The Silicon in Silicon Valley, Again

From responding to a pandemic — with novel mRNA vaccines — to building the future — with SpaceX’s rockets and Apple’s new chips — we are seeing that the material world still has the power to thrill.

But even though many appreciate physical products more than ever, this esteem has not yet prompted a broad attitude shift to recognize manufacturing as a valuable competency in the world of technology. The U.S. president and Congress are demanding that America rebuild its manufacturing capacity. Federal incentives are helpful, but government effort alone will only go so far. Success should be defined not by creating spare capacity, but by revitalizing American manufacturing capabilities. That effort will demand active participation from the tech community, which should stop dismissing manufacturing work as a “commoditized” activity not worth smart minds.

Software is fashionable, but we need more than software to reshape our physical world. Software alone will not construct manned bases on Mars. Software alone will not decarbonize the atmosphere. And software alone will not make our cities gleaming wonders. America needs to be a place that builds. Manufacturing capacity isn’t merely a nice-to-have to respond to emergencies, but is the key to realizing a technologically-accelerating civilization. The Bay Area once led in manufacturing — it was where transistors were first produced at scale, spawning the computing industry — and can do so again. Let’s once more prioritize the “silicon” in Silicon Valley. To do so we have to prize manufacturing as more meaningful work.

Technology is process knowledge

At first glance, U.S. manufacturing is in good health: real output is close to all-time highs, after all. And America had tremendous success scaling up production of mRNA vaccines, due to a massive effort by the scientific community and strong government support on both purchasing as well as production. 

But the vaccine success was an exception to a broad stagnation in manufacturing capabilities. Over the last few years, smaller companies have struggled to quickly produce relatively simple goods, larger companies are failing to master technologies that other countries have mastered, and construction of public infrastructure is regularly subject to outrageous cost overruns. It’s not just that manufacturers couldn’t produce items as uncomplicated as masks and cotton swabs early on in the pandemic. Manufacturing leaders like Intel and Boeing have also fallen on hard times: The former has faced persistent difficulties migrating to more advanced process nodes, while the latter’s 737 MAX has laid bare a challenged engineering culture.

We need to acknowledge that something fundamental in American production culture has broken. A steady deindustrialization is one reason that America can no longer build. The manufacturing workforce today has shrunk by 40% from its peak in 1980. Only around 10% of integrated circuits are manufactured on American shores. Some of the declines are due to automation and technology upgrading. But American businesses also embraced offshoring manufacturing with a greater zeal than those in other industrialized countries. While manufacturing as a share of U.S. GDP has shrunk from 25% in the 1950s to just 11% in 2020, that share in Germany, Japan, and South Korea has been basically stable around 20% in the last few decades.

To re-industrialize America, we have to start taking a more nuanced view of what technology consists of. Technology should be broken down into three components: tools (like pots, pans, and the stove); explicit instructions (like recipes); and most importantly: process knowledge. That’s the earned experience of everything that can’t be written down. Everyone knows that it’s impossible to write a memo detailed enough on all aspects of one’s job. It should be obvious that a person with no cooking experience would struggle to do something as simple as frying an egg, even given a marvelously equipped kitchen and exquisitely detailed recipes. As I’ve written previously, that know-how is everything. 

If all patents and blueprints disappeared tomorrow, we are still an advanced industrial civilization, because trained engineers are ultimately the source of technology. But if the people disappear, then no amount of detailed instructions could save us, just as a pharaoh of the past would not be able to build a modern car even if he were given the most intricate blueprint. Instead of viewing intellectual property and tools as the ultimate ends of technological progress, we should see them instead as the fruits in the training of better scientists, engineers, and technicians.

The loss of eight million manufacturing workers since 1980 has disintegrated process knowledge in industry. Closing down a factory leads to a cascade of loss. A factory is an establishment for training many skills: It’s not just line workers, but also machinists who improve tools, modelers who determine resource allocation, and industrial designers of new processes.

This point isn’t an argument that everything can or should be “made in America.” It has made good commercial logic for U.S. companies to offshore their manufacturing. In particular, it has been fantastically profitable for Apple to focus on design while having its phones assembled in China, and for Facebook and Google to abstract away the physical world and mostly work in the digital realm.

But a focus on the most profitable slice of the value chain has resulted in the broad neglect of manufacturing skills. It’s not just that the number of manufacturing workers in the U.S. has collapsed (which means the loss of a great deal of industrial process knowledge) but fresh grads are finding it much more exciting to work in finance or software, both of which scale beautifully, instead of improving industrial processes. The costs of that decay and lack of new entry are now being painfully felt, for both small and large American manufacturers.

As Germany, Japan, and South Korea have shown, it is possible to maintain a steady share of manufacturing in GDP. I concede however that they have also shed manufacturing workers while their manufacturing output has grown. In my view, however, these countries have still done something right: They responded to globalization by moving up the manufacturing value chain, while the U.S. manufacturing base responded mostly by abandoning production. What these countries grasp is that technology is ultimately about process knowledge. They have not had a dismissive attitude that manufacturing is “commoditized” and thus should be shipped to China. Instead, they show that knowledge is something that needs to be practiced for it to be maintained even at its current level. That has meant maintaining a domestic ecosystem of talent that is able to keep working out how to do new manufacturing work in better ways.

A technologically-accelerating civilization

Emphasizing a greater role for manufacturing is not just about the capacity to quickly respond in emergencies or about restoring the economic health of industrial leaders. It’s the key to making sure that America can still do big things. The Bay Area is a tight-knit network where universities, investors, talented managers, and a large pool of engineers sit next to each other. Imagine how much stronger it would be if it were also a place where technologists have easy access to manufacturing facilities to quickly produce new goods. Historians and experts on innovation clusters know that a concentration of economic linkages is part of the magic that makes an ecosystem work. It not only worked for chips — giving Silicon Valley its name — but also for automobiles, biotech, space, and more.

In the case of modern manufacturing, innovations in software and communications technology have made it easier to locate hardware far away. But by separating design and manufacturing, we’ve loosened the feedback loops between the two. It’s no longer so easy for inventors and creators to quickly exchange ideas and try new things. Ready access to manufacturing capabilities would allow knowledge to circulate, traveling up and down the stack to improve both design and production. 

This inability to manufacture deprives America of the chance to lead. There are few significant American robot makers, for example; it’s an industry dominated by Europe and Japan. If more of the world will be powered by robots in the future, shouldn’t America take the lead in both manufacturing robots and owning the IP of robot making? 

Imagine the benefits of capturing much more of the hardware production around the smartphone. The offshoring of iPhone production to Shenzhen has allowed the city to become the hardware innovator of the world. The steady training of workers for over a decade has made thousands of line engineers into the world’s greatest experts in electronics assembly. That has kickstarted the rise of Chinese smartphones, which dominate the developing world today, and made Shenzhen the leader in follow-on products — “the peace dividends of the smartphone wars” as Chris Anderson describes it — that include consumer drones, scooters, wearables, and much more.

America should not be content with being a country that invents new things but lets others actually make them. It’s not just about robots or consumer electronics. The U.S. should lead on other important technologies. Renewables are one example — the solar panels that will power the batteries that will power electric vehicles. 

The U.S. has a splendid ability to do great science and invent new things. But it is persistently outclassed by other countries in scaling them up. Restoring manufacturing will help America avoid the problems of the recent past, improve the technologies of the present, and lead the industries of the future.

• 

  Dan Wang is the Shanghai-based technology analyst for Gavekal Dragonomics. He also writes at danwang.co.

  Follow Twitter Website