Intermediateexplainers
Why Chip Shortages Happen: Fabs, Lead Times, and Concentration
Learning objectives
- Explain why chip fabrication is slow to scale
- Describe geographic concentration risks
- Explain the role of long lead times and ordering
- Identify why shortages hit autos hardest
Fabs: the hardest factories on earth
Leading-edge chip plants cost tens of billions, take years to build, and need extreme precision — a speck of dust ruins wafers. Capacity cannot respond to demand spikes; it is planned years ahead on forecasts that are often wrong.
Chip supply is nearly fixed short-term, so demand surges become price spikes and allocation queues.
Concentration risk
A huge share of advanced chips comes from a small number of fabs in one region. Any disruption — drought (fabs need water), quake, geopolitics — threatens global supply. Mature-node chips for cars are similarly concentrated among few suppliers.
Why autos suffered most
Carmakers cancelled chip orders early in the pandemic, then reordered into a queue behind booming electronics demand. Chips are cheap per car but irreplaceable — a 40,000 vehicles. Just-in-time auto supply chains had no buffer.
Panicked buyers order from multiple suppliers, inflating apparent demand and worsening the crunch — then cancellations cascade.
The cycle repeats
Shortages trigger overbuilding, which triggers gluts — the classic semiconductor cycle, now with geopolitical subsidies layered on. Expect chronic tightness at the leading edge and periodic gluts in memory. Educational content only.
This article explains semiconductor economics. It is not investment advice.
Key points
- Fabs take years and billions — supply can't flex quickly
- Geographic concentration creates systemic risk
- Long lead times plus panic ordering amplify shortages
- Autos suffered from cancelled orders and zero buffers
- The industry cycles between shortage and glut
Common misconceptions
✓ Truth: Fabs take 3-5 years; shortages resolve through demand destruction and reallocation first.
✓ Truth: Leading-edge logic, memory, and mature auto chips are distinct markets with distinct crunches.
✓ Truth: Allocation and long-term contracts, not spot gouging, rationed most supply.
Frequently asked questions
Why do fabs need so much water?
Ultrapure water cleans wafers between hundreds of process steps; droughts directly threaten output.
What is a mature node?
Older, larger-transistor processes used for autos and appliances — cheap but capacity-constrained.
How long does it take to build a fab?
Typically 3-5 years from groundbreaking to volume production.
Will subsidies end shortages?
They add long-run capacity but can't fix short-run inelasticity; overbuilding risks future gluts.