The Demise of Photolithography in the United States

Photolithography for semiconductor manufacturing was invented in the United States, with the first transistor manufactured by lithography demonstrated in 1958 and a U.S. patent granted on the process in 1959. As integrated circuits became central to the electronics industry in the 1960s and 1970s, advances in photolithography drove the scaling of device geometries to ever smaller sizes, reducing cost, increasing speed, and cutting power consumption and heat dissipation. By the mid 1980s, two U.S. companies, GCA (formerly Geophysics Corporation of America) and Perkin-Elmer, dominated the industry worldwide. Reliability problems with their products and lack of responsiveness to customer complaints and suggestions for improvements by GCA, the largest vendor, led Japanese semiconductor manufacturer Toshiba to approach Nikon to develop alternative photolithography gear. By 1984, both Nikon and Canon were producing state of the art photolithography machines, with GCA’s revenue falling 50% in 1985 due to competition from Japan, laying off 70% of their workforce over the next two years (which were boom years for the semiconductor business).

So began the collapse of the U.S. photolithography industry, with consolidation, “help” from the U.S. government and industry consortia, and the U.S. toxic management culture obsessed with quarterly earnings results, inclined to cut research and development when times are good, and to treat engineers and the know-how they possess as disposable assets during cyclical downturns, all contributing to the demise. Today, ASML in the Netherlands and Japanese companies Canon and Nikon account for around 98% of the global photolithography market.

Politicians and technology strategists who speak glibly about “on-shoring” the fundamental fabrication processes that underlie most of present-day technology and are essential to what they ironically call “national security” (in a country where millions of military-age illiterate male savages are pouring across its undefended border every year) would do well to examine precisely why these technologies, invented in the U.S., which once dominated their design and manufacture, went offshore in the first place.


—perhaps on-slushing better describes the political mindset. (expanding slush fund accounts).
We shall see how this pans out. It seems to be better than FDR make work projects …


ASML has about a 90% market share in photolithography machines and a 100% market share in the most advanced form of photolithography, extreme ultraviolet lithography.

Advancing this tradition is very expensive. ASML employs over 14,000 people in its R&D departments, out of over 39,000 employees worldwide, and spends almost $1 billion on R&D every quarter.

In 2022, ASML operated at a higher gross profit margin than Apple, despite the latter’s pricing power due to its luxury branding. Smartphone manufacturers and cutting-edge AI companies alike are ultimately dependent on the semiconductor industry and its key suppliers.


It seems to me this represents an extreme strategic vulnerability - a single point of failure - which makes the effects of the NordStream sabotage puny by comparison.


Yes. Do you remember this story about the car chip shortage because of a March '21 fire in an obscure Japanese company?

Here’s the impact on car prices:

This reminds me of Isaac Asimov’s depiction of the planet Trantor in his “Foundation” series illustrates a highly efficient but fragile society. Trantor is a planet entirely covered by a vast city, a pinnacle of technological advancement and bureaucratic organization within the Galactic Empire.

Asimov describes Asimov’s Foundation series - using Trantor as an example of an economy that has become overly dependent on centralized systems for its survival. The planet’s efficiency is derived from its intricate and interconnected networks that manage everything from food production to governance. However, this efficiency comes at a cost. Trantor’s reliance on these centralized systems makes it vulnerable to disruptions. Any disturbance, whether it’s a breakdown in technology or an external threat, could cause widespread chaos and potentially catastrophic consequences due to the planet’s lack of redundancy and resilience.

Also, XKCD:


Another example is Larry Niven’s Ringworld, where the technological civilisation of the mysterious Engineers who built the megastructure was based upon a room temperature superconductor they invented. Somehow, a mold infected the Ringworld which destroyed the superconductor, and with it the machines which allowed the Engineers to make the superconductor, so they were unable to reboot their civilisation.

I’ve heard that Niven got this idea from the “purple plague” that mysteriously destroyed some semiconductor devices in the late 1960s, while he was writing the novel. He imagined that once industry became dependent upon semiconductors to make semiconductors (as has happened, fifty years later), a plague that destroyed semiconductors could cause complete collapse of technological civilisation. The purple plague turned out to be an unexpected metallurgical phenomenon that occurred when gold was bonded to aluminium, but the idea of a semiconductor destroying organism has popped up in science fiction from time to time ever since.


And on this matter, it’s worth noticing that the father of the Euro, Bernard Lietaer, worked on complementary currencies:

Basically, for any complex to be sustainable needs to have a balance between two factors: resilience and efficiency. These two factors can be calculated from the structure of the network that is involved in a complex system. A resilient, efficient system needs to be diverse and interconnected. On the other hand, diversity and interconnectivity decrease efficiency. Therefore, the key is an appropriate balance between efficiency and resilience.


Another example of “plague” appears in plumbing:


I’ve been in the same house (an 1899 hewn log farmhouse) for 45 years, and seen too many problems caused by incompatible plumbing materials. Some inherited, some done by hired help, and a few of my own screwups before I knew better