Semiconductors: The US should be asking how we can lead
Amid a global semiconductor shortage impacting everything from appliances to smartphones, lawmakers will consider how the U.S. can gain ground. They shouldn't stop there.
The House Committee on Science, Space and Technology will hear from Intel, Micron Technology, the Lawrence Berkeley National Laboratory and Purdue University on Thursday. As the U.S. Innovation and Competition Act makes its way through Congress, the main focus is on catching up with Asian semiconductor competition. Most of the investments being talked about exploit existing technologies and knowledge, like building more fabrication plants. These have short time horizons and lower risk.
The committee should ask how do we uncover new materials, devices and manufacturing methods that will bring us the next generation of performance and energy efficiency, with radically different device structures or process technologies. This means embracing risk, fostering competition of ideas, and enabling the best new ones to get a toehold. It means adopting disruptive approaches when the future semiconductor trajectory requires ground-breaking innovations.
Why more risk-taking? Because the industry today has every incentive not to take bold risks. The costs of making a wrong technology choice are enormous. Financial pressures foster incremental steps that exploit existing know-how, rather than exploration of riskier alternatives. Rather than allocate constrained fabrication capacity to new materials or processes, it is easier to feed skyrocketing demand with tried-and-true processes. Why take risks on the unproven?
Stanford Professor James March called out this dilemma four decades ago in one of the seminal papers on innovation when he compared "exploration and exploitation." Exploration is about new possibilities, new ideas and ways of doing things, while exploitation is about established certainties - refining existing products and processes. Because the returns from exploration tend to be long-term and have a high amount of risk, while the returns from exploitation are shorter term and likely more predictable, exploitation tends to crowd out exploration.
In semiconductors, we should not let catching up (a strategy used by many developing countries) distract us from our real goal, leading the way. Building onshore fabrication plants is necessary, but not sufficient. The learning that comes from operating them is essential to ensure that new ideas are not un-manufacturable pipe dreams. But a new fabrication plant will become obsolete in four years if there is no R&D to sustain its relevancy. Manufacturing and R&D go hand in hand, in the same way that doing R&D without manufacturing is akin to building a bridge to nowhere. Exploitation of established certainties is not enough.
We can secure American leadership by fostering bold exploration and risk-taking. How do we get new devices ideas like magnetic memories (MRAMs) into high-volume production? Or create new materials and processing methods? Here are some ideas:
Foster competition and a marketplace of ideas
The Defense Advanced Research Projects Agency (DARPA) has done this successfully for decades, with an idea-driven and outcome-oriented culture. It seeks out and funds researchers who compete with each other to bring new concepts forward. DARPA program managers feed competing approaches and are adroit and tough minded in their selections. And it works, DARPA invented the digital protocols that made the internet possible.
Reduce the formidable barriers to commercializing new process technologies in the U.S.
Replicating a DARPA-like process is not enough, we need to commercialize and manufacture new innovations in volume. We could make domestic facilities available to test commercial viability by buying space, capacity and capability in existing or planned fabrication plants. Startups are often times stymied by a chicken-and-egg problem. Without prototyping capacity they struggle to demonstrate commercial viability. Without demonstrated viability, they can't get into fabs.
Underwriting some of the risk of technology development
The federal government has done this in other areas, notably in NASA's Commercial Crew and Aircraft Energy Efficiency programs, and in underwriting of clinical trials and early stage manufacturing for COVID-19 vaccines not knowing whether a vaccine would gain emergency use authorization. They bet on a range of approaches, recognizing that many would not bear fruit.
Seek leaders who are not "captured" by incumbency
If we create a new bureaucracy to coordinate these investments, let's make sure it is not bound to the status quo. This means leaders who are willing to take risks, and ensuring that "accountability" doesn't lead to risk aversion. Many R&D ideas don't pan out. Success comes from having a smart portfolio of bets.
We should make the U.S. the place people and companies come to try new ideas. We should welcome immigrants - would we give a visa these days to the next Andy Grove of Intel, from Hungary, or Satya Nadella of Microsoft, from India? American universities and research institutes attract the best and the brightest. We should encourage more people to come, not stymie those who might bring us future capabilities. We should make the U.S. a magnet for the best people and ideas.
The committee should remember that these things are historically what the U.S. does best. Let's play to our strengths.
Edlyn Levine, Ph.D., is a research associate with the Department of Physics, Harvard University.
Willy Shih is the Robert & Jane Cizik professor of management practice at the Harvard Business School.
H.-S. Philip Wong is the Willard R. and Inez Kerr Bell professor in the School of Engineering, Stanford University.
