Chip War

Chip War by Chirs Miller is a thorough history of the semiconductor industry since the mid-1950s. The relationship between business and government has shaped the technological space into what it is today.  In an industry where the U.S. has repeatedly failed to maintain dominance, it now faces its greatest challenger.

The transistor was invented in 1947 in Bell Labs by John Bardeen & Walter Brattain based on a theory proposed by William Shockley. Shortly after, Shockley refined the idea to create a switch that could replace vacuum tubes that were notorious for their large size and failure rate at the time. A little over ten years later, Jack Kilby and Bob Noyce invented the integrated circuit independent of each other which allowed for multiple transistors to be put on the same piece of silicon. These circuits became known as semiconductors/chips.

Fairchild R&D, founded by 8 former employees of Shockley including Noyce and Gordon Moore, was one of the first entrants to this new semiconductor market which was fueled by one buyer, NASA. The space race funded the early development of the semiconductor market. The government also played an important role, looking to create smarter weapons. A process called photolithography was created by Jay Lathrop and James Nall which used light and photo chemicals to produce smaller semiconductors. This process was continually refined over time to produce chips for the government. As firms advanced, they looked for ways to expand their market share. The government could only take them so far. They needed a way to reach consumers.

The Cold War brought spies to America looking for more than military secrets. Soviet physicists and engineers were capable but lacked knowledge of the semiconductor industry. At this time, the country was restricted from importing materials from many countries as there were many restrictions around communist-led countries. The Soviets planned to steal American tech and reproduce it exactly the same. This plan of course failed because by the time they could reproduce the technology, they would be five years behind what America was producing because of how quickly the tech evolved. Even though the USSR had more nuclear warheads, America won because it had the better technology. Gorbachev needed to end the war to strengthen their tech.

After WWII, Japan rose to power in the semiconductor space with Sony. Japan was restricted from developing a military, so they developed a tech sector. The U.S. and Japan became dependent on each other, the U.S. would design chips and Japan would manufacture them. This production globalization started to occur in other places with cheap labor and no unions like Hong Kong, Taiwan, Malaysia, Singapore, and South Korea. The U.S. was very supportive of Japan as it did not want them to start another war. By the 80s, Japan was outperforming American chipmakers with new firms like Toshiba, Fujitsu, and NEC dominating the market alongside Sony.

Noyce and Moore eventually left Fairchild to start Intel because Fairchild refused to give equity to its employees. Intel originally produced DRAM but as Japan became more efficient in the space, it was harder to remain competitive. Intel decided to pivot to making microprocessors which were general-use semiconductors. In 1970 the company disrupted the market and released the first microprocessor.

Around this time, Japanese semiconductors were still seen as inferior to U.S. ones. There was a belief that Japan would just rip off American innovations even though there was evidence that they had their own advancements. The FBI caught Hitachi for spying which only strengthened this mentality. This forced people not to think of the country as a threat. However, they kept growing and dominated the DRAM space. This dominance was helped by the fact that Japanese companies only used chips produced in Japan while the U.S. would manufacture in Japan as well as other countries. Japan’s government also helped the industry by giving loans so they could sell cutting-edge chips for very low prices which led to firms being unprofitable. It wasn’t until Japan’s economy tanked in the 1990s that the sector lost its footing. Japan also missed out on microprocessors.

In 1981, IBM released the first personal computer. Inside was an Intel microprocessor. This disruptive technology was ignored by most chip manufacturers for the reasons outlined in Christensen’s The Innovator’s Dilemma.

TSMC was founded in 1987 as a manufacturer of chips for fabless firms. Just as the cloud made computing more accessible, TSMC made chip production accessible. Companies no longer needed to invest hundreds of millions of dollars into setting up fabs. With help from the Taiwanese government, the company skyrocketed. China, which had fallen behind in this space because of Mao’s focus on steel production, later adopted this business model with SMIC. Both of these companies performed well in part due to the emergence of the cell phone and other portable technology.

As fab technology evolved, only one firm became able to produce the systems needed. ASML is a Dutch company that produced the EUV systems for more advanced chips in the mid-2000s. They continue to be the only company capable of building the EUV machines used in the industry. The challenge is as much a supply chain problem as an engineering one as these Extreme Ultraviolet systems contain almost a quarter million parts.

For the same reason Intel succeeded, transitioning to microprocessors, it started to fall behind. Sticking to the x86 architecture of its chips, Intel ignored ARM. ARM architecture became a big player when phones became popular. They were victims, like Japan, of the Innovator’s Dilemma. Companies like Apple produced their chips in Taiwan with TSMC. Intel then lost even more market share when it was found that Nvidia, which was able to enter the market due to not needing its own fabs, produced chips that were better with AI.

Xi Jinping, the President of the People's Republic of China, saw that China was dependent on countries like the U.S. when it came to what he called “core technologies.” He called for technological advancements in China to reduce dependence on other countries. It’s believed that it is impossible to achieve complete independence in this industry and that the goal is to have China be dependent on the least number of countries possible. China also wants to be more technologically advanced than other countries it might go to war with. Just as the U.S. won the Cold War through tech, it is speculated that the next war will be decided by who is more advanced in AI. The country plans to fund companies that bring them closer to their goal of technological independence and superiority.

Huawei saw a lot of success in the communications space and was leading 5G until security concerns were raised. The U.S. revoked their ability to use necessary tools produced in the U.S. which led to them being let go by most countries using their services.

China founded Jinhua in 2016 to create memory and DRAM and planned to learn from UMC. UMC did not know how to produce DRAM, so it stole this information from Micron. After Micron tried to act, it was countersued and banned from the Chinese market. The U.S. responded to this by banning exports of Jinhua chips to the U.S. which caused the firm to shut down.

The U.S. exercising these choke points on China has come to be known as “weaponized interdependence.” China is unable to retaliate because it is dependent on other countries like the U.S. This is also the reason they cannot try to take over Taiwan. Taiwan is too necessary for so many countries' economies, including China’s, that nobody can act upon it.

It was interesting reading about the history of semiconductors. I didn’t realize how much political decisions impacted technological development. If the U.S. had subsidized more chip fabricators in the U.S., we might not be dependent on other countries for our military and able to better restrict overseas development of the technology that could make us lose the AI war. Globalization is great when it comes to sourcing plastics and clothes but when it comes to technology, it’s scary thinking how dependent we are. If one country were to take advantage of its footing, it could be very unpleasant for a lot of people.