Chip Shortage
The chip shortage that disrupted everything from car production to game consoles was not one single event but several supply and demand shocks landing on top of a manufacturing system that has almost no spare capacity by design — semiconductor fabs are built to run near full utilization because idle capacity is enormously expensive, which means the system has very little slack to absorb a sudden shock in any direction.
The auto industry's own forecasting mistake was the spark, but not the fuel. When COVID-19 hit in 2020, automakers canceled chip orders anticipating a steep sales downturn, freeing up mature-node fab capacity almost overnight. Consumer electronics demand — laptops, monitors, webcams, and gaming hardware for people suddenly working and schooling from home — absorbed that freed capacity within months, so when auto sales rebounded faster than expected, automakers found themselves at the back of a queue for capacity that other customers had already booked on long-term contracts.
Mature-node capacity, not leading-edge capacity, was the actual bottleneck for cars. A modern vehicle can contain over a thousand chips, but the overwhelming majority are simple mature-node parts — power management, sensors, microcontrollers — built on process nodes 28 nanometers and older, the same nodes serving huge volumes of other consumer and industrial products. Because these mature nodes are cheap and thin-margin, fabs had comparatively little incentive to build new capacity there, so an auto industry now competing for the same limited mature-node wafers had nowhere to redirect demand.
Physical disruptions layered directly on top of the demand mismatch. A fire at a Renesas fab in Japan in 2021, a severe winter freeze that knocked out power to Texas fabs, and a drought that strained Taiwan's water-intensive fabs all removed capacity at the worst possible moment, each one adding weeks or months of lost production on top of an already stretched system. None of these events alone would have caused a global shortage, but arriving during a period with essentially zero slack turned each one into a multiplier on the existing problem.
Lithography and equipment lead times set the floor on how fast anyone could actually respond. Even fully funded expansion plans were gated by how quickly EUV and DUV scanners could be manufactured and delivered, since ASML and its peers can only build a limited number of these machines per year regardless of how much money customers are willing to spend. This is the structural reason the shortage lasted roughly two years rather than resolving in a single fiscal quarter: capacity additions were paced by equipment manufacturing capacity, not by capital availability.
| Cause | What it actually did | Why it mattered |
|---|---|---|
| Auto order cancellations (2020) | Freed mature-node capacity | Other customers absorbed it within months |
| Pandemic electronics demand | Filled the freed capacity fast | Left automakers at the back of the queue |
| Renesas fire, Texas freeze, Taiwan drought | Removed capacity unexpectedly | Compounded an already-thin buffer |
| EUV/DUV scanner lead times | Capped how fast new capacity could come online | New fabs still needed equipment nobody could build faster |
Read the chip shortage through a near-zero-slack lens rather than a single-cause lens: fabs run close to full utilization because idle capacity is extraordinarily expensive, so the industry had almost no buffer to absorb the auto industry's canceled orders, the pandemic's demand shift, several unrelated natural and industrial disasters, and equipment lead times that no amount of money could shorten — all landing in roughly the same eighteen-month window.