Category: Computing

Last week, my wife Rebecca, who knows what she wants and when she wants it, decided to replace her aging and worn Android phone, which she can upgrade without paying extra. Fully vaccinated and ready for a post-pandemic treat, she set off for the cellular store. Several hours later, she returned in a sour mood, her old phone still in her purse. The tech at the store told her that she couldn’t upgrade because they had no new phones to offer. The chip shortage.

Want a new car? You may have a wait. The chip shortage.

What a time for a shortage. The pandemic is winding down and the economy is winding up.

The Fourth Industrial Age

The World Economic Forum says we are entering The Fourth Industrial Age where digital models and communications combine with physical processes for speed and efficiency. In the first industrial age, beginning in the 18^th century, society began to harness energy to replace human and animal muscle; in the second age, industries were built around mass production in factories; in the third age, automated controls and computerization increased productivity.

The fourth industrial age is just in time for the global pandemic. We can be grateful that computerized gene analysis enabled development of lifesaving covid-19 vaccines in record time. Computer networks have supported productivity and commerce through quarantines and lockdowns. Distributed network management shored up shattered supply chains. As much as we complain about Zoom, social media, and video streaming, they made the lockdowns and quarantines tolerable, kept education alive, and allowed many of to continue to be productive by working at home.

Only the future will reveal where the fourth industrial revolution will take us, but one thing is clear: previous industrial ages ran on coal, oil, hydroelectricity, and nuclear power. The fourth industrial age requires energy, but more than any other commodity, advancement in the fourth age depends on more and better computer chips.

The computer chip

The computer chip started as mechanical relays invented in the early part of the 19^th century. A relay is an electrical switch controlled by another electrical circuit. The circuit that flips the switch uses only a few volts to control another stronger electrical current. When you start your car, the current to turn over the engine would quickly burn out the switch on your steering column or dash. This doesn’t happen because a relay isolates the driver operated switch driver from the massive power surge that turns the engine over.

Vacuum tubes, invented it the early 20^th century, performed many of the same relay switching functions, but faster. With tubes, came audio amplifiers, radios, and early digital computers.

Transistors, which appeared after WWII, are still faster, more compact, require far less power, and have lifespans measured in decades instead of hours, making the complex digital devices and controls of today practical.

Computer chips are tightly packed arrays of transistors on thin slices of silicon. In 1965, Intel co-founder and engineer, George Moore, predicted that the density of transistors on computer chips would double every one to two years.

This prediction was dubbed “Moore’s Law.” Since its beginning, experts have predicted the impending end of Moore’s law, but after nearly sixty years of exponential growth, the law still holds. The ingenuity of chip technologists, largely from the U.S., has been startling. Advanced chips today have over a billion transistors, that’s more than the number of grains of sand in a five-gallon bucket. Think about wiring together every grain of sand in that bucket in an exact pattern connecting each grain of sand with every other one and you get an idea of just how hard chip manufacturing is. Now, shrink the size of each grain of sand so that they all fit in a few layers the size of a postage stamp.

American engineers figured out not only how to perform this staggering task, they devised ways of using these contraptions to control cars, improve the quality of steel, discover likely covid-19 vaccines. Digital processes power Zoom meetings, deliver pizza, anticipate storm surges in sewers, and broadcast cat photos.

Today, a cutting edge computer chip is undoubtedly the most difficult manufacturing challenge on the planet, requiring hundreds of precision operations, so precise they are calibrated in wavelengths of light. Chips must be small because their speed is limited by the time required for a signal moving near the speed of light to travel from one side of the chip to the other.

Outsourced chip manufacturing

As the source of chip manufacturing technology, you would expect the U.S. to be the leading computer chip manufacturer. It is: Intel and a few other U.S. companies dominate the field. But they do and they don’t. U.S. companies design the chips and the processes to manufacture the chips, but they often outsource the fabrication to firms in Asia, primarily in South Korea and Taiwan. Companies in South Korea and Taiwan have factories on the Chinese mainland where skilled workers are plentiful and wages are low.

And there you have it: U.S. chip innovators depend on manufacturing capacity in mainland China.

Why? Chip manufacturing in China is cheap and the quality is high. The managers of the U.S. chip companies like Intel, Invidia, and AMD are obliged to optimize shareholder value. In corporate America, passing up opportunities for increased profits ends careers. Executives must manufacture chips as cheaply and efficiently as possible. They are compelled by the market to outsource to the U.S.’s leading economic and social competitor, China.

The shortage

But don’t jump to the conclusion that the chip shortage is caused by China. U.S. corporations may have shortsightedly handed chip manufacturing to the Chinese, but the shortage today is not the result of secret directives from Beijing.

The shortage was caused by the rapid progress of the fourth industrial age. From the 1970s, when computer chips first came on the scene, tech companies— computer, smartphone, and networking gear manufacturers— were the primary consumers of chips. But this has changed. Automobiles have become mobile computer data centers. The Internet of Things requires millions of computer chips in home appliances and industrial sensors and controls. Industrial robots must have chips.

Consequently, the chip market has expanded far beyond the tech sector. Adding chip production lines is difficult and slow. Think of the gargantuan private and public effort put into developing vaccine production lines. Chip production lines are more difficult, and the demand is higher. It’s not news that rapid increase in consumption of hard to manufacture commodities precedes shortages.

Covid-19 disruption

The fourth industrial revolution may have blunted the damage from the pandemic, but covid-19 entered the scene at the worst possible time for the chip industry. Let me count the ways.

Chip factories had to slow production as employees became sick with covid-19. These factories have their own supply chains for raw materials, subcomponents, and manufacturing equipment. The global pandemic disrupted these supply chains and sources as well as the factories themselves.

It gets worse. Consumers quit buying. Automobile sales plummeted and the big automakers cut back their orders, falling to the back of the queue. People quit flying. Chip manufacturing relies on cargo space on passenger flights to ship their tiny high-value products and receive materials and subcomponents, but passenger flights were cancelled. The alternative, container ships, are a slow inferior choice for shipping, and they too had their covid-19 problems.

Econ 101

Covid-19 did much more than inhibit chip production. It also increased chip demand. Kids needed computers for remote schooling. Parents needed equipment to work from home. Network usage soared, which required added network gear. In the U.S., by the end of 2020, we were using computing and the computer networks the way that the experts had predicted for 2030.

Increased demand and decreased supply. Sounds like an exercise in disaster prediction from Econ 101. Here we are. Coming out of a pandemic with a roaring fourth stage of industry demand for chips and suppliers struggling to fill orders.

How long it will take to stabilize chip production is hard to predict. Some say by the end of 2021. Things are likely get worse before they get better.

U.S. and China

This is painful, but not all bad, because it draws attention to a glaring problem. Even after the current shortage goes away, the U.S. is still in trouble. Outsourcing of sophisticated manufacturing makes managerial and profit sense, but it is a recipe for disaster. The U.S. rivalry with China is nothing like the U.S.-Soviet cold war. At the height of the cold war, the U.S. depended on the Soviet Union for fish eggs (caviar) and furs, but not much else, and the Soviet resource-based economy didn’t depend on the U.S. That left both sides free to exercise military strategies with little regard for economic consequences.

Today, the U.S. and China are economically intertwined in ways that the U.S. and the Soviets never approached. Don’t expect to see Xi Jinping pounding on a desk with his shoe like Nikita Khrushchev at the United Nations in 1960, but expect a series of confrontations and tense maneuvering for advantage. In the cold war with the Soviets, the contest was mainly ideological: state socialism vs. capitalism. Today, the superiority of capitalism is a foregone conclusion in China; the contest is between an authoritarian and a democratic state. Xi manipulates markets to achieve what he perceives as the best deal for the Chinese people. In the U.S., the people direct the market and hope they achieve their goals. If the free market says outsource to China and the people agree, so be it.

A solution

But not all Americans agree that the free market has the best solution to the chip shortage. Some folks, including me, think that we ought to identify the resources we depend upon and act for long term control of our future. They see prioritizing and supporting our own chip manufacturing base as a healthy approach to continuing democracy in the fourth industrial age.

In another venue, we can argue where private enterprise and government enterprise should prevail. But for now, I hope for a government that encourages long-term investment in chip manufacturing and discourages short-sighted profit-taking on outsourcing. We need a landscape in which every off-shore outsource has a vigorous onshore competitor. May the best contestant win, but let’s make sure that onshore contestants are on an even playing field.

We can do this.