What Caused the Global Chip Shortage? 2020–2022 Explained

In everyday life: late new cars, empty console shelves

Around 2021, the news kept reporting that new cars were slow to arrive. New game consoles were missing from store shelves. The reason given was almost always the same: a chip shortage.

A semiconductor chip is a part that switches the flow of electricity on and off in fine detail to calculate and control things. Chips are not only in phones and computers. They are also in cars, air conditioners and water heaters. When chips ran short, many kinds of products could not be built.

This article does not retell the daily news of that time. Instead, it looks at how the shortage worked. Once you understand the mechanism, news about the next shortage will be easier to read.

What happened: the numbers

Carmakers were hit hardest. In September 2021, the consulting firm AlixPartners forecast that about 7.7 million cars would go unbuilt worldwide that year. It put the lost revenue at $210 billion. That was nearly double its May forecast of 3.9 million cars. Note that both are forecasts, not counts of actual losses.

The time from ordering a chip to receiving it is called the lead time. Data from the financial firm Susquehanna Financial Group put the average at 17 weeks in April 2021. At the time, that was the longest since the firm began tracking in 2017. By May 2022 it had reached 27.1 weeks, more than half a year (as reported by Bloomberg).

Chips that control cars and home appliances had even longer waits. A small computer that controls a device, packed onto one chip, is called a microcontroller (MCU). In the same data, microcontrollers took 35.7 weeks to arrive in February 2022.

The demand side: what changed in 2020

In 2020, the COVID-19 pandemic moved work and school online almost overnight. According to the research firm IDC, about 300 million PCs shipped worldwide in 2020, up 13.1% from the year before. That was the biggest growth since 2010. Game consoles, tablets and other devices for use at home also sold well.

Carmakers moved the other way. In spring 2020, they expected car sales to fall, so they stopped factories and cut chip orders (McKinsey, 2022). But car sales came back faster than expected.

When carmakers tried to restore their orders, chip factories were already full of orders for PCs and phones. Chipmakers had given the capacity that carmakers released to consumer-electronics makers and others (Bloomberg, January 2021). Capacity once given up could not be won back quickly.

In 2021, the US Department of Commerce surveyed companies that make and use chips. It found that demand in 2021 was up to 17% higher than in 2019. Demand had grown, but the chips that reached buyers did not grow to match.

The supply side: why output can't rise quickly

1. Making chips takes months

Chips are built on a thin round slice of silicon, called a wafer, through hundreds of repeated steps. This stage is called the front end. A 2021 explainer from the Semiconductor Industry Association (SIA) says the front end alone takes about 12 weeks on average. Next comes the back end, where chips are cut apart, assembled and tested. It takes about 6 more weeks. Even if orders jump suddenly, shipments rise only months later. "How Are Chips Made?" explains the steps.

2. New factories take years and huge sums

Building a new factory takes even longer. For example, in November 2021 TSMC announced a plan to build a factory in Kumamoto, Japan. Construction began in 2022, and volume production started at the end of 2024. The initial investment was put at about $7 billion. "What Is a Semiconductor Foundry?" explains why factories cost so much.

3. Older-generation capacity was not growing

Chipmaking processes come in generations, named with numbers like "X nm." Many microcontrollers and power-management chips for cars are made with processes several generations behind the leading edge. The industry calls these proven older generations "mature nodes" or "legacy nodes."

A 2022 McKinsey analysis found that most automotive chips are made at 90 nm or larger. Capacity at 90 nm was expected to grow only about 5% a year from 2021 to 2026. According to the analysis, that would not keep up with growing demand. "What Does 2nm or 3nm Mean?" explains what the node names mean.

Events that made it worse

Demand outran supply, and no part of the chain had any slack. Then one event after another shut factories down. With no slack, a stop at a single site spreads far.

The Malaysia case was about the back end. Companies that take on back-end work for others are called OSATs, and many are in Southeast Asia. According to the Malaysian Investment Development Authority (MIDA), Malaysia has about 13% of the world's back-end capacity. Even with finished wafers, chips cannot ship until they are assembled and tested.

The bullwhip effect: orders swing harder upstream

Another reason the shortage grew was the way orders travel. A small change in sales can grow into a big change in orders. This happens as orders move back through the supply chain (upstream). This is called the bullwhip effect. A bullwhip is a long whip used to drive cattle.

Let's look at a simple example. A device maker orders from a parts maker, and the parts maker orders from a chipmaker. Each company keeps stock equal to two weeks of the orders it received that week. At first, 100 units sell each week, and each company holds 200 units. To keep things simple, we ignore delivery time here.

One week, device sales rise 10% to 110 units. The device maker replaces the 110 units it sold and raises its stock target to 220 units. So it orders 110 plus 20, or 130 units. From the parts maker's view, orders jumped 30%.

The parts maker also raises its stock target, to two weeks of 130 units, or 260 units. So it orders 130 plus 60, or 190 units. Customers bought only 10% more, yet the chipmaker received 90% more orders.

The next week, sales stay at 110. The device maker's order drops back to 110. The parts maker can lower its stock to 220, so it orders 40 fewer, only 70 units. Now the orders going upstream fall sharply.

In the real shortage, longer waits led companies to order even further ahead. According to McKinsey (2022), carmakers and their parts suppliers began ordering about 10–20% more than they needed, to guard against running out.

Why were carmakers hit hardest?

The first reason is how they hold stock. Many car plants have parts delivered only when needed, and only in the amounts needed. This keeps stock low, and it is called "just-in-time." It wastes little, but when parts stop arriving, production stops almost at once.

One number shows how thin stocks were. It comes from a US Department of Commerce survey published in January 2022, covering all industries, not only cars. Companies that use chips held a median of 40 days' worth of chips in 2019 (the middle company's value). By 2021, that had fallen to less than 5 days.

The second reason is the sheer number of chip types a car uses. A car has many small units that each control one thing, such as the engine, the brakes or the windows. These units are called ECUs (electronic control units), and most are built around a microcontroller. If just one kind of cheap chip is missing, the car cannot be finished.

The third reason is that chips cannot be swapped easily. Car parts are tested to confirm they can withstand heat, cold and vibration for many years. The best-known test standard is AEC-Q100. A change of factory or material often means repeating the tests, so carmakers cannot switch to another chip quickly.

How the shortage eased

In the second half of 2022, sales of consumer devices such as PCs and phones fell. Chip lead times began to shrink clearly in October 2022 (Bloomberg, November 2022). The bullwhip effect ran in reverse, and the extra stock that companies had piled up turned into surplus.

Companies cut orders until they had used up their surplus stock. This is called inventory adjustment. Partly because of this, world chip sales in 2023 were $526.8 billion, down 8.2% from the year before, according to SIA. This pattern of booms and slumps taking turns is called the silicon cycle.

Not every type of chip went into surplus at the same time, though. In October 2022, McKinsey warned that the gap between supply and demand for automotive chips could persist. The reason was the slow growth of mature-node capacity.

What changed afterward: government policies

The shortage was one of the triggers for many countries and regions to strengthen chipmaking at home. The main policies are listed below with their dates.

Most of the factories built under these policies started running after the shortage had eased. In other words, the policies did not fix that shortage. They are meant to prepare for the next one.

Companies changed their thinking too. In 2022, McKinsey advised carmakers to list every chip their cars use and keep track of them. It also proposed that carmakers, parts suppliers and chipmakers share demand forecasts closely. The aim is to make the bullwhip swings smaller.

Common misconceptions

Key numbers

Note that the AlixPartners car figure is a forecast made in September 2021. The lead times come from one firm's tracking, and other firms report different values.

Learn more

• Who designs chips and who makes them: "What Is a Semiconductor Foundry?" and the side-by-side table in "IDM vs Fabless vs Foundry."
• Front end and back end: "Front-End vs Back-End Semiconductor Process" sorts out the steps and the companies that do them.
• Computers inside cars and appliances: "What Is a CPU?" and "MCU vs MPU" explain what a microcontroller does.
• Repeating booms and slumps: the glossary page on the silicon cycle explains why the cycle happens.
• Reliability of car chips: see AEC-Q100 and zero defects, the goal of eliminating defects, in the glossary.