Cypress founder: US government should not interfere in chip industry

Jerry Sanders, the former president of the semiconductor Industry Association and founder of Advanced Micro Devices, who predicted in the 1970s that “semiconductors will prove to be the crude oil of the information age,” put it in understatement.

In fact, today’s mid-range cars are controlled by 100 or more chips. Without a chip, a car can’t meet legal standards for mileage or pollution, or even start, let alone connect to the internet or Display a fancy digital dashboard. A supply shortage of just one or two chips could shut down a 100-acre assembly plant with 10,000 employees. That’s what has happened to many auto plants over the past six months. Both the automotive and semiconductor industries are asking for government help, but these problems are not something that subsidies and regulation can solve.

In the late 1970s, automobiles began to enter the digital age. My first Silicon Valley employer, American Microsystems, once “threw away the cookbook” and cut off the supply of memory chips to the Lincoln Continental factory. The car cannot start without our chips. Ford later kicked us out as a supplier, punished for causing the plant to shut down.

Soon, the auto industry created a wide range of reliability and procurement methods, which avoided many of these problems, but also bogged down the industry into bureaucracy. Today, the qualification process for new chip suppliers takes 18 to 24 months or more. That’s why car companies can’t simply buy rare chips from another supplier to keep production lines running in a pinch.

A surge in demand for chips in electronics over the years has shrunk the auto industry’s share of the chip market to around 10 percent. Automotive customers now sometimes have to line up behind other big customers with faster qualification cycles and better pricing. Automakers have complained publicly about the problem, even the White House, but they are to blame for their own supply problems. We’ve seen Toyota weather the pandemic-induced shortages by only holding a few weeks’ worth of chip inventories (or by inducing its semiconductor partners to do so on their behalf).

Supply responded when auto companies cut chip orders at the start of the pandemic. But when auto demand remains firm to surprise everyone and automakers suddenly need more chips, the semiconductor industry can’t react quickly enough. Even in robotic factories, it takes an average of 12 weeks to manufacture a chip (or longer if advanced processes are required), and that’s before back-end assembly and worldwide shipment. President Biden said he was “studying” the supply chain, but everyone in the industry knows politics and subsidies are irrelevant. Market participants will fill the chip shortage before Democrats and Republicans argue over whose fault it is.

The US has had previous experience of wasting money trying to “fix” chip problems. In 1987, the Sematech consortium spent $500 million in government funding with zero benefit to the industry. Sematech’s Austin, Texas, chip factory was outdated when it opened, and layoffs began in 2002, thanks to relentless progress in unbridled competition.

Today, the Semiconductor Industry Association is repeating the Sematech era. They argue that the industry is vital to the U.S. and that its factories cost a lot of money, but wrongly conclude that the government should subsidize it. The semiconductor industry is in good shape and has high market valuations, allowing access to cheap capital. What’s more, “free government money” led to extremely inefficient spending and improper payments to executives and shareholders. Here are some phenomena that I have seen.

Advocates of the semiconductor industry argue that if the U.S. government does not help, other countries will take market share. But for decades, governments in Europe and mainland China have poured money into chip companies, but none of them have changed the competitive chip industry. Compared with Taiwan’s free-market foundries, SMIC, the largest chip company in mainland China, seems a little irrelevant. Former President Trump also last year blocked U.S. exports of advanced semiconductor equipment to China, dealing a near-fatal blow to SMIC.

It’s true that the U.S. market share in semiconductor manufacturing has slipped to 12%, but that’s not to say companies need government help to stem further declines. Integrated Circuits have been commercialized for about 60 years, and most chips have become commodities of little strategic value, and their manufacturing has been pushed overseas due to the continued need for low cost.

These losses are not terrible. Silicon Valley’s Intel, Advanced Microdevices and Nvidia make the most advanced processors. Micron Technology, based in Boise, Idaho, makes advanced memory; Analog Devices, headquartered in Wilmington, Massachusetts, still dominates the analog chip market. U.S. defense chips don’t rely on foreign sources, while Asia-based winner TSMC is building a factory in Arizona. What’s more, the industry has entered the era of software, where the code that drives a chip is often more important than the chip itself. As Google knows, the U.S. dominates the software space.

So we don’t need to hand over taxpayer money to some of the smartest and richest companies in the world. Chip companies thrive in free markets and barely survive in controlled economies. This information shouldn’t be controversial, but in 1991 my dissatisfaction with pork barrel consumption earned me the “bad boy of Silicon Valley” label on the cover of Business Week.

My proposed solution to the current chip problem? Do nothing and let the invisible hand solve the problem for free.