just asking questions

Why Congress Just Joined the Semiconductor Arms Race

Photo: Michael Short/Bloomberg via Getty Images

While moderate Democrats play a game of highly publicized footsie over historic climate legislation, lawmakers just quietly passed a massive piece of funding with significant support on both sides of the aisle. This week, the Senate and the House both green-lit the CHIPS Act, a $280 billion investment in scientific research in the U.S, including a $52.7 billion injection into the domestic manufacturing of semiconductors.

CHIPS is huge. By comparison, the current climate and energy bill approved by Senate-agenda emperor Joe Manchin — the largest climate action in congressional history — is $369 billion. So how did Democratic leaders get over a dozen GOP senators to agree to the huge spending push? First, it was championed by Republican Todd Young. Second, it has been billed by supporters as a national-security measure and a bulwark against China, a semiconductor manufacturing giant that is creeping toward potential conflict with Taiwan, another semiconductor hub. In a statement this week, President Joe Biden said the bill “will mean more resilient American supply chains so we are never so reliant on foreign countries for the critical technologies that we need for American consumers and national security.”

But what exactly are these crucial pieces of technology that have inspired bipartisan unity when virtually nothing else can? To better understand the importance of semiconductors and the world of electrical engineering, I spoke with MIT professor Jesús del Alamo about how semiconductors work and why the CHIPS Act could impact our lives in the coming decade.

So what exactly are semiconductors?
Semiconductors are a versatile class of materials with really fascinating properties that make it possible to make computer chips, light-emitting diodes, lasers, solar cells, etc. The ones we hear about most — that are in computer chips, microchips, and many of the chips we use in gadgets around us every day — are made of silicon. We’re talking about a really tiny square centimeter slab of silicon with millions, sometimes billions, of transistors in them that perform computations, signal processing, can hold memory, etc. They’re in all sorts of gadgets all around us: cars, lightbulbs, electric toothbrushes, weapons, aircraft. They’re everywhere.

I think a lot of people were surprised by the size of this bill considering that the source and supply of semiconductors wasn’t something the public thought about before pandemic shortages. How important is this bill to the industry in the U.S.?
The bill’s emphasis is on what we call leading-edge semiconductors because they enable the highest density chips and highest energy efficiency computation. They are extremely important in phones and computers and in military applications.

The leading edge is so important because we’re in a game in which, essentially, winners take all. Semiconductor-chip companies turning out leading-edge technology reap the bulk of the profits from that technology. Typically, the most advanced technology, when it comes into the market, enables a whole new class of applications and great gains in energy performance and efficiency in applications that already exist. So this is extremely attractive for computer-makers and manufacturers of very advanced products with very large markets to move the products over to the latest technology because they can deliver more to the customer at many times a lower price. Being on the leading edge first, deploying a new technology first, is really critical to really get significant profits.

And what is the implication of that? You can plow those profits back to foster innovation and make sure you remain in first for the next wave of technology. Right now, we have lost that. The big profits and hot products are being made on technologies that are fabricated in Taiwan and Korea.

Is the story of how the U.S. lost the lead in semiconductor manufacturing part of the larger story of manufacturing moving abroad, or is there something else going on here?
For a long time in Asia, countries have seen the strategic value of these technologies, and they have been supporting their companies in R&D and setting up manufacturing facilities on their own soil. The U.S. has done none of that, and that’s what needs to be corrected. We need to match the incentives that Taiwan, South Korea, and China are offering to their manufacturers to set up plants onshore to get to the leading edge and stay there.

These advanced chips are extremely difficult to manufacture. Why is that?
The technology is unbelievably complex. It takes many many years of research and development to bring a new technology to the marketplace. If you look at the numbers that companies are investing in Arizona, Texas, and Ohio, we’re talking about manufacturing facilities that are way in excess of $10 billion. One facility, as much as $17 billion. Imagine the bet it represents for a company to decide to set up a new plant. This is a gigantic bet that could sink the company if you make the wrong decision. These are difficult decisions that require you to be gutsy and to be backed up by your engineers.

This has been described as a national-security bill. Does that sound right to you?
The most advanced technology, roughly 80 to 90 percent of those chips, comes from Taiwan. Some of that technology is used in Defense, and some of it permeates U.S. society. So even if it doesn’t have military use, it’s extremely important because our whole country could come to a halt. It is a huge vulnerability when we’ve seen the supply issues caused by the invasion of Ukraine. When we contemplate what could happen in Taiwan, it’s extremely scary, and it’s very important that we bring as much semiconductor manufacturing to the U.S. as possible.

What sort of exciting research is out there right now that this bill could help support?
One area is artificial intelligence, which is solving superhuman problems and permeating all kinds of applications. Because of this explosion of AI, there is now a big effort to develop unique technology to perform AI functions more efficiently.

The second big revolutionary technology that is emerging is something called chiplet integration. If you look at an electronic board, you’ll see many chips laid out on them or connected together somehow. It is pretty much spread out and is relatively large and the chips are relatively far from each other and are connected through conducting lines that are slow and power hungry. This new chiplet-integration approach is a way to assemble a system with individual chips in an extremely compact way with very high energy efficiency. This is going to allow very large and complex systems to be way cheaper to design and make than before.

But it takes about ten years in microelectronics for a new technology to hit the marketplace. So if we want, ten years from now, to continue to improve on performance, we need to start developing new technology today.

From that decade-out perspective, how does this bill look?
This is going to help a lot. But it’s not going to be a panacea. This is a first and necessary step to change the course of things. In other words, instead of decaying in terms of the presence of the United States in the manufacturing of leading-edge products, this is going to change the trends. But it really needs to be sustained. Most other countries have been sustaining their semiconductor manufacturing for a long time, and the U.S. has to be willing to do the same. This is a first step, but we hope that as we see the returns of this investment over the next five years become clear, the government will realize they have to stay put. This will continue to be a set of technologies to foster and nurture, with a skilled workforce, new technologies and critical start-ups coming out of university labs. We need to support this system because the rest of the world is doing the same.

This interview has been edited for clarity and length.

Why Congress Just Joined the Semiconductor Arms Race