The United States has lost two industries of critical strategic importance: semiconductors and communications-infrastructure equipment. Only about 9% of the world’s semiconductor devices are currently produced in the US and no significant telecom-equipment manufacturers remain in that country.
Different reasons are responsible for these losses, but government support will be required in both cases if the United States is to keep its industrial competitiveness.
Since these industries were largely pioneered in the US, we should examine how this happened, and ask further how valid are concerns about America’s apparent loss of technological leadership.
The semiconductor industry
The invention of the transistor at AT&T Bell Labs in 1946 launched the digital age by replacing vacuum tubes with solid-state devices. Because the patents were broadly licensed, major vacuum-tube manufacturers like RCA and Sylvania became early transistor manufacturers. Others followed.
Manufacturing transistors required a new production technology, which each market entrant jealously guarded as a competitive asset. Early applications replaced vacuum tubes in radios and in computers that were emerging as mass-market products.
Building computers required assembling discrete transistors for specific computational requirements. This was a costly process that limited the number of commercial applications.
The first such computers had military applications. The Burroughs Corporation built one of the biggest computers by assembling germanium transistors, for control of the Atlas ballistic missile in the 1950s.
Intel’s invention of integrated-circuit chips in the late 1960s greatly increased transistor applications and gave us the foundation of the general computer industry. Integrated circuits combined on one silicon chip the size of a thumbnail interconnected transistors, forming digital processing circuit.
These circuits today enable practically all electronic products. Integrated circuits dramatically reduced the cost of computing systems, because expensive mechanical transistor interconnections were eliminated.
Because the processing capability of a chip is related to the number of transistors at work and their switching speed, increasing transistor density became the key technology driver. The race was on to increase the number of transistors on a silicon chip to enable ever more processing power at lower cost.
This was accomplished over decades by the development of ever more complex technologies that gradually reduced transistor size. Over a period of 50 years, the density of transistors per unit area on silicon chips doubled every two years, following Moore’s law. Transistor density went from a few dozen to several billion on a single chip.
But these innovations came at great cost in research and development and capital equipment. The consequence of this transistor shrinkage was that the production technology became ever more sophisticated, requiring ever more costly production equipment.
For example, the most advanced lithographic equipment to pattern transistors cost US$2 million in 1985. In 2020 the cost was $160 million. In 1985, it was possible to build a viable integrated-circuit production plant for about $200 million. In 2021 the cost of a state-of-the-art factory is more than $10 billion, and could be double that.
As a result, innovative new products could not be brought to market by new companies unless they had access to very large sums for manufacturing facilities – beyond the capabilities of venture capital.
That is why industry dynamics changed in a fundamental way, as standard industry production processes emerged in the 1980s, allowing chip designs to be separated from proprietary production processes. Designers gradually adopted rules that enabled production in plants equipped to process chips designed along those rules.
What also emerged was specialized companies prepared to manufacture such chips on contract. Taiwan Semiconductor Manufacturing Corporation, founded in 1987, has emerged as the leading company solely focused on manufacturing the chip designs of others. Over time its production technology has improved to the point that today it among the three leaders in the world, along with Samsung and Intel.
Thus innovative companies focused on product design and marketing, while outsourcing production to TSMC secure in the knowledge that their proprietary designed products are protected. This greatly reduced the cost of entry to the industry, and dozens of new companies were founded with venture capital, including several funded by Warburg Pincus.
The emergence of the “chip foundry” model gave huge impetus to innovation, and many great new companies were founded, mostly in the US. New products could benefit from the most advanced production technology.
So while manufacturing was not in the US, the value created domestically by new fabless companies was immense, and this kept the most advanced design skills in the US. Looking at production volume, more than 70% 0f the world’s advanced-logic chip production is in Taiwan in 2021.
US companies still made commitments to internal chip fabrication, not to invest in new technology, but rather to reduce the cost of plants and their operations. Hence they searched for international locations willing to subsidize the cost of new plants.
Intel built plants in Asia and Israel, for example. Other companies followed a similar strategy. That explains the reduction of US semiconductor production while domestic design activity continued to grow. Domestic production dropped while corporate revenues continued to grow profitably.
For the US communications-equipment industry, the story is different.
AT&T, while it was a monopoly, was a world-leading pioneer of satellite, fiber-optic and wireless communications. To ensure competition, the US government mandated the breakup of AT&T in 1984 into service companies and equipment suppliers.
This eventually destroyed Lucent, the equipment supplier that was the successor to Western Electric, the Bell System’s manufacturing arm. It lost its exclusive domestic customer base and for the first time had to compete both domestically and internationally without the immense resources it needed.
The breakup also spelled the eventual demise of Bell Labs, then the world’s largest and best communications laboratory, which had lost the guaranteed funding it had enjoyed while AT&T was a monopoly. It was that guaranteed funding that allowed basic and applied research to be conducted on the long time scale required for communications technology.
Combined, these two effects left the Regional Bell Operating Companies to find their own technology strategies as foreign suppliers came in. Lucent did not have the massive resources to fund Bell Labs to research and develop an innovative competitive international market position as the Internet and growing wireless services revolutionized the industry requirements.
A series of acquisitions and mergers did not help. Lucent is now part of Nokia, a Finnish company that is a relatively modest competitor to the two largest communications-equipment makers, Huawei and Ericsson.
The long-term picture
The next question is: Is the loss of semiconductor fabrication and communications infrastructure worrisome for the US economy in the long term? The answer is yes.
As David Goldman and I argued in 2018, semiconductors are the key enablers of practically all systems. At the current pace, the US government will have to rely on imported products for key defense requirements for high-performance devices. This is not an acceptable situation for a major power.
As we discussed in the article linked above, the move to offshore production was driven by capital-requirement decisions of the industry because the technology was largely US-originated. This strategic need has now been addressed with a $50 billion US government fund available to the industry for production capital support.
Furthermore, programs have been put in place to attract new plants to the US. For example, TSMC has announced a new plant to be built in Arizona, and Samsung may follow with a new US plant, while Intel will upgrade its plants.
It has become widely recognized that communications infrastructure is a key component of the economic system. It is noted that fifth-generation (5G) wireless technology is becoming a core enabler of new services, but is controlled outside the United States. US companies have no role in its evolution, nor in what will become 6G some years hence. It is recognized that rebuilding the industry is not practical. The technology is largely overseas.
To establish a new open-market technology impetus, a large number of corporations, with support of the US government, have established an organization focused on developing a sophisticated suite of open-sourced software to operate networks using generic equipment instead of the proprietary equipment offered by the current suppliers. It is called the Open Radio Access Network Policy Coalition.
If successful, the O-RAN program will allow network operators to build their systems using equipment from many vendors and with new software, and tailor their systems to suit their requirements rather than be restricted by the proprietary solutions now offered by Huawei and Ericsson. O-RAN has so far gotten off to a slow start, and it will take some years to determine its success in changing the industry’s competitive condition.
In conclusion, what has been learned from the history of the semiconductor and communications-equipment industries is that free market capitalism sometimes needs government support when it comes to strategic industries with immense capital needs. Making financial returns the priority in order to satisfy investors may not be in America’s best long-term interest.
Henry Kressel is a technologist, inventor and long-term Warburg Pincus private equity investor. Among his technological achievements is the pioneering of the modern semiconductor laser device that enables modern communications systems.