Semiconductors are vital to the modern economy, powering everything from video games and cars to supercomputers and weapons systems. The Biden administration is investing $39 billion to help companies build more factories in the United States to bring more of this supply chain back home.
But even after U.S. facilities are built, chip manufacturing will remain decidedly global.
The international journey of one kind of chip, made by the American semiconductor manufacturer Onsemi and used to power electric vehicles, demonstrates how difficult it will be to decouple from East Asia and other regions that dominate the chip market.
The first steps for making this particular semiconductor, known as a silicon carbide chip, happen in a factory in New Hampshire. The chip ends up in cars driven on American roads and elsewhere. But in between, the process will depend on raw materials, machinery and intellectual property from dozens of foreign suppliers and factories.
The first step begins inside Onsemi’s New Hampshire plant, with a jet black powder of silicon and carbon from Norway, Germany and Taiwan. The powder is added to graphite and gases that come from the United States, Germany and Japan, then heated to a temperature close to that of the sun, producing a crystal that will form the backbone of millions of chips.
This crystal, which is almost as hard as a diamond, is sent to a factory in the Czech Republic to be sliced into thin wafers using special machinery from the United States, Germany, Italy and Japan.
The wafers are shipped to an ultraclean factory in South Korea, where mechanized pails carry them between complex machines from the Netherlands, the United States and Japan. The machines use chemicals, gases and intricate patterns of light to create channels just a few atoms wide for electrons to move through as they convey information.
The wafer is then cut into tiny chiplets, which travel to facilities in China, Malaysia and Vietnam for finishing touches and testing. Then the chips head to global distribution centers in China and Singapore.
Finally, the chips are sent to Hyundai, BMW and other automakers in Asia and Europe, which put them into power systems for electric vehicles. Other chips are sold to auto parts suppliers in Canada, China and the United States.
The first computer chips were invented in the United States, but by the late 1960s parts of the supply chain began to move overseas as companies looked to save costs. With the help of generous subsidies, Asian companies eventually began to manufacture chips that were cheaper and more advanced than those made in the West.
America’s share of world chip manufacturing has fallen to just 12 percent today from 37 percent in 1990, according to industry figures.
The United States is trying to win back more chip production to make its supply chains more resilient, and avoid the kind of expensive and economically damaging shortages of semiconductors seen during the pandemic. But with other countries also continuing to spend heavily on their chip industries, American investments — as big as they are — will go only so far to change the global picture.
One 2020 study by the Boston Consulting Group and the Semiconductor Industry Association estimated that an infusion of $50 billion would increase the American share of manufacturing to 13 or 14 percent by 2030, helping the United States to hold on to at least a portion of the global market. Without the funding, the U.S. share would fall to 10 percent, the study said.
For the most cutting-edge chips, including those that are helping to power a boom in artificial intelligence, U.S. officials now say that new investments will put the country on track to produce roughly 20 percent of the world’s leading-edge logic chips by the end of the decade.
Still, chip and electronics production is likely to be centered in Asia for the foreseeable future, Moody’s Analytics said in a recent report. Tech companies are under intense competitive pressure to keep costs down while innovating, meaning they are driven to work with most skilled manufacturers in Asia, said Chance Finley, Onsemi’s vice president of global supply chains.
The incredible cost of chip manufacturing facilities — which range from $5 billion to $20 billion to build, more than a nuclear power plant — encourages domestic chipmakers to outsource manufacturing to foreign facilities rather than build their own, he said. Chips also happen to be small and light, which makes them easy to move around the world.
Onsemi is looking to the new U.S. investments in the chip industry to help it grow, and yet considering sites in the United States, the Czech Republic and in South Korea for a $2 billion expansion.
The Onsemi factory in Bucheon, South Korea.
Jun Michael Park for The New York Times
Many stages of Onsemi’s manufacturing process are done in house. That is somewhat unusual for chip companies, which often outsource certain production stages. Other chip supply chains are different, but no less international. Many run through Taiwan, which produces more than 60 percent of the world’s chips, and more than 90 percent of the most advanced chips.
One 2020 study by the Global Semiconductor Alliance and Accenture found that chips and their components could cross international borders 70 times or more before reaching the final consumer, traveling more than 25,000 miles in the process.
Another study by the Boston Consulting Group and the Semiconductor Industry Association looked at creating a self-sufficient chip supply chain in the United States, and estimated it would take $1 trillion and sharply increase prices for chips and products made with them.
“The idea that we’ll be somehow self-sufficient, that is not realistic,” said Bindiya Vakil, the chief executive of Resilinc, which maps supply chains for the semiconductor and other industries. “We are part of this global supply chain, whether we like it or not.”
