of the best investments the government can make. If we step back and look at the bigger picture, I think the most valuable thing the U.S. government can do is fund long-term research. The Department of Defense’s Very High-Speed Integrated Circuit Program in the 1980s became one of the foundations of modern semiconductor design and manufacturing processes. There are numerous other examples. It has always been important for the government to fund long-term, risky R&D that is beyond what private companies can do. I fear we have moved to a more short-term transactional view of the industry with recent cuts to R&D that were mandated in the CHIPS Act.
Does the semiconductor industry depend on international cooperation, or is the U.S. poised for success in domestic markets alone?
The semiconductor industry and the broader electronics industry are both highly globalized. Inside a modern fab, there are materials and tools that come from all over the world. If the U.S. wanted to be 100 percent self-sufficient, it would take two decades and would be prohibitively expensive. We would have to build a lot of specialty skills that we just don’t have.
The other thing we have to think about is: where is the demand for semiconductors? A lot of it is centered in Asia because, over the last 25 years, that’s where a lot of electronics assembly moved. The U.S. is still the richest market in the world, but American consumers are not inclined to pay extra for something to be made domestically (and many can’t afford that privilege anyways). So, the markets for semiconductors are still global and interconnected.
When you think about the semiconductor industry 10 years from now, what makes you the most optimistic?
What I am most optimistic about is the creativity and ingenuity being applied not only to the manufacture of semiconductors, but to how they are used. I continue to be amazed at some of the feats of atomic-scale engineering: we are literally making devices that are a few atoms in size or making structures at a level of complexity that is hard to really appreciate.
What gives you pause about the future of semiconductor innovation?
The main thing that gives me pause is that the cost to design and deliver the most advanced chips has gotten to the point that it risks becoming inaccessible to all but the best-funded firms, companies like Nvidia, Apple, Google, or Amazon. If we could use machine learning and modeling to accelerate moving down the learning curve or reduce the minimum efficient scale, that would be really exciting.
